UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A survey of underwater wireless optical communication Zeng, Zhaoquan 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2016_february_zeng_zhaoquan.pdf [ 1.82MB ]
Metadata
JSON: 24-1.0220823.json
JSON-LD: 24-1.0220823-ld.json
RDF/XML (Pretty): 24-1.0220823-rdf.xml
RDF/JSON: 24-1.0220823-rdf.json
Turtle: 24-1.0220823-turtle.txt
N-Triples: 24-1.0220823-rdf-ntriples.txt
Original Record: 24-1.0220823-source.json
Full Text
24-1.0220823-fulltext.txt
Citation
24-1.0220823.ris

Full Text

A Survey of Underwater WirelessOptical CommunicationbyZhaoquan ZengB.Eng., Tianjin University, P. R. China, 2013A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF APPLIED SCIENCEinTHE COLLEGE OF GRADUATE STUDIES(Electrical Engineering)THE UNIVERSITY OF BRITISH COLUMBIA(Okanagan)December 2015c© Zhaoquan Zeng, 2015AbstractUnderwater wireless communication refers to transmitting data in unguided water environmentthrough the use of wireless carriers, i.e., radio-frequency wave, acoustic wave, and optical wave.We focus, in this thesis, on the underwater wireless optical communication (UWOC) that employsoptical wave as the transmission carriers. In comparison to RF and acoustic counterparts, UWOChas a much higher transmission bandwidth, thus providing much higher data rate. Due to thishigh-speed transmission advantage, UWOC has attracted considerable attention in recent years.Many potential applications of UWOC systems have been proposed for environmental monitoring,offshore exploration, disaster precaution, and military operations. However, UWOC systems alsosuffer from severe absorption and scattering introduced by underwater channel. In order to overcomethese technical challenges, several new system design approaches, which are different from theconventional terrestrial free-space optical communication, have been explored in recent years. In thisthesis, we provide a comprehensive survey of the state-of-the-art of UWOC research in three aspects:channel characterization, channel modulation and coding techniques, and practical implementationsof UWOC. Based on the comprehensive understanding of UWOC, we also investigate the outageperformance for vertical buoy-based UWOC with pointing errors. Closed-form outage probabilitywith zero boresight pointing errors and outage probability bounds with nonzero boresight pointingerrors have been derived.iiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixList of Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiiAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xivDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvChapter 1: Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Overview of Underwater Wireless Optical Communication . . . . . . . . . . . . . . . 11.2 Advantages and Challenges of UWOC . . . . . . . . . . . . . . . . . . . . . . . . . . 71.3 Thesis Organization and Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . 9Chapter 2: UWOC Channel Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . 122.1 Light Propagation in Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122.2 Modeling of Aquatic Optical Attenuation in UWOC . . . . . . . . . . . . . . . . . . 202.2.1 Aquatic Optical Attenuation in LOS Configuration . . . . . . . . . . . . . . . 202.2.2 Aquatic Optical Attenuation in NLOS Configuration . . . . . . . . . . . . . . 252.3 Modeling Geometric Misalignment of UWOC . . . . . . . . . . . . . . . . . . . . . . 262.4 Modeling Link Turbulence of UWOC . . . . . . . . . . . . . . . . . . . . . . . . . . . 282.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29iiiTABLE OF CONTENTSChapter 3: UWOC Channel Modulation and Coding Techniques . . . . . . . . . . 313.1 Modulation Schemes of UWOC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313.2 Channel Coding of UWOC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37Chapter 4: Experimental Setups and Prototypes of UWOC . . . . . . . . . . . . . 394.1 Typical LOS/NLOS UWOC systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 394.2 Retroreflectors in UWOC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434.3 Smart Transceivers of UWOC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454.4 UWOC for Underwater Vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474.5 Hybrid Acoustic/Optical UWC Systems . . . . . . . . . . . . . . . . . . . . . . . . . 484.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52Chapter 5: Outage Performance for Underwater Wireless Optical Links WithPointing Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545.1 Pointing Errors Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545.1.1 Pointing Errors Model with Zero Boresight . . . . . . . . . . . . . . . . . . . 545.1.2 Pointing Errors Model with Nonzero Boresight . . . . . . . . . . . . . . . . . 595.2 Beam Spread Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605.3 Outage Probability with Zero Boresight Pointing Errors . . . . . . . . . . . . . . . . 625.4 Bounds of Outage Probability with Nonzero Boresight Pointing Errors . . . . . . . . 635.4.1 Lower Bound of Outage Probability with Nonzero Boresight . . . . . . . . . . 645.4.2 Upper Bound of Outage Probability with Nonzero Boresight . . . . . . . . . 665.4.3 Discussion on the Tightness of the Outage Probability Bounds . . . . . . . . 685.5 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 705.5.1 Outage Probability with Zero Boresight Pointing Errors . . . . . . . . . . . . 705.5.2 Outage Probability Bounds with Nonzero Boresight Pointing Errors . . . . . 735.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75Chapter 6: Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 766.1 Summary of Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 766.2 Suggested Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80ivTABLE OF CONTENTSAppendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100Appendix A: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101Appendix B: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102vList of TablesTable 1.1 Comparison of underwater wireless communication technologies [1]. . . . . . . 10Table 2.1 Summary of absorption and scattering characteristics of seawater [2] . . . . . 18Table 2.2 Typical values of a(λ), b(λ), and c(λ) for different water types . . . . . . . . . 19Table 2.3 Summary of literatures on UWOC channel modeling . . . . . . . . . . . . . . 30Table 3.1 Summary of literatures on UWOC modulation schemes . . . . . . . . . . . . . 38Table 3.2 Summary of literatures on UWOC channel coding . . . . . . . . . . . . . . . 38Table 4.1 Summary of literatures on experimental setups and prototypes of UWOC . . 53Table 5.1 Summary of simulation parameters . . . . . . . . . . . . . . . . . . . . . . . . 70viList of FiguresFigure 1.1 The “transparent window” for light aquatic attenuation is shown with blueand green color. Please refer to the colored version of this thesis. Figure 1.1is adapted from [3]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Figure 1.2 An underwater wireless sensor network with aerospace and terrestrial com-munication. Figure 1.2 is adapted from [4]. . . . . . . . . . . . . . . . . . . . 4Figure 1.3 Link configurations of UWOC. . . . . . . . . . . . . . . . . . . . . . . . . . . 5Figure 2.1 Geometry of inherent optical properties for a volume ∆V . Figure 2.1 isadapted from [5]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Figure 2.2 Optical absorption spectra for different ocean components. The data in Fig-ure 2.2 is from [6–8]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16Figure 2.3 Optical scattering spectra for different ocean components. The data in Figure2.3 is from [6–8]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Figure 3.1 Illustration of OOK, PWM, PPM and DPIM. Figure 3.1 is adapted from [9]. 32Figure 4.1 A typical laboratory LOS UWOC system based on intensity-modulationdirect-detection (IM/DD) technique. . . . . . . . . . . . . . . . . . . . . . . . 40Figure 4.2 Demonstration of corner and spherical retroreflectors. . . . . . . . . . . . . . 44Figure 4.3 Modulating retroreflector link. . . . . . . . . . . . . . . . . . . . . . . . . . . 44Figure 4.4 Two types of hybrid acoustic/optical UWC links. . . . . . . . . . . . . . . . 50Figure 5.1 Comparison between corrected and non-corrected PDF of ocean slopes. . . . 56Figure 5.2 Geometry of the buoy-based UWOC. Figure 5.2 is adapted from [10]. . . . . 57Figure 5.3 PDF of Hoyt distributed radial displacement r with L = 5m and differentvalues of wind speed U . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Figure 5.4 PDF of Beckmann distributed radial displacement r with µx = 0.01, µy =0.02, L = 5m and different values of wind speed U . . . . . . . . . . . . . . . 59viiLIST OF FIGURESFigure 5.5 Geometry for BSF. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60Figure 5.6 BSF results for L = 3.63m with different attenuation coefficients c. Modelresults shown as lines. Experimental data shown as points. Relative intensityis the received power with displacement r normalized by the power withoutdisplacement B(L, r)/B(L, 0) [11]. Figure 5.6 is reprinted from [11]. . . . . . 61Figure 5.7 BSF values for L = 5m and c = 0.3 with different values of transmissionpower Pt. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61Figure 5.8 Demonstration of outage region. . . . . . . . . . . . . . . . . . . . . . . . . . 63Figure 5.9 Integrating region for lower bound of outage probability. . . . . . . . . . . . 64Figure 5.10 Coordinates of nth circumscribed rectangle on the upper semicircle. . . . . . 64Figure 5.11 Divide half of the non-outage region into several rectangles with the sameheight. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66Figure 5.12 Coordinates of nth inscribed rectangle on the upper semicircle. . . . . . . . . 67Figure 5.13 Demonstration of the area Sshade =∑2 (SLB − SUB). A factor of 2 indicatesthe symmetry of rectangle area in lower and upper semicircles. . . . . . . . . 69Figure 5.14 Outage probability of a vertical buoy-based UWOC system with zero bore-sight pointing errors. Link distance L = 5m . . . . . . . . . . . . . . . . . . 71Figure 5.15 Outage probability of a vertical buoy-based UWOC system with zero bore-sight pointing errors. Link distance L = 10m . . . . . . . . . . . . . . . . . . 72Figure 5.16 Outage probability and its bound with nonzero boresight pointing errors.µx = 0.01, µy = 0.02, U = 2m/s, c = 0.3m−1, L = 5m, D = 5cm, γth = 0.01. 74viiiList of AcronymsAcronyms DefinitionsAMOUR Autonomous Modular Optical Underwater RobotAOPs Apparent Optical PropertiesAPD Avalanche PhotodiodeAUVs Autonomous Underwater VehiclesBCH Bose-Chaudhuri-HocquenghemBER Bit-Error RateBPSK Binary Phase-Shift KeyingCDMA Code Division Multiplexing AccessCDOM Colored Dissolved Organic MaterialCORK-OTS Circulation Obviation Retrofit Kit Optical Telemetry SystemCRC Cyclic Redundancy CheckCSAIL Computer Science and Artificial Intelligence LaboratoryDPIM Digital Pulse Interval ModulationDPPM Differential Pulse Position ModulationDPSK Differential Phase-Shift KeyingDT Dynamic ThresholdFEC Forward Error CorrectionFOV Field of ViewFSO Free Space OpticalixList of AcronymsGbps Gigabit per SecondGMSK Gaussian Minimum Shift KeyingHDL Hardware Description LanguageHG Henyey-GreensteinIM/DD Intensity Modulation/Direct DetectionIOPs Inherent Optical PropertiesISI Inter Symbol InterferenceKbps Kilobit per SecondLD Laser DiodeLDPC Low-Density Parity-CheckLED Light-Emitting DiodeLOS Line-of-SightLT Luby TransformMAC Medium Access ControlMbps Megabit per SecondMEMS Micro-Electromechanical SystemMIMO Multiple-Input Multiple-OutputMIT Massachusetts Institute of TechnologyML Maximum LikelihoodMPPM Multi-Pulse Pulse Position ModulationNLOS Non-Line-of-SightNRL Naval Research LaboratoryNRZ-OOK Non-Return-to-Zero On-Off KeyingOFDM Orthogonal Frequency-Division MultiplexingOOK On-Off KeyingxList of AcronymsOOK On-Off KeyingOWC Optical Wireless CommunicationsPC Personal ComputerPDF Probability Density FunctionPIN Positive-Intrinsic-NegativePolSK Polarization Shift KeyingPPM Pulse Position ModulationP-PPM Polarized- Pulse Position ModulationPSK Phase-Shift KeyingPWM Pulse Width ModulationQAM Quadrature Amplitude ModulationQPSK Quadrature Phase-Shift KeyingRF Radio FrequencyRF-EM Radio-Frequency ElectromegneticROVs Remotely Operated Underwater VehiclesRS Reed-SolomonRTE Radiative Transfer EquationRZ-OOK Return-to-Zero On-Off KeyingSIM Subcarrier Intensity ModulationSIMO Single-Input Multiple-OutputSISO Single-Input Single-OutputSNR Signal-to-Noise RatioSPF Scattering Phase FunctionTDMA Time Division Multiplexing AccessUTROV Untethered ROVxiList of AcronymsUWC Underwater wireless communicationUWOC Underwater Wireless Optical CommunicationUWSNs Underwater Wireless Senor NetworksVSF Volume Scattering FunctionxiiList of SymbolsSymbols Definitionsa(λ) The absorption coefficientb(λ) The scattering coefficientc(λ) The attenuation coefficientI0(·) The modified Bessel function of the first kindJ0(·) The Bessel function of the first kind with order 0p(·) The Hankel transformerf(x) The error functionQ1(·, ·) The first order Marcum Q-functionln(·) The log function with base elim The limit of function| · | The absolute value of the argument∇ The divergence operatorxiiiAcknowledgementsI am deeply grateful to my thesis supervisor Dr. Julian Cheng for his enthusiasm, guidance,advice, encouragement, support, and friendship. I will continue to be influenced by his rigorousscholarship, clarity in thinking, and professional integrity.I owe many people for their generosity and support during my master study at the University ofBritish Columbia. I would like to thank my dear colleagues for sharing their academic experiencesand constructive viewpoints generously with me during our discussions. I would also like to thankmy dear friends for sharing my excitement and encouraging me when I was in frustration duringthis journey.Finally, I would like to thank my parents for their patience, understanding, support, and loveover all these years. All my achievements would not have been possible without their constantencouragement and support.xivTo My Loving ParentsxvChapter 1Introduction1.1 Overview of Underwater Wireless Optical CommunicationTwo thirds of the earth’s surface is covered with water. During the past thousands of years,humans have never stopped the exploration of the ocean. In recent years, with an increase of globeclimate change and resource depletion of land, there has been a growing interest in the researchof ocean exploration system. Underwater wireless communication (UWC) technology enables therealization of ocean exploration systems, and thus attracts more and more attention. UWC refersto transmitting data in an unguided water environment through the use of wireless carriers, i.e.radio-frequency (RF) waves, acoustic waves, and optical waves. Considering the limited bandwidthof RF and acoustic methods and the increasing need for high-speed underwater data transmission,underwater wireless optical communication (UWOC) has become an attractive and viable alterna-tive. In fact, light has been used as a wireless communication method for thousands of years invarious forms. For instance, the ancient Chinese used beacon towers in order to deliver militaryinformation around 1,000 BC, and the ancient Greek and Roman armies used polished shields toreflect sunlight for signaling around 800 BC. In 1880, Alexander Graham Bell developed a newwireless telephone system that used sunlight as the transmission medium. This system is regardedas the first optoelectronic communication system in the world [12, 13]. In the 1960s, the inventionof laser as an ideal optical source has changed the future of optical wireless communication (OWC)[14]. From that time on, a flurry of terrestrial OWC applications appeared. But due to the severeattenuation effects of seawater to visible light and the limited knowledge of aquatic optics, the earlydevelopment of UWOC was far behind the terrestrial free-space optical (FSO) communications.Based on nearly 20 years experimental and theoretical study of light propagation in the sea,in 1963, Duntley proposed that seawater shows a relatively low attenuation property to light withwavelengths from 450nm to 550nm which corresponds to the blue and green spectrum (Figure1.1) [15]. This finding was then experimentally confirmed by Gilbert et al. [16]. The existenceof the blue-green light transmission“window” in water provides a foundation for the development11.1. Overview of Underwater Wireless Optical Communicationof future UWOC. The early applications of UWOC are mainly for military purpose, especially inthe area of submarine communications. In 1976, Karp evaluated the feasibility of wireless opticalcommunications between underwater and above surface (satellite) terminals [17]. In 1977, theresearchers in the Lawrence Livermore Laboratory of the University of California proposed an one-way optical communication system from shore to submarine [18]. The transmitter of the UWOCsystem employed blue-green laser source to generate light pulses. It was flexible to be carried bya land vehicle or an airplane due to its compact architecture. The transmitter can also focus itsoutput light beam on a relay satellite, which then reflects the beam to a submarine [18]. OtherUWOC tests of the plane-to-submarine and satellite-to-submarine topologies were established byUS and Russian navy during 1980s and 1990s [19]. Mixed optical communication links with abouttens of kilometers FSO link and several tens of meters UWOC link were achieved.Over the decades, the interest of UWOC is still limited to military applications [18, 20]. Themassive market promotion of UWOC has not been achieved so far. Only a few limited UWOCproducts were commercialized in the early 2000s, such as the BlueComm UWOC system which canachieve 20 Mbps underwater data transmission over 200m distance, and the Ambalux UWOC systemwhich can provide the same data transmission rate in a shorter range of 20m [21]. In order to satisfythe increasing demands for ocean exploration with efficient high bandwidth data transmission,researchers have proposed the concept of underwater wireless sensor networks (UWSNs). Theproposal of UWSNs has greatly facilitated the development of UWOC. Thus the market of UWOChas begun to show a future promise. The basic UWSNs consist of many distributed nodes suchas seabed sensors, relay buoys, autonomous underwater vehicles (AUVs) and remotely operatedunderwater vehicles (ROVs) (Figure 1.2). These nodes have capabilities to accomplish sensing,processing, and communication tasks that maintain the collaborative monitoring to the underwaterenvironment [4]. In Figure 1.2, sensors located at the bottom of the seabed collect data andtransmit via acoustic or optical links to the AUVs and ROVs. Then, AUVs and ROVs relay signalsto ships, submarines, communication buoys and other underwater vehicles. Above the sea surface,the onshore data center processes data and communicates with satellite and ships through RF orFSO links.Based on link configurations between the nodes in UWSNs, UWOC can be divided into fourcategories (Figure 1.3) [22]: a) Point-to-point line-of-sight (LOS) configuration, b) Diffused LOSconfiguration, c) Retroreflector-based LOS configuration, and d) Non-line-of-sight (NLOS) configu-ration.21.1. Overview of Underwater Wireless Optical CommunicationFigure 1.1: The “transparent window” for light aquatic attenuation is shown with blue and greencolor. Please refer to the colored version of this thesis. Figure 1.1 is adapted from [3].31.1. Overview of Underwater Wireless Optical CommunicationOnshore data centerSubmarineROVROVROVShipShipSatelliteROVSensorSensorSensorCommunication buoyUndersea obstacleSensorSea surfaceOptical/Acoustic linkRF/FSO linkFigure 1.2: An underwater wireless sensor network with aerospace and terrestrial communication.Figure 1.2 is adapted from [4].41.1. Overview of Underwater Wireless Optical Communication(a) Point-to-point LOS configuration.(b) Diffused LOS configuration.Retroreflector(c) Retroreflector-based LOS configuration.ObstaclesSea Surface(d) NLOS configuration.Figure 1.3: Link configurations of UWOC.51.1. Overview of Underwater Wireless Optical Communicationa) Point-to-point LOS configuration (Figure 1.3(a)) is the most commonly used link configurationin UWOC [23]. In point-to-point LOS configuration, the receiver detects the light beam in thedirection of the transmitter. Since the point-to-point LOS UWOC system commonly employslight sources with a narrow divergence angle, such as a laser, it requires precise pointing betweentransmitter and receiver. This requirement will limit the performance of UWOC systems inturbid or turbulent water environments and then becomes a severe problem when the transmitterand the receiver are non-stationary nodes, such as AUVs and ROV [22].b) Diffused LOS configuration employs diffused light sources with large divergence angle such ashigh-power light-emitting diodes (LEDs) to accomplish broadcasting UWOC from one nodeto multiple nodes (Figure 1.3(b)). Broadcasting method can relax the requirement of precisepointing. However, compared with the point-to-point LOS configuration, the diffused-light basedlink suffers from aquatic attenuation due to the large interaction area with water. Relatively shortcommunication distances and lower data rates are the two major limitations of this configuration.c) Retroreflector-based LOS configuration, as shown in Figure 1.3(c), can be regarded as one specialimplementation of point-to-point LOS configuration. This configuration is suitable for duplexUWOC systems having limited power and weight budget, such as an underwater sensor node.In modulating retro-reflector link, the transmitted light is reflected back from a modulatedretro-reflector. During this process, the information that the retroreflector responses to thetransceiver will be encoded on the reflected light. Since there is no laser or other light sources inthe retroreflector end, its power consumption, volume and weight will be tremendously reduced.One limitation of this configuration is that the backscatter of the transmitted optical signalmay interfere the reflected signal, thus degrading the system signal to noise ratio (SNR) andbit-error-rate (BER). Moreover, since the optical signals will go through the underwater channeltwice, received signal will experience additional attenuation.d) NLOS configuration (Figure 1.3(d)) overcomes the alignment restriction of LOS UWOC. In thisconfiguration, the transmitter projects the light beam to the sea surface with an angle of incidencegreater than the critical angle, so that the light beam experiences a total internal reflection [24].The receiver should keep facing the sea surface in a direction that is approximately parallel withthe reflected light to ensure proper signal receiving. The major challenge of NLOS links is therandom sea surface slopes induced by wind or other turbulence sources [25]. These undesirablephenomena will reflect light back to the transmitter and cause severe signal dispersion.61.2. Advantages and Challenges of UWOC1.2 Advantages and Challenges of UWOCUWOC systems are used for high speed underwater communications between multiple fixed ormobile nodes. They have great potential for applications in the UWSNs. Conventionally, there arethree UWC choices for implementing UWSNs: acoustics, RF and optics [1]. In order to emphasizethe unique advantages and characterizations of UWOC, we will compare the UWOC with RF andacoustic methods in the following of this section.The acoustic method is the most widely used technology in UWC. It has a long applicationhistory that can be dated to late 1800s. After an extensive expansion of military applications duringthe two World Wars, underwater acoustic communication system has become a popular proventechnology that has been applied to almost every aspect of UWSNs [26]. Considering the extremebroadness of ocean and the strong attenuation effect of seawater to other transmission sources likeoptical wave and RF wave, the most attractive advantage of underwater acoustic communication isthat it can achieve a long link range up to several tens of kilometers [27]. Although acoustic methodis the most popular method to achieve UWC, it also has certain intrinsic technical limitations.Firstly, since the typical frequencies associated with underwater acoustics are between 10 Hz and 1MHz, the transmission data rate of acoustic link is relatively low (typically on the order of kbps) [4].Secondly, due to the slow propagation speed of sound wave in water (about 1500m/s for 20 Celsiuspure water), the acoustic link suffers from severe communication delay (typically in seconds). Thusit can’t support applications which require real-time large volume data exchange. Thirdly, acoustictransceivers are usually bulky, costly and energy consuming. They are not economical for largescale UWSNs implementations [28]. Furthermore, acoustic technology can also impact marine lifewhich uses sound waves in order to accomplish communication and navigation [29].The underwater RF electromagnetic (EM) communication can be seen as an extension of theterrestrial RF-EM communication. The underwater RF communication has two major advantages.First, compared with acoustic wave and optical wave, the RF wave can perform a relatively smoothtransition through air/water interface . This benefit can be used to achieve the cross-boundarycommunication which combines the terrestrial RF communication system and underwater RF-EMcommunication system together. Second, RF-EM method is more tolerant to water turbulence andturbidity than optical and acoustic methods [1]. The fatal limitation that impedes the developmentof underwater RF-EM method is its short link range. Since seawater that contains lots of salt isa conductive transmission media, the RF waves can only propagate a few meters at extra-low fre-71.2. Advantages and Challenges of UWOCquencies (30-300Hz) [4]. Moreover, the underwater RF-EM systems also require huge transmissionantenna and costly, energy-consuming transceivers.Compared with the acoustic approach and RF-EM approach, UWOC has the highest transmis-sion data rate, lowest link delay and the lowest implementation costs. UWOC can achieve a datarate on the order of Gbps over moderate distances of tens of meters. This high-speed advantagewill guarantee the realization of many real-time applications such as underwater video transmission.Since the transmission speed of light in water is much higher than acoustic wave, UWOC links areimmune to link latency. UWOC also has higher communication security over the acoustic and RFmethods. Most UWOC systems are implemented in LOS configuration, rather than the diffusedbroadcasting scenario like acoustic and RF wave. It becomes more difficult to be eavesdropped.Furthermore, UWOC is much more energy efficient and cost-effective than its acoustic and RFcounterparts. Instead of using large and expensive acoustic and RF transceivers which are highlyenergy consuming, relatively small and low-cost optical underwater transceivers, such as laser diodesand photo diodes, can be implemented in UWOC systems. This benefit can improve the large scalecommercialization of UWOC, and accelerate the implementations of UWSNs.Although UWOC enjoys many advantages over the acoustic and RF methods, achieving UWOCremains as a challenging task. The main challenges of UWOC are listed as follows.a) Optical signal suffers from severe absorption and scattering. Although the wavelength of trans-mission light has been carefully selected in the blue and green spectrum [15] to minimize thetransmission attenuation coefficient, due to the inevitable photon interactions with the watermolecules and other particulate matters in water, absorption and scattering still severely atten-uate the transmitted light signal and cause multi-path fading. Due to the impact of absorptionand scattering, UWOC suffers from poor BER performance over a few hundred meters link dis-tance in turbid water environment. In underwater environment, matters such as chlorophyll arecapable of absorbing the blue and red lights. These matters and other colored dissolved organ-ic material (CDOM) can increase the turbidity of the water, and thus shrink the propagationdistance of the light. Moreover, the concentration of CDOM will also change with ocean depthvariations, thus change the corresponding light attenuation coefficients [30]. These undesirableimpacts will increase the complexity of UWOC systems.b) Underwater optical links will be temporarily disconnected due to misalignment of optical transceiver-s. In several UWOC systems, blue/green lasers or LEDs have been implemented as the light81.3. Thesis Organization and Contributionssources due to their narrow divergence feature; however, a precise alignment condition is required.As the underwater environment is turbulent at relatively shallow depths, link misalignment willtake place frequently, especially in the vertical buoy-based surface-to-bottom UWOC applica-tions. Random movements of sea surface will cause serious connectivity loss problem.c) Implementation of UWOC systems requires reliable underwater devices. The underwater en-vironment is complex. The flow, pressure, temperature and salinity of seawater will stronglyimpact the performance and lifetime of UWOC devices. Considering that no solar energy canbe exploited undersea and the long undersea operation time of UWOC devices, the reliability ofdevice batteries and efficiency of device power consumption are critical.In Table 1.1, we summarize the benefits and limitations of the three popular techniques choicesto achieve UWC.1.3 Thesis Organization and ContributionsThis thesis contains six chapters. A summary of each chapter and its contributions are presentedas follows.Chapter 1 presents several background knowledge and recent development of UWOC. As thedemand for high-speed underwater data transmission increases, UWOC becomes a promising alter-native technology for underwater applications such as oil exploration and submarine communica-tions. However, the optical signal suffers from severe intensity attenuation introduced by inevitableabsorption and scattering effects of water. To overcome these limitations and improve the UWOClink performance, it is imperative to study the characterizations of underwater channel, optimalmodulation and coding schemes, as well as the properties of transmission devices.Chapter 2 focuses on the channel modeling of UWOC. Firstly, we introduce several basic proper-ties of light propagation in water that constructs the fundamental of UWOC channel modeling. Sec-ondly, we present and classify four comprehensive aspects of UWOC channel modeling: LOS aquaticattenuation, NLOS aquatic attenuation, link misalignment, and channel turbulence. In each aspec-t, detailed channel modeling method and representative research works have been demonstrated.Finally, we summarize the related literatures on the topic of UWOC channel modeling.In Chapter 3, we study the channel modulation and coding techniques that can be appliedin UWOC. Since most UWOC systems are based on intensity modulation and direct detectionmechanisms, several conventional intensity modulation schemes, such as on-off keying and pulse91.3. Thesis Organization and ContributionsTable 1.1: Comparison of underwater wireless communication technologies [1].UWC technologies Benefits LimitationsAcoustic· Most widely used UWCtechnology· Long communicationrange up to 20 km· Low data transmission rate(on the order of kbps)· Severe communication latency(on the order of second)· Bulky, costly and energyconsuming transceivers· Harmful to some marine lifeRF· Relatively smooth transitionto cross air/water boundaries· More tolerant to waterturbulence and turbidity· Loose pointing requirements· Moderate data transmissionrate (up to 100 Mb/s) atvery close distance· Short link range· Bulky, costly and energyconsuming transceiversOptical· Ultra-high data transmissionrate (up to Gbps)· Immune to transmissionlatency· Low cost and small volumetransceivers· Can’t cross water/airboundary easily· Suffers from severe absorptionand scattering· Moderate link range(up to tens of meters)101.3. Thesis Organization and Contributionsposition modulation, have been widely implemented in both theoretical and experimental UWOCresearch. Classic forward error correction technologies, such as the Reed-Solomon code and theTurbo code, have also been embedded into many UWOC systems. We will briefly introduce thecharacterizations of each modulation and coding schemes and demonstrate their applications inUWOC.In Chapter 4, we study the recent development of experimental setups and prototypes of UWOC.Typical UWOC experimental testbeds that include different link configurations such as point-to-point LOS, diffused LOS, and NLOS are demonstrated. Besides these typical UWOC experiments,we also discuss several popular topics of experimental UWOC research which includes retroreflector-based UWOC, smart transceivers, UWOC for underwater vehicles and hybrid UWOC systems. Atthe end of this chapter, we also summarize and classify the contributions of experimental UWOCin recent years.Chapter 5 studies the outage performance of vertical buoy-based UWOC links using intensi-ty modulation/direct detection (IM/DD) on-off keying with zero and non-zero boresight pointingerrors. A closed-form expression for the outage probability with zero boresight pointing errors isachieved. We also derive the closed-form bounds for the outage probability with non-zero boresightpointing errors.Chapter 6 summarizes the entire thesis and states our contributions in this work. In addition,future work and potential research directions of UWOC are also suggested.11Chapter 2UWOC Channel ModelingIn this chapter, we will firstly present background knowledge related to light propagation prop-erties in the underwater environment. Then, UWOC channel modeling techniques, which includelink attenuation modeling, geometric misalignment modeling, and turbulence modeling, will bepresented. Finally, we will summarize this chapter and organize the related literatures in one table.2.1 Light Propagation in WaterCompared with terrestrial FSO communication channels, UWOC channels have several uniquecharacteristics. The existing terrestrial FSO channel models are not suitable for underwater environ-ment; therefore, new reliable channel models must be proposed and studied. In order to derive newchannel models for UWOC, we have to firstly understand the basic properties of light propagationin the underwater environment.According to Mobley’s statements in [5], the optical properties of water can be classified intotwo different groups: inherent optical properties (IOPs) and apparent optical properties (AOPs).IOPs can be understood as the optical parameters that only depend on the transmission mediumitself, more specifically the composition of that medium and particulate substances present withinit [22]. They are independent of the characterizations of light sources. The major IOPs of waterare the absorption coefficient, the scattering coefficient, the attenuation coefficient, and the volumescattering function [31]. AOPs, on the other hand, are known as the optical parameters that dependnot only on the the transmission medium itself, but also the geometrical structure of the light fieldsuch as diffusion and collimation [22]. The three major AOPs of water are radiance, irradianceand reflectance [31]. In a UWOC system, IOPs are typically used in determining communicationlink budgets, whereas AOPs are used to calculate ambient light levels for communication systemsnear the ocean surface [22]. Since IOPs have a greater impact on the link performance. In therest of this section, we will focus on IOPs. The details of AOPs that include their definitions andmeasurements can be found in [5, 31–34].122.1. Light Propagation in WaterPI PA PTPS△D△V(λ)θFigure 2.1: Geometry of inherent optical properties for a volume ∆V . Figure 2.1 is adapted from[5].Absorption and scattering coefficients are the two major IOPs that determine the underwaterlight attenuation. Absorption is an energy transfer process in which photons lose their energy andconvert it into other forms, such as heat and chemical (photosynthesis). Scattering is caused byvariations in the refractive index that changes the propagation direction of photons [35]. Generally,the impacts of absorption and scattering to a UWOC system can cause three undesirable effects.First, in the presence of absorption, the total propagation energy of light is continuously decreasing,which will limit the link distance of the UWOC. Second, in the presence of scattering, since thesize of optical aperture is finite, scattering will spread the light beam and result in a reductionof the number of photons collected by the receiver. This will lead to degradation of SNR of thesystem. Third, due to the light scattering in an underwater environment, each photon may arriveat the receiver panel in different time slots, and multi-path dispersions will occur. The undesirableimpacts of multi-path phenomenon include inter symbol interference (ISI) and timing jitter.In order to derive the absorption and scattering coefficients mathematically, we introduce thesimple model in Figure 2.1. We assume that a volume of water ∆V with thickness ∆D is illuminatedby a collimated light beam with wavelength λ. We denote the power of incident light as PI . Aportion of the incident light power PA is absorbed by water, and another portion of light power PS1is scattered. PT is the remaining light power that will propagate as desired. According to the law1The reflected light component is included in PS .132.1. Light Propagation in Waterof conservation, we get [5, 36]PI = PA + PS + PT . (2.1)Based on (2.1), we define the ratio between absorbed power and incident power PAPI as absorbance.Similarly, the fraction between scattered power and incident power PBPI as scatterance. The sub-sequent absorption coefficient and scattering coefficient are then calculated by taking the limit ofabsorbance and scatterance as water thickness ∆D becomes infinitesimally small [5, 36]a(λ) = lim∆D→0PAPI∆D, (2.2)b(λ) = lim∆D→0PSPI∆D. (2.3)In underwater optics, the overall attenuation effects of absorption and scattering can be describedby the attenuation coefficient 2 c(λ) which can be expressed as [38]c(λ) = a(λ) + b(λ). (2.4)The unit of attenuation coefficient is m−1. In addition, the author of [2] states that the underwaterlight absorption coefficient can be further represented as the summation of four absorption factors[2]a(λ) = aw(λ) + aCDOM (λ) + aphy(λ) + adet(λ) (2.5)where aw(λ) is the absorption due to pure seawater, aCDOM (λ) is the absorption due to CDOM,aphy(λ) denotes the absorption due to phytoplankton, and adet(λ) represents the absorption due todetritus.The absorption effect of pure seawater is introduced from two sources: the water molecules anddissolved salt in water such as NaCl, MgCl2, Na2SO4, and KCl [39]. Pure seawater is absorptiveexcept around a 400nm-500nm window, the blue-green region of the visible light spectrum. Thecorresponding absorption spectrum of pure seawater is shown in Figure 2.2(a).CDOM 3 refers to colored dissolved organic materials with dimensions smaller than 0.2 mm[40]. In Figure 2.2(b), it shows that the CDOM presents highly absorptive to blue wavelengths(420nm-450nm) and less absorptive to yellow and red light [41].The absorption effects due to phytoplankton are mainly caused by photosynthesising of chloro-phyll. For different phytoplankton species, the characteristics of the absorption effect are also2Also known as extinction coefficient in some optical literatures such as [5, 37]3In several optical literatures, it’s also represented as gelbstoff, yellow substances or gilvin.142.1. Light Propagation in Waterdifferent [42]. Figure 2.2(c) shows a typical absorption coefficient profile shared by all species. Wecan observe that the aphy(λ) shows a high absorption in the 400-500 nm region and a further peakat about 660 nm.Detritus includes living organic particles, such as bacteria, zooplankon, detrital organic matterand suspended inorganic particles such as quartz and clay. These substances are grouped togetherdue to their similar absorption behaviour [43]. Figure 2.2(d) shows a absorption curve similar tothat if CDOM.The scattering coefficient for underwater light propagation can also be presented as a summationof different scattering factorsb(λ) = bw(λ) + bphy(λ) + bdet(λ) (2.6)where bw(λ) is the scattering due to pure seawater, bphy(λ) denotes the scattering due to phytoplank-ton, and bdet(λ) represents the scattering due to detritus. Compared with absorption, scattering isrelatively independent of wavelength. The dominant factor that impacts scattering is the densityof particulate matters.In pure seawater, since the refractive index will change with the variations of flow, salinity andtemperature, the scattering coefficient will also change. Compared with the size of water molecules,the wavelength of light is relatively large, thus the Rayleigh scattering model can be used to describethe scattering induced by pure seawater. The corresponding scattering spectra is shown in Figure2.3(a).Phytoplankton and detritus account for more than 40% of the total scattering effects [44]. Sincethe scattering light caused by phytoplankton and detritus propagates mainly in the forward direc-tion, Mie scattering model can be used to approximate these two types of scattering [5]. In practice,the exact scattering coefficients highly depends on the density of phytoplankton and detritus [45].In Figure 2.3(b) and Figure 2.3(c), we present the scattering spectra due to phytoplankton anddetritus with different densities. A summary of the above discussion on seawater absorption andscattering characteristics is presented in Table 2.1.152.1. Light Propagation in WaterWavelength (nm)200 300 400 500 600 700 800Absorption coefficient of pure seawater (/m)00.511.522.533.5Absorption spectra of pure seawater(a) Absorption spectra of pure seawater.Wavelength (nm)200 300 400 500 600 700 800Absorption coefficient of CDOM (/m)01234567Absorption spectra of CDOM(b) Absorption spectra of CDOM.Wavelength (nm)400 450 500 550 600 650 700Absorption coefficient of phytoplankton (/m)00.20.40.60.811.2Absorption spectra of phytoplankton(c) Absorption spectra of phytoplankton.Wavelength (nm)200 300 400 500 600 700 800Absorption coefficient of detritus (/m)00.511.522.53Absorption spectra of detritus(d) Absorption spectra of detritus.Figure 2.2: Optical absorption spectra for different ocean components. The data in Figure 2.2 isfrom [6–8].162.1. Light Propagation in WaterWavelength (nm)200 300 400 500 600 700 800Scattering coefficient of pure seawater (/m)00.020.040.060.080.10.120.140.160.18Scattering spectra of pure seawater(a) Scattering spectra of pure seawater.Wavelength (nm)200 300 400 500 600 700 800Scattering coefficient of phytoplankton (/m)00.511.522.5Scattering spectra of phytoplanktonDensity of phytoplankton 0.05Density of phytoplankton 0.5Density of phytoplankton 1.0Density of phytoplankton 5.0(b) Scattering spectra of phytoplankton.Wavelength (nm)200 300 400 500 600 700 800Scattering coefficient of detritus (/m)00.20.40.60.811.2Scattering spectra of detritusDensity of detritus 0.5Density of detritus 1.0Density of detritus 2.0Density of detritus 3.0(c) Scattering spectra of detritus.Figure 2.3: Optical scattering spectra for different ocean components. The data in Figure 2.3 isfrom [6–8].172.1. Light Propagation in WaterTable 2.1: Summary of absorption and scattering characteristics of seawater [2]Compositions Absorption coefficient Scattering coefficientWaterInvariant at constanttemperature and pressure.Strongly depends on λRayleigh scattering.Small variance comparedwith absorption.Strongly depends on λSea saltsNegligible in visible spectrum.Increase towards short λRayleigh scattering.Doesn’t depend on λCDOMVariable with the density ofCDOM.Increase towards short λNegligiblePlankton and detritusVariable with the densityof plankton and detritus.Increase towards short λMie scattering.Variable with the densityof plankton and detritus.Increase towards short λBased on the attenuation coefficient that has been introduced, Beer-Lambert law provides thesimplest and most widely used scenario to describe the light attenuation effects in underwaterenvironment [46] asI = I0e−c(λ)z (2.7)where I0 is the power of transmitted light, z denotes the light transmission distance, I representsthe power of light after transmitting z distance, and c(λ) stands for the attenuation coefficient. Theexact value of attenuation coefficient c(λ) will change with different water types and water depth.The typical values of a(λ), b(λ), and c(λ) associated with four major water types are given in Table2.2 [5, 11, 37, 47]. In pure seawater, absorption is the main limiting factor, the low scattering coeffi-cient makes the beam free from divergence. In clear ocean waters, there is a higher concentration ofdissolved particles that affects scattering. In coastal ocean water, high concentrations of plankton,detritus and minerals are the dominant sources of absorption and scattering. Turbid harbor waterhas the highest concentration of dissolved and in-suspension matters, which will severely attenuate182.1. Light Propagation in Waterthe light propagation [37]. More details of water types and variations of attenuation coefficient withother parameters such as depth, pressure, and salinity can be found in [5, 30, 44, 48, 49].Table 2.2: Typical values of a(λ), b(λ), and c(λ) for different water typesWater types a(λ) (m−1) b(λ) (m−1) c(λ) (m−1)Pure sea water 0.053 0.003 0.056Clear ocean water 0.114 0.037 0.151Costal ocean water 0.179 0.219 0.298Turbid harbor water 0.295 1.875 2.17From (2.4) and (2.7), we know that the Beer lambert’s law contains two implicit assumptions.First, the transmitter and receiver are perfectly aligned. Second, all the scattered photons are losteven though in reality some of the scattered photons can still arrive at the receiver after multiplescattering events. This assumption severely underestimates the received optical power, especiallyin the scattering dominant situation. In order to describe the scattering effects more accurately,another important IOP volume scattering function (VSF) is introduced. It is defined as [36]β(θ, λ) = lim∆D→0lim∆Ω→0PS(θ, λ)∆D∆Ω(2.8)where PS(θ, λ) is the fraction of incident power scattered out of the beam through an angle θ into asolid angle ∆Ω centered on θ (Figure 2.1). VSF is the scattered intensity per unit incident irradianceper unit volume of water. In the view of physics, the VSF can also be interpreted as the differentialscattering cross section per unit volume [36].Integrating β(θ, λ) over all directions (solid angles) gives the scattering coefficient [36]b(λ) =∫β(λ, θ)dΩ = 2pi∫ pi0β(λ, θ) sin(θ)dθ. (2.9)Normalizing (2.8) with the scattering coefficient, we obtain the scattering phase function (SPF),which is defined as [36]β˜(θ, λ) =β(λ, θ)b(λ). (2.10)The scattering phase function is also an important IOP. Considering the difficulty of measuringscattering phase function (SPF), the Henyey-Greenstein (HG) function is commonly introduced to192.2. Modeling of Aquatic Optical Attenuation in UWOCpresent the SPF as [50–54]β˜(θ, λ) = PHG(θ, g) =1− g24pi(1 + g2 − 2g cos θ) 32(2.11)where g is the average cosine of β in all scattering directions.To this end, we have introduced the concept of absorption and scattering coefficients, Beer Lam-bert’s law, as well as VSF. These concepts provide a theoretical basis for more complex UWOCchannel models [37]. In a UWOC link, the optical signal launched from the transmitter will experi-ence various losses before reaching the receiver. They include system loss introduced by transceivers,link loss results from water attenuation, geometric misalignment, and water turbulence. Since theloss introduced by the transceiver is mainly characterized by device parameters and design specifi-cations, it is challenging to characterize the loss in a comprehensive and uniform approach. Thus,in Sections 2.2, 2.3 and 2.4, we will focus on the modeling techniques of the aforementioned lossesin UWOC links.2.2 Modeling of Aquatic Optical Attenuation in UWOCAs we have presented in Section 2.1, without considering the link configuration, transceiverarchitecture and alignment condition, the two major IOPs that will attenuate light propagation inUWOC systems are absorption and scattering. Thus, the modeling of UWOC aquatic optical linkattenuation can be regarded as a task that accurately describes the absorption and scattering effectswith specific link configurations. In the remaining of this section, we will introduce the models ofaquatic optical attenuation in two categories: LOS configuration and NLOS configuration.2.2.1 Aquatic Optical Attenuation in LOS ConfigurationFor simplicity, several researchers utilized the Beer Lambert’s law to model LOS UWOC. In[55] and [56], the authors evaluated the performance of a UWOC system based on Beer Lambert’slaw in different water types and different communication ranges. The impacts of environmentalvariability, such as variations of refractive index with depth, were taken into account.Another general theoretical model of aquatic optical attenuation in UWOC is radiative transferequation (RTE). As we have presented in Section 2.1, the VSF is an important IOP that describesthe scattering characterizations of photons. However, the VSF is difficult to be measured in practice[57]. Furthermore, the VSF can only determine the scattering properties of a single photon at one202.2. Modeling of Aquatic Optical Attenuation in UWOCsingle refractive index condition. It’s not suitable to model the scattering properties of large numberof photons [35]. Considering these two facts, most UWOC researchers employ RTE in their UWOCchannel modeling research. Without considering the temporal dispersion of light, the typical two-dimensional RTE can be expressed as [58–60]~n · ∇L(λ,~r, ~n) = −cL(λ,~r, ~n) +∫2piβ(λ, ~n, ~n′)L(λ,~r, ~n)d~n′ + E(λ,~r, ~n) (2.12)where ~n is the direction vector, ∇ is the divergence operator, L(λ,~r, ~n) denotes the optical radianceat position ~r towards direction ~n, β(λ, ~n, ~n′) is the VSF, and E(λ,~r, ~n) represents the source radi-ance. RTE is capable of describing the energy conservation of a light wave that is passing througha steady medium [59, 60]. The derivations of RTE are complex and lengthy, and they can be foundin [36] and [61]. The RTE can be solved both analytically and numerically. Since the RTE is anintegro-differential equation involving several independent variables [58, 60], it is difficult to findan exact analytical solution. Thus only few analytical RTE models have been proposed in recentyears. In [62], Jaruwatanadilok devised an analytical solution of RTE employing the modified Stokesvector. This model takes both multiple scattering and light polarization effects into account. Basedon this model, numerical results show that the ISI and BER are as functions of data rate and linkdistance. This finding can be further used to predict several performance parameters of UWOCsystems such as the maximum communication distance with certain data rate and BER. In [11] and[63], Cochenour et al. proposed a beam-spread function for laser-based UWOC by solving the RTEanalytically. The small angle approximation was performed to simplify the derivation. This ana-lytical model reveals the relationship between received optical power versus link range for varioustransmitter/receiver pointing accuracies. It was also validated through watertank experiments.Besides utilizing analytical solutions, numerical methods are preferred to solve the RTE. In fact,for many practical UWOC applications, finding an exact analytical solution of RTE is even morechallenging [58]. Moreover, since a series of assumptions and approximations have been made tosimply the RTE, the analytical solutions will also suffer from numerous limitations [64]. In view ofthis, most of the researchers focused on developing powerful numerical RTE solvers [46, 60]. Themost popular numerical approach to solve RTE is Monte Carlo simulation. It is a probabilisticmethod to mimic the loss of underwater light propagation by sending and tracking large number ofphotons [65, 66]. The Monte Carlo method benefits from its easy programming, accurate solutionand high flexibility, but it also suffers from random statistical errors and low simulation efficiency[36]. In [67], Leathers et al. from the U.S. Naval Research Laboratory (NRL) reported a practical212.2. Modeling of Aquatic Optical Attenuation in UWOCguide to generate Monte Carlo computer simulations for typical ocean optics applications. Thismethod has been referred by many other UWOC researchers and has been proved to be robust.In recent years, lots of researchers have employed Monte Carlo approach to solve the RTE orstudy the characterization of UWOC channels. In [68], Li et al. built a Monte Carlo simulator tomodel the impulse response of UWOC channel. Within this simulator, several receiver parameterssuch as aperture size and field of view (FOV) were taken into account. The authors utilized thisMonte Carlo simulator in order to evaluate the channel capacity of a UWOC system with differentlink distances, water conditions, and transceiver parameters [69]. Simulation results indicate thatthe bandwidth of UWOC for clean water, coastal water and harbor water are on the order ofhundreds of MHz, tens of MHz and MHz respectively [69]. Chadi et al. from Institut Fresnelutilized a Monte Carlo approach to solve the RTE and provided a channel model that can be usedto appropriately predict different design parameters of UWOC systems [37]. As a continuance of[37], the authors in [50] proposed a channel impulse response of UWOC system by solving the RTEthrough Monte Carlo simulation. The authors quantified the channel time dispersion for differentwater types, link distances, and transmitter/receiver characteristics. A two-dimensional HG phasefunction was employed to model the VSF asPTTHG(θ) = αPHG(θ, gFWD) + (1− α)PHG(θ,−gBKWD) (2.13)where PHG(·, ·) is the HG function defined in (2.11); α is the weight of the forward-directed HGfunction; and gFWD and gBKWD are the asymmetry factors for the forward- and backward-directedHG phase functions, respectively [50]. Based on this numerical channel model, the authors con-cluded that the channel time dispersion can be neglected when operating at a moderate distance(20m) in a clean water environment. However in highly turbid water, the channel time dispersioncan impact the data transmission when operating over a large distance (100m). Based on thisconclusion, the system will experience less ISI in the received signal when the transmission distanceis short and the water is clear. As a result, complex signal modulation and demodulation can beavoided. In order to validify the Monte Carlo approach for UWOC channel modeling, Frank etal. made a comparison between the results of Monte Carlo simulation and laboratory experiments[47]. The results of the Monte Carlo simulation and the water-tank experiment exhibited reasonableagreement. Up to one Gbps data rate was achieved in a two-meter long water pipe. In [70], theauthors employed Monte Carlo approach to solve the RTE and calculate the impulse response for aUWOC system over different operation environments. Another similar comparison between Monte222.2. Modeling of Aquatic Optical Attenuation in UWOCCarlo simulation and experimental measurements can also be found in [51]. The authors deviseda numerical Monte Carlo simulation tool that is capable of computing received power of a UWOCsystem by considering the receiver aperture size, FOV, and pointing-tracking losses. This simulatoris based on modeling a complex probability density function (PDF) (such as the lightfield distribu-tion underwater) by its known individual components (such as the scattering distance of photons inwater) [51]. By randomly sampling these known processes, the unknown PDF can be approximatedusing these discrete samples [51]. The accuracy of this simulator was validated through comparingthe simulation results with the experimental data from [11, 71]. The author in [72] also made theMatlab source code of this simulation tool available to the public.Besides the probabilistic Monte Carlo approach, there are also two deterministic methods thatcan be used to solve the RTE numerically: the discrete ordinates method and the invariant imbed-ding method [5, 46]. But only few researchers employed these two approaches as alternatives ofMonte Carlo simulation. In [59] and [60], Li et al. developed an efficient RTE solver based onthe deterministic numerical approach. This solver employs the matrix free Gauss-Seidel iterativemethod in order to calculate the received power of UWOC systems. It can also process highlyforward peaked VSF that can not be handled well by the discrete ordinates approach. Accordingto the simulation results, this method can achieve the same accuracy as the Monte Carlo approachbut with a much shorter simulation time. The referred Matlab source code of this method can befound in the appendix of [59].The majorities of aquatic optical attenuation models for UWOC are based on solving the RTE.However, instead of solving RTE, several stochastic models have also been proposed from the prob-abilistic nature of photon trajectory. In [73], Zhang et al. from Tsinghua University demonstrateda stochastic channel model to represent the spatial-temporal probability distribution of propagatedphotons for non-scattering and single scattering 4 components of UWOC links. The authors adoptedthe HG function as the probability density function of light scattering angle to simplify the analysis.The proposed stochastic model also exhibited reasonable agreement with the numerical results ofMonte Carlo simulation. Based on [73], the same research group further proposed a more generalstochastic UWOC channel model in [74] by taking into account of all three components of propa-gated photons, which include non-scattering, single scattering and multiple scattering5 components.4Single scattering components refer to photons that experience only one scattering event during the propagationfrom source to destination.5Multiple scattering components refer to photons that experience more than one scattering event during thepropagation from source to destination.232.2. Modeling of Aquatic Optical Attenuation in UWOCThis comprehensive channel model fits well with the Monte Carlo simulations in turbid water en-vironment, such as in coastal or in harbor waters. Following the similar stochastic approach of [73]and [74], the Tsinghua researchers also presented a closed-form expression for the angle of arrival(AOA) distribution in [75]. This AOA model characterizes how the received intensity of ballisticand single scattering components is distributed over AOA with respect to unit transmission power[75]. Numerical results have validated the proposed AOA distribution by Monte Carlo approach inclear and turbid coastal and harbor water with relatively short link range.Semi-analytical modeling approach has also been employed by several UWOC researchers. In[76] and [77], based on the results of a Monte Carlo simulation, Tang et al. adopted a closed formdouble-Gamma function to represent the channel impulse response of the UWOC [76] ash(t) = C1∆te−C2∆t + C3∆te−C4∆t, (t ≥ t0) (2.14)where ∆t = t − t0. t is the time scale and t0 = L/v is the propagation time which is the ratio oflink range L over light speed v in water [76]. The parameter set (C1, C2, C3, C4) in (2.14) can becomputed from Monte Carlo simulation results as [76](C1, C2, C3, C4) = arg min(∫[h(t)− hmc(t)]2 dt)(2.15)where h(t) is the double Gamma functions model in (2.14) and hmc(t) is the Monte Carlo simulationresults of impulse response; arg min(·) is the operator to return the argument of the minimum. Eq.(2.15) can be solved through a numerical curve fitting approach [76]. This semi-analytical impulseresponse is capable of describing the temporal dispersion of light in turbid underwater environments.It can be used to carry out a performance evaluation for calculating the BER and 3-dB channelbandwidth of a UWOC system. As an extension of [76] and [77], the authors applied a similarcurve fitting approach to derive the the impulse response for LOS UWOC links with multiple-inputmultiple-output (MIMO) configuration [78]. Weighted double Gamma functions have been derivedas the impulse response of a 2-by-2 LOS MIMO UWOC system in turbid water environment.During the past ten years, a lot of research has focused on UWOC aquatic optical attenuationmodeling. However, to this date, only a few models are capable of providing an end-to-end simulatorfor the UWOC designers [64]. There’s still several “barriers” between the UWOC channel modelersand hardware engineers [64]. In [79], Doniec et al. presented an end-to-end model that can simulatethe signal strength and communication distance in any propagation directions. This generic modelincorporates all the components of a UWOC system that includes information of light source,242.2. Modeling of Aquatic Optical Attenuation in UWOCdetectors, amplifiers, and analog-to-digital converters [79]. The authors also verified this modelthrough an autonomous underwater optical robotic system. Since this model takes into accountall the relevant components of a UWOC system as well as the attenuation properties of water,it provides a direct and complete reference for UWOC designers to estimate the overall systemperformance.2.2.2 Aquatic Optical Attenuation in NLOS ConfigurationAs shown in Figure 1.3(d), in NLOS implementations, transceivers can utilize reflection ofthe sea surface to overcome link obstacles. Compared with channel modeling of LOS UWOC,investigations of NLOS UWOC channel modeling have received less attention. Light propagation inNLOS configuration experiences the same attenuation effects as in LOS configuration. The majordifference between LOS and NLOS channels is the reflection effects introduced by wavy sea surface.Thus accurately describing the reflection effect of sea surface is considered as the most critical partof NLOS channel modeling. Several models that describe the slopes of random sea surface can befound in [10, 80, 81]. Similar to channel modeling work of LOS configuration, channel models ofNLOS link can also be derived both analytically and numerically. To the best of our knowledge,most channel models of NLOS configuration were derived through numerical approaches such asthrough Monte Carlo simulations. As an example of an analytical approach, Shlomi et al. in[23] and [24] proposed a novel concept of NLOS UWOC network. Each node inside this networkcan communicate with each other through reflection at the ocean-air interface. Communicationfrom one single node to multiple nodes can also be achieved. The authors derived a mathematicalmodel for the NLOS channel by considering the link attenuation, sea surface slopes and receiverFOV. Numerical simulation was also performed to test the validity of this NLOS UWOC channelmodel. Simulation results show that an increase in node separation distance dramatically increasesthe BER of the NLOS UWOC system. By applying the numerical Monte Carlo method, theauthors of [25] proposed a path loss model for NLOS UWOC links. The effects of both randomsea surface slopes and scattering properties of seawater have been taken into account. Numericalresults suggest that the random surface slopes induced by wind or other turbulent sources maystrongly corrupt the received signal. However, this effect can be alleviated when the received signalcontains multiple dominant scattering light components. In [82] and [83], Jagadeesh et al. proposedan impulse response for NLOS UWOC based on Monte Carlo simulation. A two-dimensional HGangle scattering function was employed in this simulation process in order to model the multiple252.3. Modeling Geometric Misalignment of UWOCscattering effects of light. Based on this impulse response, the authors also evaluated the systemperformance with different water types and receiver FOV.2.3 Modeling Geometric Misalignment of UWOCAs introduced in Section 1.2, the undiffused point-to-point UWOC links suffer from temporalmisalignment. This undesired effect will degrade the system performance and induce temporalcommunication interruptions. In fact, link misalignment is unavoidable in any UWOC systems,and there are three major reasons that will tighten the system alignment requirements.a) Limitations of transceivers: In order to achieve a higher data rate and longer communicationrange, many UWOC applications utilize the laser diode and photo diodes as transmitters andreceivers respectively. However, due to the narrow divergence angle of laser diodes and limitedFOV of photo diodes, these UWOC systems require precise alignment.b) Relative motions caused by underwater vehicles, ocean current, and other turbulent sources:UWOC links suffer from severe misalignment when communicating with an AUV or ROV. Sincethe AUV or ROV keeps moving, the transceivers should always keep tracking with each other.Thus, link misalignment is more likely to occur. Ocean currents and wind can introduce randommovements of transceivers in underwater environment, possibly causing link interruptions.c) Variations of refractive index: The refractive index will change with water depth, temperature,salinity, and other environmental conditions. This phenomenon usually occurs in surface-to-bottom UWOC links and will cause the non-straight light propagation, thus aggravate linkmisalignment of UWOC [30].Similar to the modeling work of UWOC aquatic optical attenuation, both analytical and numer-ical methods can be implemented in the modeling of UWOC link misalignment. For the analyticalcases, without focusing on the pointing error caused by slight jitter of the transceivers, the au-thors in [84] employed the beam spread function to model the link misalignment when the receiverdeviates in a larger region [11, 84]BSF (L, r) =E (L, r) exp(−cL) +∫ ∞0E (L, v) exp(−cL)×{exp[∫ L0bs (v(L− z)) dz]− 1}J0(vr)vdv(2.16)262.3. Modeling Geometric Misalignment of UWOCwhere BSF (L, r) is the irradiance distribution of the receiver plane; E(L, r) and E(L, v) are theirradiance distributions of the laser source in spatial coordinate system and spatial frequency do-main, respectively [11]; L presents the distance between the source and the receiver plane; r is thedistance between the receiver aperture center and the beam center on the receiver plane which isassumed to be perpendicular to the beam axis; b and c are the attenuation and scattering coeffi-cients respectively; s(v) is the scattering phase function. Through this model, the authors evaluatedthe BER performance of UWOC under misalignment condition. Numerical results indicated that,regardless of water type, an appropriate amount of misalignment will not cause severe performancedegradation with sufficiently large transmission power. A similar conclusion was also drawn fromthe experiment of [85]. As an extension of [84], Dong et al. in [10] have presented a model of randomsea surface slopes that concerns the link pointing errors caused by slight jitter of the transceiversfor a vertical buoy-based UWOC system. The PDF of random sea surface slopes is expressed as[10]P (sx, sy) =12piσuσcexp[−(s2x2σ2u+s2y2σ2c)](2.17)where sx = ∂z/∂x and sy = ∂z/∂y are defined as wave slopes of up/downwind and crosswinddirections in the Cartesian coordinate (x, y, z) respectively; σ2u and σ2c denote the mean square slopein the up/downwind and crosswind directions, respectively. The authors employ this model andbeam-spread function to evaluate the BER performance of the system. Numerical results suggestthat the BER deteriorates as the pointing errors increase. This performance degradation can bereleased through an increment in the seawater turbidity. Zhang et al. have also employed a similarPDF of random sea-surface slope as [10] but in the form of angle to model the pointing errors of buoy-based downlink UWOC systems [86]. The authors utilized this model and evaluated the channelcapacity of downlink buoy-based UWOC multiple-input multiple-output systems. Numerical resultssuggest that more turbid water, larger link range and larger inter-spacing may reduce the channelcapacity, and meanwhile more turbid water and larger link range can weaken the effects of randomslopes on the channel capacity [86].Numerical methods have also been employed to model UWOC link misalignment. By using aMonte Carlo approach, the authors of [87] studied the impact of link misalignment on the receivedpower of a point-to-point LOS UWOC system. This numerical model was validated through water-tank experiments. Since misalignment effects in LOS UWOC can also be caused by variations ofrefractive index. In [30] and [49], Laura et al. proposed a profile of the refractive index with thevariation of depth. This profile was then used in a numerical ray-tracing simulation in order to272.4. Modeling Link Turbulence of UWOCevaluate the tolerence of UWOC link offset. From the numerical results, a 0.23m link offset wasallowed for 500nm laser with 0.57 degree FOV and 200m link distance. This work provides aneffective reference for modeling link misalignment induced by refractive index variations [88].2.4 Modeling Link Turbulence of UWOCTurbulence in UWOC is defined as the event that makes water experience rapid changes inthe refractive index [22]. This phenomena is commonly caused by ocean currents which will in-duce sudden variations in the water temperature and pressure. Most studies on UWOC channelmodeling have focused on providing an accurate description of absorption and scattering effects,but the impact of underwater optical turbulence is commonly ignored. In fact, underwater opticalturbulence can also cause considerable degradation to the performance of a UWOC system. Thus,more attention should be paid to this subject. The modeling of underwater optical turbulence ismainly based on research results of atmospheric optical turbulence channel models in free-space op-tical communications. Considering the similarity of underwater optical turbulence and atmosphericoptical turbulence, several researchers directly applied or modified the classical atmospheric opticalturbulence models as the underwater optical turbulence models. In [89], Hanson et al. adopted theKolmogorov spectrum model to present the underwater optical turbulence. Motivated by [89], theauthors of [90] characterized a UWOC channel model that includes both scattering/absorption andunderwater optical turbulence. The proposed underwater optical turbulence model was adoptedfrom the classical lognormal turbulence model used in FSO communication [90]fI(I) =1√2piσtIexp(−(ln I − µ)22σ2t), I > 0 (2.18)where I is the received optical irradiance, µ is the mean logarithmic light intensity, and σ2t is thescintillation index. Based on this channel model, the authors developed a single-input-multiple-output (SIMO) UWOC system. Monte Carlo numerical simulation was performed to evaluate thesystem’s BER performance. Numerical results show that the SIMO scheme can effectively reducechannel fading and extend communication range. Instead of using classic turbulence models ofFSO communication, the authors of [91] employed oceanic turbulence model from [92] and [93] toinvestigate the temporal statistics of irradiance in the moving ocean with weak turbulence. Theyderived a closed-form expression that described the relationship of temporal correlation, propagationdistance, and average velocity for the moving medium. Based on this expression, the authors alsoevaluated the temporal correlation of irradiance in specific turbulent ocean environments. Numerical282.5. Summaryresults show that the velocity of ocean flow is the dominant factor that causes turbulence and affectstemporal statistics of irradiance in a UWOC system [91].2.5 SummaryIn this chapter, we presented several background knowledge on light propagation properties inthe underwater environment. The concept of absorption, scattering, and VSF were introduced.Then, we demonstrated the UWOC channel modeling techniques in three aspects: link attenua-tion modeling, geometric misalignment modeling, and turbulence modeling. For link attenuationmodeling, we presented the RTE and two viable methods. For geometric misalignment modelingand turbulence modeling, we explained the corresponding modeling mechanisms. At the end of thischapter, we summarized the literature related to channel modeling of UWOC in Table 2.3.292.5. SummaryTable 2.3: Summary of literatures on UWOC channel modelingUWOC channel models Related references CommmentsBeer Lambert’s law [55, 56, 94]Simplest UWOC channel modelwithout considering temporaldispersion.Chlorophyll-based Monte-Carlo model [30], [42], [53, 95, 96]Chlorophyll concentrationseverely affect underwater lightattenuation.Analytical RTE model [11, 36, 61–63, 68] Difficult to be solved.Numerical RTE model[47], [37], [46], [58], [59], [36],[65–67], [69], [50], [70]Monte Carlo simulations arewidely used.Analytical Stochastic model [73]Derived from the nature ofphoton trajectory in the senseof probability.Numerical Stochastic model [51, 72, 76–78]Open source simulator in [72].[77] implemented MIMO.End-to-end system model [79]Incorporates all the componentsof a UWOC system.NLOS model [10, 24, 25, 80–83, 97][10], [80], and [81] illustratedprobability model of sea surfaceslopes.Misalignment model [10, 49, 84, 85, 87, 88]Includes transceivers limitations,relative motions, and variationsof refraction index.Turbulence model [89–93]Turbulence models of FSOcommunications can be applied.30Chapter 3UWOC Channel Modulation andCoding TechniquesIn this chapter, we will first give a brief introduction of several digital modulation techniquesthat implemented in UWOC systems. The advantages and limitations of each modulation schemewill be presented. Then, we will discuss the channel coding techniques of UWOC. Finally, wewill summarize this chapter and classify the related literatures on UWOC channel modulation andcoding topics.3.1 Modulation Schemes of UWOCUWOC channel modulation techniques have attracted much attention in recent years due to itscapability to impact the system performance considerably. Since UWOC can be regarded as imple-menting FSO in underwater environment, the conventional intensity modulation (IM) techniquesthat used in FSO communication systems can also be applied to UWOC systems. On-off keying(OOK) modulation is the most popular and simplest IM scheme in FSO communication system.This modulation scheme can also be implemented in UWOC systems. The OOK modulation is abinary level modulation scheme. During an OOK transmission, an optical pulse which occupiespart of or entire bit duration represents a single data bit “1”. On the other hand, the absence ofan optical pulse represents a single data bit “0”. There are two pulse formats in OOK modulationscheme: return-to-zero (RZ) format and non-return-to-zero (NRZ) format. In the RZ format, apulse with duration that only occupies a part of the bit duration is defined to present “1”; however,the pulse duration occupies the whole bit duration in the NRZ scheme. The RZ-OOK has beenproved having higher energy efficiency than the NRZ-OOK, but at the expense of consuming morechannel bandwidth. Due to the severe absorption and scattering effects in underwater environment,the transmitted OOK signal suffers from various channel fading. In order to alleviate these impactsand achieve an optimal OOK signal detection, dynamic threshold (DT) techniques can be applied313.1. Modulation Schemes of UWOC000 001 100 101 OOK PWM PPM DPIM Binary BitsFigure 3.1: Illustration of OOK, PWM, PPM and DPIM. Figure 3.1 is adapted from [9].323.1. Modulation Schemes of UWOCin most UWOC OOK receivers. The DT is determined based on the estimation of channel fading.Several channel estimation techniques of FSO communication systems such as pilot symbol method,symbol-by-symbol maximum likelihood (ML) method, and ML sequence method [98] can also beemployed by the UWOC OOK systems. The two major drawbacks of UWOC OOK scheme arelow energy efficiency and spectral efficiency. But considering its simplicity, OOK modulation is stillthe most popular IM scheme in UWOC. It has been implemented in a number of theoretical andexperimental UWOC research works [62, 99, 100].Pulse position modulation (PPM) scheme is another popular modulation technique used inUWOC systems. Compared with OOK modulation, PPM has much higher energy efficiency anddoesn’t require dynamic thresholding, but at the expense of lower bandwidth utilization rate andmore complex transceivers. In PPM, every transmitted M bits will be modulated as one singlepulse in one of 2M time slots, and the pulse position represents the transmitted information (Figure3.1). The main drawback of PPM modulation is the tight timing synchronization requirement.Any timing jitters or asynchronization will severely degrade the BER of system. In recent years,several researchers have studied the performance of PPM scheme over UWOC channel models. Theauthors of [101] have investigated the performance of 4-PPM scheme over numerical RTE channelmodel. They have found that the corresponding BER for PPM scheme is almost equal to thatof OOK modulation and with much higher energy and spectrum efficiency. More complex PPMsuch as 8-PPM or 16-PPM can be used to improve higher bandwidth efficiency. In [102], Sui etal. proposed a modified PPM scheme for UWOC. This modified PPM can maintain the similarpower efficiency and anti-noise performance as the conventional PPM. It also has improved thebandwidth utilization rate of the system. Besides theoretical studies, PPM was also applied inseveral experimental UWOC implementations. The related literatures can be found in [103–109].Similar to PPM, pulse width modulation (PWM) also utilizes the relative positions of pulses torepresent data symbols. In L-ary PWM, optical pulses will only appear in the first L consecutivetime slots to represent one symbol, where L is equal to the decimal of symbol bits (Figure 3.1). Sincethe PWM extends the total pulse time during the transmission of one symbol, the peak transmissionpower of each pulse is reduced (Figure 3.1). The PWM scheme also benefits from better spectralefficiency and stronger resistance to ISI. However, these two advantages will be counterbalanced byhigher average power requirements that increase with number of slots per symbol [98].Digital pulse interval modulation (DPIM) is also widely implemented in UWOC. In this modu-lation, an “On” optical pulse slot is sent and followed by a number of “Off” slots. The number of333.1. Modulation Schemes of UWOC“Off” slots depends on the decimal value of the transmitted symbol, and an additional guard slotis commonly added in order to avoid sending consecutive “On” pulses (Figure 3.1) [9]. Comparedwith PPM and PWM which require slot and symbol level synchronization, digital pulse intervalmodulation is an asynchronous modulation scheme with variable symbol length [98]. Furthermore,with variable symbol length, DPIM also has higher spectral efficiency than PPM and PWM [98].The most critical problem of DPIM is the error spread in demodulation. From Figure 3.1, we noticethat if an “Off” slot is demodulated as “On”, then all the succeeding symbols will also be wrong.Applications of DPIM can be found in several UWOC applications of ROVs and AUVs such as[110–112].Similar to the comparison of IM schemes that we have made, the authors of [9] also performeda comparison study on different IM techniques for UWOC. This comparison included OOK, PPM,PWM, DPIM and other derivative IM schemes such as multi-pulse PPM (MPPM) and differentialPPM (DPPM). Simulation results show that with the same link distances, PPM is the most energyefficient modulation scheme. DPIM has better bandwidth efficiency over PPM and OOK but at theexpense of more complex demodulation devices. Other similar comparison of IM for UWOC canalso be found in [108] and [113].Coherent modulation schemes have also been implemented in several UWOC systems. In con-trast to the direct IM schemes, coherent modulations encode the information on the amplitude,polarization or phase of optical carriers. At the receiver side, the same synchronized optical carrierwill mix with the received optical signals and accomplish demodulation. Compared with IM, coher-ent modulations benefit from higher receiver sensitivity, higher system spectral efficiency and betterrejection on background noise, but at the expense of higher implementation complexity and highercost [98]. Due to the high dispersion effect of seawater, coherent modulation at optical frequenciesis difficult to be achieved in UWOC systems. In order to overcome this limitation, intensity modu-lation has to be imposed on the pre-modulated signals [114]. Typical coherent modulations used inUWOC systems include quadrature amplitude modulation (QAM), phase shift keying (PSK), andpolarization shift keying (PolSK).In [114], Cochenour et al. presented an experimental comparison study of binary PSK (BPSK),quadrature PSK (QPSK), 16-QAM and 32-QAM in a UWOC system. The authors evaluated thelink performance for different coherent modulations with different levels of water turbidity. A sum-marization of constellation diagrams for each modulation techniques was demonstrated. Similarly,the authors of [115] compared the coherent PSK, frequency shift keying modulations with several343.2. Channel Coding of UWOCIM schemes such as OOK and PPM. Simulation results have demonstrated that PSK modulationperforms the best over other modulation schemes in terms of data rate and BER. But it also suffersfrom poor power efficiency. A binary PolSK (BPolSK) modulation for UWOC has been introducedin [116]. In BPolSK, the signal is modulated by changing the polarization of the light. Since po-larization states of light are less sensitive than the amplitude, phase or intensity of optical signals,BPolSK has higher tolerance to underwater turbulence and other channel interference such as am-bient light. This property is ideal for UWOC in low SNR environment. PolSK can also be used tosuppress backscatter of the transmitter in a duplex system and has better immunity to phase noiseof lasers [98, 116]. Although PolSK is ideal for optical wireless communications, it still suffers fromshort transmission distance and low data rate. To overcome these limitations, Dong et al. presenteda novel polarized pulse position modulation (P-PPM) [117, 118]. This modulation scheme combinesconventional PPM and PolSK together by transmitting a series of PPM symbol in different polar-ization directions [118]. Numerical results show that P-PPM benefits from the virtues of both PPMand PolSK. It can increase the transmission bandwidth and distance of a UWOC system.Another coherent modulation scheme implemented in UWOC is the subcarrier intensity modu-lation (SIM). The interest of SIM is the much higher spectral efficiency [9]. But SIM also requirescomplex modulation/demodulation devices and suffers from poor average power efficiency [119].By using SIM, orthogonal frequency-division multiplexing (OFDM) can also be achieved in UWOCsystems [119, 120].3.2 Channel Coding of UWOCAs was discussed in the previous chapters, due to the severe absorption and scattering effectsinduced by sea water, the transmitted optical signal will experience considerable attenuation. Thisundesirable effect will directly degrade the BER performance of UWOC system. In order to mitigatethe impact of aquatic optical attenuation and maintain a low BER in low SNR underwater envi-ronment, forward error correction (FEC) channel coding techniques can be implemented in UWOCsystems.FEC coding is an error-control technique that adds redundant bits into the transmitted sequenceso that the receiver can correct a limited number of errors in the received message. A properlydesigned FEC code is capable of improving the power efficiency of a communication system, but atthe expenses of decreased bandwidth efficiency. For a UWOC system, these benefits are presented as353.2. Channel Coding of UWOClower transmitter power requirements or extended link range. Generally, FEC codes can be dividedinto two categories: block codes and convolutional codes [121]. Researchers have employed severalclassical block codes into the UWOC systems due to their simplicity and robustness. The first blockcode that implemented in UWOC system is the Reed-Solomon (RS) code [122]. In [122], Cox et al.demonstrated an experimental UWOC system that utilized (255, 129) RS FEC code. This systememployed 405 nm laser diode and RZ OOK modulation to achieve a 500 kbps UWOC link in a 3.66meters long water tank. The experimental results suggested that the coded system can reduce therequired power to achieve a BER of 10−4 by approximately 8 dB compared with an uncoded OOKsystem. Based on [122], the same research group from North Carolina State University upgradedtheir system in [123]. In the upgraded system, a 5 Mbps UWOC link was established also using RScode in three and seven meters long water tanks. The experimental results show that the (255,129)RS and (255,223) RS codes are capable of improving the SNR of received signals about six and fourdB respectively at a given BER of 10−6. Another similar experimental UWOC system that utilized(2720,2550) RS and SIM can be found in [120]. In this system, the (2720, 2550) RS code performedan error correction that reduced the input BER of 1.5 · 10−3 to 10−9.Besides the RS code, other classical block code and error-detecting code such as Bose-Chaudhuri-Hocquenghem (BCH) code and cyclic redundancy check (CRC) code have also been implementedin UWOC systems to improve the BER performance in low SNR underwater environment. In[124], the authors simulated the anti-noise performance of BCH and RS codes with simple OOKmodulation. Numerical results indicated that the RS code outperformed the BCH code in errorcorrection capability, but at the expense of lowering transmission data rate. In [106], a UWOCsystem based on hardware description language (HDL) was demonstrated. In this system, theauthors referred the architecture of IEEE 802.15.4 and IEEE 802.11 protocols and implementedCRC code in the medium access control (MAC) layer of the system. The BER performance at thereceiver was improved compared with uncoded systems. Instead of applying one layer of FEC codefor byte-level error correction, packet-level error correction coding schemes were also developed tomaintain the robustness of UWOC system. In [79] and [125], Doniec et al. embedded a two-layererror correction coding scheme into a UWOC video transmission system. In this two-layer channelcoding scheme, the transmitted video frames were firstly encoded on Manchester codes and LubyTransform (LT) codes to mitigate packet-level losses, then CRC and RS codes will be employedsequentially for byte-level error correction at the physical layer. Experimental results show thatthis multi-layer coding scheme can greatly improve the robustness of the UWOC system in a turbid363.3. Summarywater environment. But trade-offs may be taken between system performance and complexity.Although several block codes are simple to be implemented, they are not capable of providingthe optimal performance for UWOC, especially in the environment with strong interference. Thus,more complex and powerful channel coding schemes such as low-density parity-check (LDPC) codeand Turbo code are employed. LDPC code is a highly efficient linear block code. It is constructedby employing sparse parity check matrices and can provide an error-correction performance closeto the Shannon limit [98]. Turbo code is a parallel concatenated code. It combines two or moreconvolutional codes and an interleaver to produce a block code that can also achieve a BER close tothe Shannon limit. Although lots of research works on implementing LDPC and Turbo codes in FSOcommunication system have been proposed, there is still little investigation on applying LDPC andTurbo codes in UWOC. Everett demonstrated the performance of RS, LDPC and Turbo codes inUWOC systems both theoretically and experimentally [126]. The author explained the mechanismof RS, LDPC and Turbo codes in details, and compared their performance for UWOC such as BER,power efficiency, and link distance extension. This work provides a relatively complete descriptionof implementing channel coding techniques in UWOC.3.3 SummaryIn this chapter, we first introduced several IM schemes that implemented in UWOC systemssuch as OOK, PPM, PWM, DPIM and so on. Then, applications of coherent modulation techniquesin UWOC systems were presented. Besides the modulation schemes applied in UWOC, we alsointroduced the applications of channel coding methods. These applications include the simpleclassical block codes such as RS code and BCH code, as well as more powerful LDPC and Turbocodes. At the end of this chapter, we will summarize all the related literatures of UWOC channelmodulations and coding techniques in Table 3.1 and Table 3.2 respectively.373.3. SummaryTable 3.1: Summary of literatures on UWOC modulation schemesUWOC modulations Literature CommmentsOOK [62], [99], [100], [113] Simple but with low efficiency.PPM [101–109] High energy efficiency.DPIM [9], [110–112] Higher bandwidth efficiency.PSK [114, 115, 119]Combined with intensitymodulation.QAM [114]Combined with intensitymodulation.PolSK [116–118]Higher tolerence tounderwater turbulence.SIM [119, 120]Increase system capacity;low cost.Table 3.2: Summary of literatures on UWOC channel codingUWOC channel codes Literature CommmentsRS [79], [124], [125], [127] Simple and robust block codes.BCH [124] Simple and robust block codes.CRC [79], [106], [125] Simple error-detecting codes.LT [79], [125] Practical fountain code.LDPC [126] Complex linear block code.Turbo [126] Complex convolutional code.38Chapter 4Experimental Setups and Prototypesof UWOCIn this chapter, we will study the experimental setups and prototypes of UWOC from differentaspects. Firstly, we are going to introduce several typical LOS/NLOS experimental setups andprototypes of UWOC. Secondly, we will review the research of UWOC implementations in severalspecific topics, which include retroreflector, smart transceiver design, UWOC for underwater vehiclesand the hybrid UWOC systems. Finally, we will summarize this chapter and propose the literatureclassification of experimental UWOC systems.4.1 Typical LOS/NLOS UWOC systemsAs we have mentioned in Chapter 1, although a few commercial UWOC products were proposedin the early 2000s, the large scale commercial applications of UWOC systems have not been realizedso far. Most of the UWOC systems are experimental demonstrations and prototypes in laboratoryenvironment. In the remaining of this section, we will provide a comprehensive summary of therecent progress on experimental UWOC research. The purpose of this summary is not to introduceall the UWOC experimental literatures in details, but to provide a general description of the mostrecent works on UWOC experiments that concern different applications and approaches.According to the link configurations, experimental setups and prototypes of UWOC can bedivided into two categories: LOS experimental setups and NLOS experimental setups. Due tothe simplicity of implementation, most UWOC experimental systems utilize LOS configuration. InFigure 4.1, a typical laboratory LOS UWOC system based on intensity-modulation direct-detectionis demonstrated. The configuration of LOS UWOC link is similar to the FSO communication setups[98]. On the transmitter side, the information bits are generated by a personal computer (PC), andthen modulated onto optical carriers. In several UWOC experiments, the modulated optical signalwill be further amplified by an optical amplifier and then transmitted through lens that are precisely394.1. Typical LOS/NLOS UWOC systemsInformation SourceAutomatic gain control amplifierDriverModulatorLens/Optical instrumentLens/Optical instrumentWater tank/pipeTransimpedance amplifierPhoto detectorOptical amplifierTransmitter SideRecevier SideLPFDemodulatorDecoderLight sourcePC/Data analyzerFigure 4.1: A typical laboratory LOS UWOC system based on intensity-modulation direct-detection(IM/DD) technique.aligned to focus the light. Water tank or pipe is used to model the underwater transmission link. Inorder to mimic the different refractive condition and turbidity of underwater environment, Maaloxis added in the water to act as a scattering agent for attenuating the light beam [94, 127]. On thereceiver side, the optical signal will go through an optical filter and focusing lens. It will then becaptured by the photodiode. Since photodiode can only transform the variations of light intensityinto corresponding current changes, a trans-impedance amplifier is cascaded as the following stageto convert current into voltage. The transformed voltage signals will then go through a low passfilter to reduce the thermal and ambient noise levels [98]. Further signal processing programs thatinclude demodulation and decoding will be performed at the last two stages of the receiver. Therecovered original data will finally be collected and analyzed by a PC or BER tester for evaluatingseveral important performance parameters such as BER.There are two light sources that are commonly used in typical UWOC experimental systems:light-emitting diodes (LEDs) and laser diodes (LDs). As it was stated in Chapter 1, blue orgreen wavelength has been chosen for the light sources to minimize the aquatic optical attenuation.Compared with LED, LD has higher output power intensity, better collimated properties, narrowerspectral spreading, and much faster switching speeds. But at the expense of higher cost, shorter404.1. Typical LOS/NLOS UWOC systemslifetime and dependence on temperature. Thus LDs are more appropriate to be implemented inapplications of high-speed UWOC that has strict alignment requirement. In several LD-basedUWOC applications, optical diffusers are implemented to reduce the system pointing requirements.Compared with the LD-based UWOC system without diffusers, this diffused LD-based UWOCsystem can benefit from both high speed and relatively low pointing requirements. On the otherhand, since LEDs offer lower output power intensity, wider divergence angles, and lower bandwidths.They can be installed in several diffused UWOC applications with short-range, low-speed linkrequirement.At the receiver, there are also two types of photodiodes that are widely used in UWOC ex-periments: P-i-N (PIN) diode and avalanche photodiode (APD). The major difference betweenthese two devices is in the noise performance. For PIN photodiodes, the dominant noise is thermalnoise, while for the APDs, the performance is mainly limited by shot noise [98]. Since APD canprovide higher current gain, it can be implemented in longer UWOC links (tens of meters), butat the expense of more complex auxiliary circuits. Besides PIN diodes and APDs, photomultipliertubes (PMT) have also been implemented in several UWOC experiments [70, 105, 106, 128–131].Compared with photodiodes, PMT benefits from higher sensitivity, higher optical gain and lowernoise levels. But it also suffers from high voltage supplies (on the order of hundred volts) and highunit cost. Moreover, PMT is susceptible to shocks and vibration. It can be easily damaged bythe overexposure to light. The cost of PMTs are also much higher than photodiodes. Thus PMTsare commonly used in static experimental UWOC systems. Based on the typical LOS UWOC linkconfiguration and the critical devices that we have introduced, lots of experimental UWOC links fo-cusing on channel model verification and system performance analysis have been proposed in recentyears.Since LED benefits from its low cost and stable performance in various environments, severalresearchers preferred to employ it as the light sources in experimental UWOC systems [109, 120, 124,132]. In [7], Chancey proposed a UWOC system based on high power Gallium Nitride LEDs. Thisexperimental demo is capable of achieving 10 Mbps video transmission over a distance of 12 meters.Also in [8], Simpson from the North Carolina State University demonstrated a UWOC system withsignal processing capabilities that utilized high-power LED as the light source. Experimental resultsshow that 1 Mbps data rate is achievable over a distance of 3.66 meter long. Similarly, the authorof [133] has also developed a UWOC system that utilized a high-power blue LED as transmitterand a blue enhanced photodiode as receiver. This system successfully accomplished a 3 Mbps data414.1. Typical LOS/NLOS UWOC systemstransmission in a 13-meter long water tank. By using the mirrors folding architecture, the authorsof [95] tested their LED based UWOC system over a distance of 91 meters, a maximum data rateof 5 Mbps was accomplished. Recently, researchers from the Massachusetts Institute of Technology(MIT) presented a bidirectional UWOC system named AquaOptical [110]. The transmitter of thesystem consisted of six five watts LEDs with 480 nm wavelength. The researchers tested thisdemo system in both pool and ocean environment. Experimental results showed in clear poolwater, the AquaOptical can achieve a data rate of 1.2 Mbps at distances up to 30 meters; whilein turbid water with only three meters visibility, the system achieved a data rate of 0.6 Mbps overnine meters. As an upgraded version of AquaOptical, AquaOptical II can establish a bidirectionalunderwater communication link between each transceivers [111]. Since AquaOptical II is designedusing a software defined radio, it has more powerful signal processing capabilities than its previousgeneration and can also achieve a data rate of two Mbps over a distance of 50 meters. Severaltheoretical channel models have been validated through this testbed [79]. Furthermore, the MITresearchers also performed a real-time video delivery experiment by employing AquaOptical II [125].Using the same design approach of software defined radio, Cox et al. from North Carolina StateUniversity have also built up a UWOC experiment based on LED [119]. Since software definedradio system is more configurable than conventional hardware implementations, it is convenient totest various modulation formats or digital filtering schemes on UWOC [119]. The authors examinedthe performance of BPSK and Gaussian minimum shift keying (GMSK) schemes and accomplisheda data rate of one Mbps over a range of 3.66 meters. Most recently, a typical cellular UWOCnetwork prototype based on LEDs was demonstrated in [99]. The authors implemented code divisionmultiplexing access (CDMA) techniques in this prototype and tested the network performance invarious water conditions. Besides the experiments that we have already mentioned, other similarrecent experimental UWOC systems and prototypes that utilized LEDs as light sources can also befound in [100, 108, 134].Instead of using LEDs, several experimental setups also utilized lasers as the light sources dueto its high bandwidth and lower noise floor [63, 85]. Although laser and laser diodes were inventedin the early 1960s, only few early laser-based UWOC experiments have been performed in the 1990s[103, 135]. In recent years, with the cost reduction and popularization of laser devices, there isa surge of laser-based UWOC experimental systems. Cox et al. constructed a laboratory testbedbased on a 405 nm blue laser diode and PMT. This setup can provide up to one Mbps underwaterdata transmission in a distance of 12 meters [136]. Hiskett et al. proposed a laser-based UWOC424.2. Retroreflectors in UWOCsystem by utilizing 450nm laser diode and APD [137]. One 40 Mbps wireless communication link wasestablished over one meter water tank. In [128], the authors demonstrated a real-time underwatervideo transmission system by implementing 488 nm blue laser and PMT. Five Mbps high-speedvideo stream was successfully transmitted through 4.5 meters long underwater channel. Severalresearchers employed laser to study the spatial and temporal dispersion effects of UWOC links overdifferent modulations, coding schemes and water conditions [138–141].Compared with typical LOS UWOC experimental systems, only few UWOC experiments havefocused on the diffused and NLOS link configurations. Pontbriand et al. from the Woods HoleOceanographic Institution (WHOI) demonstrated a broadcasting diffused UWOC system [142].This system can achieve a data rate up to 10 Mbps over a maximum vertical distance of 200meters and horizontal radius up to 40 meters [142]. In [143], Cochenour et al. employed 532 nmlaser and a diffuser to generate diffused light. A diffused UWOC link with up to 1 Gbps datarate in 7.72 meters long water tank was established. On the other hand, the experimental NLOSUWOC links are mainly focused on the applications of underwater ranging and imaging. In [144],Alley et al. proposed a NLOS imaging system that utilized 488 nm blue laser as the illuminator.Experimental results demonstrate that, compared with the conventional LOS imaging system, thisNLOS configuration significantly improves the SNR of imaging. A similar experimental approachthat employed modulated pulse laser in NLOS configuration for underwater detection, ranging,imaging, and communications was presented in [145].4.2 Retroreflectors in UWOCRetroreflector is an optical device that can reflect arbitrary incident light back to its source(Figure 4.2). Utilizing this beneficial characterization, a modulating retroreflector UWOC systemwas introduced. In the modulating retroreflector link (Figure 4.3), the active transceiver projects alight beam into the retroreflector. During the reflection process, the modulator will modulate thelight beam and add information on it. This information will later be captured and demodulated bythe active transceiver. The most significant advantage of modulating retroreflector UWOC systemis that most of the power consumption, device weight, volume and pointing requirements are shiftedto the active end of the link, thus the passive end will benefit from small dimensions, relatively lowpower and pointing requirements [146]. There are lots of sensor nodes and underwater vehicles inUWSNs. Each sensor node and underwater vehicle is required to have long enough cruising time due434.2. Retroreflectors in UWOCFigure 4.2: Demonstration of corner and spherical retroreflectors.Transmission DevicesInformation SourceReceiving Devices Photodiode/PMTRetroreflectorInterogated BeamModulated BeamLight SourceModulatorActive transceiver Passive retroreflectorFigure 4.3: Modulating retroreflector link.to the difficulty of recharging battery. In this sense, modulating retroreflector becomes an attractivechoice. In addition to the challenges that involved in a direct UWOC link such as absorptionand scattering, retroreflector-based UWOC systems have several additional limitations. Unlike thetypical UWOC links, the retroreflector based UWOC links have to transmit through the underwaterchannel twice, so the link will experience higher attenuation and interference. Furthermore, thebackscattered light generated from the interrogating beam can be significant in turbid water whereit will eventually surpass the desired retro-reflected signals.Although the concept of implementing retroreflector into FSO communication system has al-ready been proposed for almost 20 years [147], only few research works on UWOC retroreflec-tor system were demonstrated recently. The first institute that implemented retroreflector FSO444.3. Smart Transceivers of UWOCcommunication in marine environment is the U.S. Naval Research Laboratory (NRL). Since late1990s, the NRL began to launch the research of retroreflector applications in FSO communicationand successfully achieved shore-to-shore, boat-to-shore and sky-to-ground retroreflector FSO links[146, 148, 149]. Based on these achievements, NRL researchers then applied the retroreflector linkinto underwater environment. In [150] and [151], Mullen et al. employed a polarization discrimina-tion technique to overcome the impact of backscattering on the interrogating light. An experimentaltest was also performed in laboratory water tank to evaluate the system performance. The authorscompared the experimental results of polarized and non-polarized setups with different transceiverFOV and link ranges. Experimental results showed that, by utilizing polarization discriminationtechnique, the backscatter level can be greatly reduced. This fact will then increase the commu-nication range of retro link. In [127], Cox et al. from North Carolina State University proposed ablue/green retro-reflecting modulator for UWOC based on micro-electromechanical system (MEM-S). The authors deployed the retroreflector link in a 7.7 meters long water tank and evaluated thesystem performance with various water turbidities. Experimental results show that 1 Mbps and 500kbps data rates can be achieved in 2.7 attenuation length 6 and 5 attenuation length respectively.4.3 Smart Transceivers of UWOCAs shown Figure 4.1, in a UWOC system, the information waveforms are generated by a sourceand then transferred by an optical transmitter through the water channel to a specific destination.At the other end of the link, the receiver will collect the optical signal and recover the originalinformation. Although the transmission wavelength is carefully selected in blue/green transparentwindow to minimize the attenuation effect of sea water, several other factors such as misalignmentwill still severely degrade the link performance. As we have stated in Chapters 1 and 2 , most UWOCsystems utilize point-to-point configuration, and thus precise pointing and tracking requirementsare necessary. However, link misalignment is an inevitable phenomena in underwater environment,any variations of refractive index or turbulence of ocean can cause link misalignment and interruptcommunication. Especially in mobile UWOC applications such as AUVs and ROVs, the two endsof a link are all in nonstatic condition, which makes the alignment more difficult to be achieved.Conventionally, there are three common methods to relief the pointing requirements of a UWOCsystem: using diffused light beam, increasing receiver aperture size, or implementing a dedicated6The attenuation length defined as the product of attenuation coefficient and link distance.454.3. Smart Transceivers of UWOCgimbal system. Diffused light beam can effectively increase the illuminated area of a light source, butthe communication range also shrinks. Although large aperture can increase the receiver FOV and ithas already been implemented in several UWOC systems such as [142], the extra introduced ambientlight and limited transceiver size requirement will still restrict the application of this method.Dedicated gimbal system can be used in several applications that have less size limitation andenergy requirements, but for compact UWOC systems that don’t have much volume and energybudget, this approach is not practical.Considering the limitations of each compensation method that we have introduced, a compactadaptive smart UWOC transceiver that can relax the misalignment requirement with minimizedvolume and energy cost. In [152], Simpson et al. proposed a novel UWOC front-end that intro-duced the concept of smart transmitter and receiver. The smart quasi-omnidirectional transmittercan estimate the water condition according to the backscattered light captured by the adjacentsmart receiver. Based on specific water conditions, the transmitter can take several actions suchas changing transmission light wavelength to improve link performance. The transmitter can alsoelectronically switch the beam direction according to the angle of arrival of detected signal. Onthe receiver side, segmented lens array architecture was implemented to increase the total FOV. Byusing the information of angle of arrival estimation, the smart receiver can also adjust and steerthe FOV towards the direction of desired signals to improve the the SNR of the received signal.Moreover, the CDMA technique has also been implemented in both transmitter and receiver ends toreinforce the system performance in multi-user environment. The authors installed the prototypingsmart transceivers in a 3.66-meter long laboratory water tank to evaluate the system performance.Experimental results demonstrate that the smart system can effectively increase the total FOV ofthe receiver. The preliminary algorithm for angle of arrival estimation and backscatter estimationwas also verified to work properly. Other performance aspects such as diversity combining andmulti-user CDMA approach were also tested and proved to be effective. This novel trial of smarttransceivers provides an adaptive solution to handle the impact of dynamic nature of underwaterenvironment to the UWOC systems. It can be applied to different underwater platforms such asAUVs, ROVs, and other sensor nodes embedded with UWOC system. Several theoretical researchworks focusing on smart or adaptive UWOC transceivers were also proposed. In [107], Tang et al.presented an adaptive gain control scheme for UWOC receivers based on APD. The authors deriveda close-form expression that can describe the relationships of optimal gain of APD, link range andreceiver offset distance. This result can be further applied to practical design of UWOC transceiver464.4. UWOC for Underwater Vehiclesfor improving the link reliability.4.4 UWOC for Underwater VehiclesWith the increasing demands of human underwater activities, underwater vehicles such as AUVsand ROVs have been widely applied to perform different tasks such as undersea resource exploration,wreck rescue missions and maintenance of oil production facilities. In the perspective of commu-nication methods, underwater vehicles can be divided into three categories: tethered underwatervehicles, wireless underwater vehicles, and hybrid underwater vehicles. Tethered underwater vehi-cles are usually controlled robots that connected to the surface control platform through an opticalfibre or electrical cable. The tethered system has long endurance time and can provide reliablehigh-speed data communication, but at the expense of higher manufacture cost and limited opera-tion range. On the other hand, conventional wireless underwater vehicles are usually autonomousoperated robots that utilize acoustic wave as the viable communication carrier [153]. Since this kindof vehicle is free from the limitations of connection cable, it has more flexibility and can operate ina vast area. The bottle neck of this approach are the low bandwidth, high latency, and complexenergy-consuming acoustic transceivers. Hybrid underwater vehicles integrated both tethered andwireless systems together [154, 155]. It has the optimal flexibility and reliability, but it’s not suitablefor the large-scale implementations in UWSNs due to the high unit cost, bulky instrumentations,and large number of cables.In order to satisfy the needs of UWSNs for compact, endurable, and high-bandwidth under-water vehicles, several researchers have embedded UWOC into AUVs and ROVs to overcome thelimitations of conventional underwater vehicles. A team from the Computer Science and Artifi-cial Intelligence Laboratory (CSAIL) of MIT firstly proposed a prototyping AUV system calledautonomous modular optical underwater robot (AMOUR) [156]. The AMOUR was designed toperform tasks that including underwater monitoring, exploration, and surveillance. Since AMOURis based on a stack-up modular design approach, it also has the capabilities to deploy and recoversensor nodes in the sensor networks. The earliest version of AMOUR employs LEDs as the lightsource and can achieve one Kbps data rate over a distance of two meters. After the presentation ofthe first AMOUR prototype, CSAIL researchers upgraded the AMOUR system. Several new fea-tures based on UWOC such as remote control, localization, and time division multiplexing access(TDMA) have been implemented into this underwater vehicles systems [157, 158]. In [159], the474.5. Hybrid Acoustic/Optical UWC Systemsresearchers demonstrated a cooperative UWSN that employed AMOUR, another different kind ofAUV named Starbug and several underwater sensor nodes [160]. During the experiment, cooper-ative tasks such as data transmission, cooperative localization and navigation, as well as physicalconnection were performed. This work has proved the feasibility of cooperations among differentunderwater vehicles and sensor nodes. It also verified the plausible approach to achieve long-termoperation of UWSNs. In [112], an upgraded underwater vehicle system AMOUR VI that embed-ded with UWOC module to achieve real-time control link was demonstrated. In this experiment,the CSAIL researchers used blue/green LEDs as light source and tested the system in a shallowswimming pool where ambient light existed. Human input device was used to control the orienta-tion of the vehicle. Compared with the conventional acoustic ROVs which has a data rate up tohundred Kbps and latencies of hundred milliseconds, this UWOC system can achieve data rate onthe order of Mbps and latencies on the order of one millisecond in a range of tens of meters. Thearchitecture of the vehicle is compact. Both the transmitter and receiver modules are sealed in atransparent plastic cylinder with approximate length of 30cm and weight of two kilograms. Thevehicles can also move in arbitrary directions with the embedded thruster system. The details ofthis thruster algorithm and its corresponding configurations were presented in [161]. Besides theseresearch results that were proposed by MIT CSAIL, other UWOC research groups also presentedseveral demos and prototypes on optical wireless underwater vehicles such as [6, 162, 163]. Hybridcommunication systems which included both acoustic and optical modules were also implementedin several underwater vehicles [164, 165]. We will introduce them in the following section.4.5 Hybrid Acoustic/Optical UWC SystemsThe performance of UWOC systems can be severely degraded by the absorption and scatteringeffects of sea water, channel turbulence, misalignment errors and other impact factors. All of theseundesirable factors can cause frequent communication failure. Thus the reliability of UWOC systemshould be enhanced. Based on hybrid RF/FSO communication systems [98], one plausible method toincrease the reliability of UWOC system is to employ acoustic wave as back-up scenario. Comparedwith UWOC, underwater acoustic communication method benefits from its mature technology,long link range and lower pointing requirements, but suffers from low data rate, low security andbulky instruments. On the other hand, UWOC systems can achieve high speed point-to-point datatransmission, but can’t operate in long distance and turbid environment. Considering the pros484.5. Hybrid Acoustic/Optical UWC Systemsand cons of these two methods, two typical hybrid link configurations have been proposed (Figure4.4). The first configuration (Figure 4.4(a)) utilizes both acoustic wave and optical wave as duplextransmission medium. In this configuration, the two ends of the link are all mobile underwatervehicles that equipped with both acoustic and optical transceivers. When the two nodes of the linkare in short distance and water condition is clear, the system will use optical wave as carrier toachieve high speed data transmission. If there is large distance between the two nodes or the wateris turbid, the system will instead employ acoustic methods in order to accomplish connectivity.The virtue of this implementation is the high flexibility and reliability, but at the expense of highpower consumption and bulky instruments due to acoustic transceivers on both ends. In the secondconfiguration shown in Figure 4.4(b), the system is configured by one static control platform andseveral mobile sensor nodes. Acoustic wave is used as a broadcasting method to transmit controlinformation in the downlink from control platform to each sensor node. While optical wave isapplied in the communication links between each senor node, as well as uplinks from sensor nodesto main control platform. This hybrid UWC system utilizes the advantages of each communicationmethod. Since acoustic wave has diffusion property and long propagation distance, it can coverthe area that is distributed with senor nodes. Moreover, in the downlink from control platformto sensor nodes the transmitted information are low-speed control signals, which are suitable foracoustic communication. On the other hand, in the uplink of the system the large volume of oceanicmonitoring data is transmitted through high speed UWOC links.Based on these two hybrid UWOC link configurations, several related research have been pro-posed in recent years. In [164], researchers from the CSAIL of MIT reported a novel hybrid UWSNsthat is capable of accomplishing long-term underwater monitoring tasks including video stream ofsea floor, real time data of water temperature, pressure and other parameters. This hybrid UWSNsconsist of two types of sensor nodes: static nodes and mobile nodes. Point-to-point communicationbetween each node is accomplished using high-speed optical wave, while diffused broadcasting linksare achieved through acoustic method. Moreover, the mobile nodes can also locate and move to thestatic nodes for data muling. The authors also demonstrated several optimizing algorithms and thecorresponding hardware implementations of the system. Experimental results show that this hy-brid data collection system can provide a much higher power and data transmission efficiency thanthe conventional acoustic underwater networks with multihop protocol. In [164], Vasilescu et al.proposed a more advanced hybrid underwater senor network [166]. This underwater sensor networkconsists of several sensor nodes called AquaNodes. Each of the nodes contain an RF bluetooth494.5. Hybrid Acoustic/Optical UWC SystemsOptical LinksAcoustic Links(a) Mutual hybrid UWOC configuration.Acoustic LinkOptical Links(b) Boradcasting hybrid UWOC configuration.Figure 4.4: Two types of hybrid acoustic/optical UWC links.504.5. Hybrid Acoustic/Optical UWC Systemsmodule, an acoustic module and an optical module that operates with green light. Depending onthe application environment, the users can communicate with each node through different mean-s. For instance, in clear shallow water using optical signal, in water/air interface using bluetoothmethod. Furthermore, TDMA and self-synchronization technologies are also implemented in eachnode. According to the experimental results, this system can achieve a 400-meter long acoustic linkwith 300 bps data rate in ocean environment and establish an optical link up to eight meters with330 kbps data rate. Based on this transmission speed, the data collected through the embeddedsensors which include temperature, pressure, and water chemistry information can be continuouslytransmitted to the communication buoy, which will then relay the information to the onshore datacenter for following processing and analysis. Except these two hybrid UWOC systems demonstratedby MIT, scholars from other research institutions also proposed several discussions on hybrid UWOCtopic. Farr et al. from the WHOI presented the operation concept of an untethered ROV (UTROV)that employs both optical and acoustic communication methods [167]. This vehicle can accomplishtypical survey and reconnaissance tasks over a long distance using a low bandwidth acoustic mo-dem. It can also communicate optically by employing a small ship-based or seafloor-based relay.Based on UTROV, the authors also demonstrated a seafloor borehole observatory system calledcirculation obviation retrofit kit optical telemetry system (CORK-OTS) in [168]. The CORK-OTSconsist of several underwater ROVs and a surface vessel with lowered cable. At the lowered cableend, an acoustic modem was integrated with the optical system to send control information, whileoptical transceivers were implemented in both ROV and lowered cable end to achieve high speeddata transmission. Although the details of testing this platform was not discussed, this researchstill provided a novel method to achieve seafloor observation.Besides the experimental setups of hybrid UWOC, little theoretical research on this topic hasoccurred. In [169], the authors proposed a hybrid duplex optical-acoustic communication system.In this system, the downlink from the base station to the AUVs is a diffused acoustic link with lowbandwidth and the uplinks are highly directional optical links with high bandwidth. The authorsalso discussed the factors that limited the performance of system which include refractive indexvariation, optical pointing error and acoustic latency. Another theoretical evaluation of hybrid un-derwater optical-acoustic network can be found in [170]. The researchers proposed a communicationnode equipped with both acoustic and optical transceivers. A transmission algorithm was designedand applied to each nodes to ensure the link alignment and connectivity. Simulation results showthat the hybrid system has better energy efficiency than the pure acoustic system and is more514.6. Summaryflexible to be applied in different water conditions.4.6 SummaryIn this chapter, we studied the the experimental setups and prototypes of UWOC from dif-ferent aspects. We demonstrated the typical experimental setups of LOS UWOC systems andintroduced the characterization of light sources and receivers that are commonly used in UWOC.We also discussed the development and state-of-the-art of several popular UWOC applications suchas retroreflector, smart transceivers, underwater vehicles and hybrid UWOC systems. At the endof this chapter, we summarize all the discussed experimental UWOC systems in Table 4.1.524.6. SummaryTable 4.1: Summary of literatures on experimental setups and prototypes of UWOCSpecified topics Literatures CommmentsTypical LED-based LOS UWOC[95], [79], [99, 100], [105, 106],[108–111], [119, 120], [124, 125][7, 8, 131–134]Relatively low cost; easy to beimplemented; moderate speedand communication range.Typical Laser-based LOS UWOC [63, 85, 103, 128, 135–141, 171]Higher cost; high speed;long communication range;Strict pointing.Diffused LOS UWOC [142], [143] To achieve broadcasting UWOC.NLOS UWOC [144], [145]To overcome underwaterobstacles; few experimentswere performed.Retroreflector-based UWOC [127, 150, 151]Light and compact architecturewith low cost and energy budget.UWOC smart transceivers [152], [107]Adaptive transmission toimprove link performance.UWOC in underwater vehicles [6, 112, 156–163]With higher speed and lessinstruments budgets thanacoustic method.Hybrid UWOC systems [164–170] Improve system reliability.53Chapter 5Outage Performance for UnderwaterWireless Optical Links With PointingErrorsIn this chapter, we study the outage performance of vertical buoy-based UWOC links usingIM/DD OOK with zero and nonzero boresight pointing errors. Firstly, we introduce the pointingerror models with zero and nonzero boresight. Secondly, we introduce the beam spread functionwhich describes the propagation of light in underwater environment. Thirdly, we analyze the outageprobability with zero and nonzero boresight pointing errors. A closed-form outage probability withzero boresight pointing errors is achieved. We also derive the closed-form outage probability boundswith nonzero boresight pointing errors. These bounds can be made arbitrarily tight and approachthe exact outage probability.5.1 Pointing Errors ModelsSimilar to the definition of pointing errors in FSO communication, pointing errors in UWOCalso include two components: boresight and jitter. The boresight is the fixed displacement betweenbeam center and center of the detector, and the jitter is the random offset of the beam center atdetector plane, which is mainly caused by random sea surface slopes [172]. In this section, we willfirst present the the pointing errors model with zero boresight, and then introduce the pointingerror model with nonzero boresight.5.1.1 Pointing Errors Model with Zero BoresightThe statistics of sea surface slopes has been studied during the past decades. Cox and Munkfirst measured the wave slope of random sea surface generated by wind. They concluded that the545.1. Pointing Errors Modelssurface slope approximately follows the Gaussian distribution with Gram-Charlier series 7 correctionby taking into account the skewness and kurtosis [174]Pr(sx, sy) =12piσuσcexp[−(s2x2σ2u+s2y2σ2c)]1 + ∞∑i,j=1cijHi(sx)Hj(sy) (5.1)where sx = ∂z/∂x and sy = ∂z/∂y are defined as wave slopes of up/downwind and crosswinddirections in the Cartesian coordinate (x, y, z) respectively; Hi(sx)Hj(sy) is the product of Hermitepolynomials with coefficients cij to be determined; σ2u and σ2c denote the variance of slopes in theup/downwind and crosswind directions, respectively. Based on the experimental results of Cox andMunk, σ2u and σ2c for the clean surface can be estimated asσ2u = 0.003 + 0.00192U ± 0.002, σ2c = 0.00316U ± 0.004 (5.2)where U is the wind speed which is between 1 m/s and 14 m/s [174]. According to [174, eq. (18)],we expand the first 5 terms of the series in (5.1) as:P ′r(sx, sy) =12piσuσcexp[−(s2x2σ2u+s2y2σ2c)]×[1− 12c21(s2xσ2u− 1)s2xσ2c− 16c03(s3yσ3c− 3sxσc)+124c40(s4xσ4u− 6 s2xσ2c+ 3)+14c22(s2xσ2u− 1)(s2yσ2c− 1)+124c04(s4yσ4c− 6 s2yσ2c+ 3)+ · · ·].(5.3)We substitute the values of c21, c03, c40, c22, c04 of (5.3) from [174] and let U = 5m/s. The curves of(5.3) and two-dimensional Gaussian without multiplying the correction series are demonstrated inFigure 5.14(a) and Figure 5.14(b), respectively. Since σ2u and σ2c are small, the Gaussian functionhas a sharp peak, which means that the multiplied series have little impact on the PDF valuesfor larger sx and sy. Considering these facts, we can simplify (5.3) as two dimensional zero-meanGaussian distribution with different variancesPˆr(sx, sy) =12piσuσcexp[−(s2x2σ2u+s2y2σ2c)]. (5.4)7Since Gram-Charlier series is an approximation to the unknown probability distribution using well-known distri-bution(Gaussian) and correction polynomials, the integral of the probability density function needs not integrate toone [173].555.1. Pointing Errors Models(a) Plot of corrected PDF (with series).(b) Plot of non-corrected PDF (2D Gaussian without series).Figure 5.1: Comparison between corrected and non-corrected PDF of ocean slopes.565.1. Pointing Errors ModelsReceiver ApertureOCBABeam AxisSea SurfaceSea WaterAirxzyθφTransmitterFigure 5.2: Geometry of the buoy-based UWOC. Figure 5.2 is adapted from [10].According to [10], the geometry of buoy-based UWOC is demonstrated in Figure 5.2. A compactlight source is placed on a sea surface buoy illuminating the receiver at the bottom plane. x andy axes represent the up/downwind and crosswind directions respectively [10]. xOy plane is thebottom plane parallel to the horizontal sea surface and z is the vertical axis perpendicular to thexOy plane. The centers of light source and receiver aperture are located at points of C and Orespectively. A is the beam center at xOy plane and line AC represents the direction of beamaxis with zenith angle θ measured from z axis. Azimuth angle of slope between OA and x axis isdenoted as ϕ. OB is the normal of AC. The beam axis of the source remains the same direction asthe normal of sea facet distributed as (5.4). When the surface is calm without any slopes, the linkis assumed to be precisely aligned, which means the center of the light source is on the z axis [10].When the wind speed is low (1 ≤ U ≤ 7 m/s), the slope angles are small, the light sourcehas its center C approximately on z axis during the random slopes, i.e., BC ≈ OC = L [10].Then in the case of small slope angle and the receiver with wide open field of view (FOV), thereceiver can be treated as perpendicular to the beam axis with offset distance r from beam axis asr ≈ OB = L√x2a + y2a/√x2a + y2a + L2 where (xa, ya) is the position of A in the Cartesian coordinate[10]. From Figure 5.2, (xa, ya) is given by xa = L tan θ cosϕ and ya = L tan θ sinϕ. Then, the surface575.1. Pointing Errors Modelsslopes can be expressed as sx = ∂z/∂x = tan θ cosϕ and sy = ∂z/∂y = tan θ sinϕ, which denotethe relationship as (xa, ya) = (Lsx, Lsy). By substituting this relationship into (5.4), the PDF of(xa, ya) can be presented as [10]Pr(xa, ya) =12piσuσcL2exp[− 12L2(x2aσ2u+y2aσ2c)]. (5.5)From (5.5), we observe that xa and ya are independent zero mean Gaussian distributed RVs withdifferent variances, i.e., xa ∼ N (0, σ2uL2), ya ∼ N (0, σ2cL2). After several mathematical derivationsin Appendix I, the radial displacement r =√x2a + y2a which follows Hoyt or Nakagami-q distributioncan be expressed as [175]fr,HT (r) =rqHσ2uL2exp[−(1 + q2H)r24q2Hσ2uL2]I0((1− q2H)r24q2Hσ2uL2)(5.6)where qH = σc/σu ∈ (0, 1], I0(·) is the modified Bessel function of the first kind with order zero.The plot of (5.6) is shown in Figure 5.3.Figure 5.3: PDF of Hoyt distributed radial displacement r with L = 5m and different values ofwind speed U .585.1. Pointing Errors Models5.1.2 Pointing Errors Model with Nonzero BoresightAlthough typical buoy-based vertical UWOC systems are initially installed with near zero bore-sight errors, mooring displacement of the buoy, ocean current and wind will still generate consid-erable boresight errors. Thus, it’s necessary to take into account the impact of boresight errors inthe vertical buoy-based UWOC system. Based on the foregoing derivations of (5.5), the pointingerrors expression with nonzero boresight can be presented as:Pr(xa, ya) =12piσuσcL2exp[− 12L2((xa − µx)2σ2u+(ya − µy)2σ2c)](5.7)where µx and µy are the means of two independent Gaussian random variables xa and ya, respec-tively. Using polar coordinates, i.e., xa = r cos θ and ya = r sin θ, we obtain the radial displacementr =√x2a + y2a following the Beckmann distributionfr,BM (r) =r2piσuσcL2∫ 2pi0exp[−(r cos θ − µx)22σ2uL2− (r sin θ − µy)22σ2cL2]dθ. (5.8)The Beckmann distribution is also known as generalized Rician distribution, which is used to de-scribe the PDF of fading channels in general [172]. It specializes to the Hoyt distribution in (5.6)when µx = µy = 0, σu 6= σc. The plot of (5.8) is shown in Figure 5.4.Figure 5.4: PDF of Beckmann distributed radial displacement r with µx = 0.01, µy = 0.02, L = 5mand different values of wind speed U .595.2. Beam Spread Function5.2 Beam Spread FunctionLight propagation in underwater environment are severely impacted by absorption and scatter-ing. The effects of absorption and scattering can be modeled by the beam-spread function (BSF)as [11]B(L, r) =Pt2piσ20exp(−r22σ20)exp(−cL) + 12pi∫ ∞0Pt2piσ20exp(−r22σ20)exp(−cL)×{exp[∫ L0bp (v(L− z)) dz]− 1}J0(vr)vdv(5.9)where Pt is the transmission power; σ20 is the variance of Gaussian beam; L denotes the link distance,c presents the light attenuation coefficient in underwater environment; b is the light scatteringcoefficient; and v denotes the spatial frequency. In (5.9) J0(·) is Bessel function of the first kind oforder 0. p(v) = 12∫ pi0 p(β)J0(vβ)βdβ is the Hankel transform of the scattering phase function. Thecommonly used Henyey-Greenstein (HG) function is adopted as the scattering phase functionp(β) =1− g24pi(1 + g2 − 2g cosβ) 32(5.10)where g is the average cosine of β in all scattering directions. We adopt g = 0.924 [50]. Thephysical meaning of B(L, r) is the light irradiance with displacement r from the beam center axiswith perpendicular distance L (Figure 5.5).OCrBeam AxisSea SurfaceSea WaterAirxzyLight SourceLFigure 5.5: Geometry for BSF.605.2. Beam Spread FunctionThe values of B(L, r) with different values of r have also been calculated and verified throughwater tank experiments in [11]. The normalized values of (5.9) can be found in Figure 5.6 [11]. Wealso plot the non-normalized values of (5.9) with respect to r for different transmission power Pt inFigure 5.7.Figure 5.6: BSF results for L = 3.63m with different attenuation coefficients c. Model resultsshown as lines. Experimental data shown as points. Relative intensity is the received power withdisplacement r normalized by the power without displacement B(L, r)/B(L, 0) [11]. Figure 5.6 isreprinted from [11].Figure 5.7: BSF values for L = 5m and c = 0.3 with different values of transmission power Pt.615.3. Outage Probability with Zero Boresight Pointing ErrorsFor typical optical detector, the receiver aperture is on the order of centimeters which is narrowenough. Hence, we can derive the received power Pr directly from (5.9) without integrating B(L, r)in the aperture area as [10]Pr(r) =piD24B(L, r) (5.11)where D is the diameter of the aperture, r is a random variable follows Hoyt or Beckmann distri-bution in (5.6) and (5.8) respectively.5.3 Outage Probability with Zero Boresight Pointing ErrorsIn this section, we derive the outage probability with zero boresight pointing errors. We assumethat there’s a displacement r on the receiver plane between the beam center and the receiver aperturecenter. According to (5.11), the received power is Pr(r). When Pr(r) is less than a threshold γth8,outage occurs. Thus we define the outage probability asPout = Prob(Pr(r) < γth). (5.12)From Figure 5.6, we observe that Pr(r) is monotonically decreasing in r. Equivalently, we havePout = Prob(r > P−1r (γth)). (5.13)Since deriving the inverse function r = P−1r (γth) is intractable, we can obtain the value of r bylooking up the numerical table of B(L, r) such as Figure 5.7.In Figure 5.8, the receiver aperture is located at the center of the circle O. The radius of thecircle is P−1r (γth). The probability that the beam center doesn’t locate inside the circle is the outageprobability which can be derived by integrating the PDF of Hoyt distribution in (5.6)Pout,HT = Prob(r > P−1r (γth)) =∫ ∞P−1r (γth)fr,HT (r)dr = 1−∫ P−1r (γth)0fr,HT (r)dr (5.14)where∫ P−1r (γth)0 fr,HT (r)dr is the cumulative distribution function (CDF) of Hoyt distribution eval-uated at P−1r (γth). After mathematical derivations shown in Appendix II, we obtain the outage8γth can be regarded as a constant here. It is determined by the inverse function of instantaneous capacity C−1(R0)[172],where R0 is the transmission data rate.625.4. Bounds of Outage Probability with Nonzero Boresight Pointing ErrorsXYNon-outage regionOutage regionOContour line of 2-D Gaussian with zero mean and different varianceFigure 5.8: Demonstration of outage region.probability with zero boresight pointing errors asPout,HT =1− 1√1 + q2HQ1√1+qH1−qH√1− q4HP−1r (γth)2qH√1 + q2HσuL,√1−qH1+qH√1− q4HP−1r (γth)2qH√1 + q2HσuL−Q1√1−qH1+qH√1− q4HP−1r (γth)2qH√1 + q2HσuL,√1+qH1−qH√1− q4HP−1r (γth)2qH√1 + q2HσuL(5.15)where Q1(a, b) =∫∞b xexp(−x2+a22)I0(ax)dx is the first-order Marcum Q-function.5.4 Bounds of Outage Probability with Nonzero BoresightPointing ErrorsFollowing the similar approach of deriving the outage probability with zero boresight pointingerrors, we can also express the outage probability with nonzero boresight pointing errors asPout,BM =Prob(r > P−1r (γth)) =∫ ∞P−1r (γth)fr,BM (r)dr = 1−∫ P−1r (γth)0fr,BM (r)dr=1−∫ P−1r0r2piσuσcL2∫ 2pi0exp[−(r cos θ − µx)22σ2uL2− (r sin θ − µy)22σ2cL2]dθdr(5.16)635.4. Bounds of Outage Probability with Nonzero Boresight Pointing Errorswhere∫ P−1r (γth)0 fr,BM (r)dr is the CDF of Beckmann distribution evaluated at P−1r (γth). We noticethat the double integral in (5.16) is intractable, thus it’s necessary to derive the lower and upperbounds of the outage probability with nonzero boresight instead.5.4.1 Lower Bound of Outage Probability with Nonzero BoresightOrYXDABCGFEHIJKLMContour line of 2-D Gaussian with different non-zero mean and varianceQPNSTFigure 5.9: Integrating region for lower bound of outage probability.22 (n 1) ( 1),r n rr BN N          22 ( 1) ,n r nrr AN N         22 (n 1) ( 1),r n rD rN N         22 ( 1) ,n r nrC rN N        Figure 5.10: Coordinates of nth circumscribed rectangle on the upper semicircle.In Figure 5.9, we construct several rectangles with the same height circumscribed the upper semi-circle of non-outage region. Since the radius of the circle is r and HO = IH = JI = KJ = LK = r5 ,we utilize pythagorean theorem to derive the coordinates of each rectangle vertexes as: M (r, 0),A(r, r5), Q(√r2 − ( r5)2, r5), B(√r2 − ( r5)2, 2r5 ), P (√r2 − (2r5 )2, 2r5 ), C (√r2 − (2r5 )2, 3r5 ),N(√r2 − (3r5 )2, 3r5 ), D(√r2 − (3r5 )2, 4r5 ), S (√r2 − (4r5 )2, 4r5 ), T (√r2 − (4r5 )2, 5r5 ). Othercoordinates of rectangle vertexes in the quadrant II of Figure 5.9 can be obtained through thesymmetry property.645.4. Bounds of Outage Probability with Nonzero Boresight Pointing ErrorsConsidering the general case, we circumscribe N(N ≥ 2) rectangles with the same height ofrN =P−1r (γth)N on the upper semicircle in Figure 5.9. Following the derivation process of N = 5, thevertex coordinates of nth rectangle can be presented as Figure 5.10. Thus the integration of (5.7)over nth rectangle region can be expressed asPunit,BM(P−1r (γth))=∫ nP−1r (γth)N(n−1)P−1r (γth)N∫ √P−1r (γth)2−( (n−1)P−1r (γth)N )2−√P−1r (γth)2−((n−1)P−1r (γth)N)2 12piσuσcL2× exp[− 12L2((xa − µx)2σ2u+(ya − µy)2σ2c)]dxadya, N ≥ 2.(5.17)Since xa and ya are independent, we integrate xa and ya separately and obtainPunit,BM(P−1r (γth))=14[erf(µy − (n−1)P−1r (γth)N√2σcL)−erf(µy − nP−1r (γth)N√2σcL)]×erfµx +√P−1r (γth)2 −((n− 1)P−1r (γth)N)2√2σuL−erfµx −√P−1r (γth)2 −((n− 1)P−1r (γth)N)2√2σuL , N ≥ 2.(5.18)where erf(x) = 2√pi∫ x0 e−t2dt is the error function. Taking summation to (5.18), we derive theintegration of (5.7) over the whole rectangle region circumscribed the upper semicircle in Figure5.9 asPLB1,BM(P−1r (γth))=14N∑n=1[erf(µy − (n−1)P−1r (γth)N√2σcL)−erf(µy − nP−1r (γth)N√2σcL)]×erfµx +√P−1r (γth)2 −((n− 1)P−1r (γth)N)2√2σuL−erfµx −√P−1r (γth)2 −((n− 1)P−1r (γth)N)2√2σuL , N ≥ 2.(5.19)Following the similar process of deriving (5.19), we express the integration of (5.7) over the entirerectangle region circumscribed the lower semicircle as655.4. Bounds of Outage Probability with Nonzero Boresight Pointing ErrorsPLB2,BM(P−1r (γth))=14N∑n=1[erf(µy +nP−1r (γth)N√2σcL)−erf(µy +(n−1)P−1r (γth)N√2σcL)]×erfµx +√P−1r (γth)2 −((n− 1)P−1r (γth)N)2√2σuL−erfµx −√P−1r (γth)2 −((n− 1)P−1r (γth)N)2√2σuL , N ≥ 2.(5.20)According to (5.19) and (5.20), the lower bound of Beckmann CDF evaluated at P−1r (γth) ispresented asFr,LB(P−1r (γth))= PLB1,BM(P−1r (γth))+ PLB2,BM(P−1r (γth)). (5.21)Thus the lower bound of outage probability with nonzero boresight pointing errors can be expressedasPout,BM ≥1− Fr,LB(P−1r (γth))= 1− PLB1,BM(P−1r (γth))− PLB2,BM (P−1r (γth)) . (5.22)5.4.2 Upper Bound of Outage Probability with Nonzero BoresightContour line of 2-D Gaussian with different non-zero mean and variance. OrYXDABCGFEHIJKLMNPQFigure 5.11: Divide half of the non-outage region into several rectangles with the same height.In Figure 5.11,we construct several rectangles with the same height inscribed the upper semi-circle of non-outage region. Since the radius of the circle is r and OH = IH = JI = KJ =665.4. Bounds of Outage Probability with Nonzero Boresight Pointing ErrorsLK = r5 , we utilize pythagorean theorem to derive the coordinates of each rectangle vertexes as:M(√r2 − ( r5)2, 0), A(√r2 − ( r5)2, r5), B(√r2 − (2r5 )2, 2r5 ), Q(√r2 − (2r5 )2, r5),C(√r2 − (3r5 )2, 3r5 ), P (√r2 − (3r5 )2, 2r5 ), D(√r2 − (4r5 )2, 4r5 ), N (√r2 − (4r5 )2, 3r5 ). Othercoordinates of rectangle vertexes in the quadrant II of Figure 5.11 can be obtained through thesymmetry property.Considering the general case, we inscribe N − 1(N ≥ 2) rectangles with the same height ofrN =P−1r (γth)N on the upper semicircle in Figure 5.11. Following the derivation process of N = 5,the vertex coordinates of nth rectangle can be presented as Figure 5.12.22 ( 1),nr n rr BN N         22 ,nr nrr AN N         22 n ( 1),r n rD rN N        22 n ,r nrC rN N        Figure 5.12: Coordinates of nth inscribed rectangle on the upper semicircle.Following the similar process in Section 5.4.1, we express the integration of (5.7) over the entirerectangle region inscribed the upper semicircle asPUB1,BM(P−1r (γth))=14N−1∑n=1[erf(µy − (n−1)P−1r (γth)N√2σcL)−erf(µy − nP−1r (γth)N√2σcL)]×erfµx +√P−1r (γth)2 −(nP−1r (γth)N)2√2σuL−erfµx −√P−1r (γth)2 −(nP−1r (γth)N)2√2σuL , N ≥ 2.(5.23)For the integration over the entire rectangle region inscribed the lower semicircle, we have675.4. Bounds of Outage Probability with Nonzero Boresight Pointing ErrorsPUB2,BM(P−1r (γth))=14N−1∑n=1[erf(µy +nP−1r (γth)N√2σcL)−erf(µy +(n−1)P−1r (γth)N√2σcL)]×erfµx +√P−1r (γth)2 −(nP−1r (γth)N)2√2σuL−erfµx −√P−1r (γth)2 −(nP−1r (γth)N)2√2σuL , N ≥ 2.(5.24)According to (5.23) and (5.24), the upper bound of Beckmann CDF evaluated at P−1r (γth) ispresented asFr,UB(P−1r (γth))= PUB1,BM(P−1r (γth))+ PUB2,BM(P−1r (γth)). (5.25)Thus the upper bound of outage probability with nonzero boresight pointing errors can be expressedasPout,BM ≥1− Fr,UB(P−1r (γth))= 1− PUB1,BM(P−1r (γth))− PUB2,BM (P−1r (γth)) . (5.26)5.4.3 Discussion on the Tightness of the Outage Probability BoundsIn (5.22) and (5.26), we have, respectively, showed the lower and upper bounds of outage proba-bility with nonzero boresight pointing errors. In order to prove that the outage probability boundscan be made arbitrarily tight and approach the exact outage probability, we are required to provethat the bounds of Beckmann CDF approaches the exact CDF of Beckmann distribution. Thus, weneed to prove that for any given radius of circle r(r ≥ 0)limN→∞|Fr,LB (r)− Fr,UB (r) | = 0 (5.27)or equivalently,limN→∞[∑2 (SLB − SUB)]= 0 (5.28)where SLB and SUB denote the integration area of one rectangle for calculating lower and upperbounds of Beckmann CDF respectively (Figure 5.13).685.4. Bounds of Outage Probability with Nonzero Boresight Pointing ErrorsO XYFigure 5.13: Demonstration of the area Sshade =∑2 (SLB − SUB). A factor of 2 indicates thesymmetry of rectangle area in lower and upper semicircles.In Figure 5.9, the area of one rectangle is SLB =2rN√r2 − (n− 1)2 r2N2. Similarly, we can alsoobtain the area of one rectangle in Figure 5.11 as SUB =2rN√r2 − n2 r2N2. Thus (5.28) can beexpressed aslimN→∞∑2 (SLB − SUB) = limN→∞[N−1∑n=1(4rN√r2 − (n− 1)2 r2N2− 4rN√r2 − n2 r2N2)+4r2N2]= limN→∞[N−1∑n=14r2N2(√N2 − (n− 1)2 −√N2 − n2)+4r2N2] (5.29)whereN−1∑n=1(√N2 − (n− 1)2 −√N2 − n2)= N −√2N − 1. Then (5.29) can be presented as:limN→∞2 (SLB − SUB) = limN→∞[4r2N2(N −√2N − 1)+4r2N2]= limN→∞[4r2N2(N −√2N − 1)]+ limN→∞(4r2N2)=0.(5.30)695.5. Numerical ResultsThus we have limN→∞∑2 (SLB − SUB) = 0. We conclude that the limit in (5.27) holds and that thelower and upper bounds of Beckmann CDF in (5.21) and (5.25) can be made approach the exactCDF of Beckmann distribution. The bounds of outage probability in (5.22) and (5.26) can alsoapproach the exact outage probability in (5.16) with arbitrary accuracy.5.5 Numerical ResultsIn this section, we adopt a 532 nm laser source and the system settings shown in Table 5.1[10, 11]. Based on these parameters, we carry out the outage performance of a vertical bouy-basedUWOC system.Table 5.1: Summary of simulation parametersParameters c(λ) σ0 L D γth Rx FOV Tx divergenceValues 2.19/0.3m−1 0.1 5m/10m 5 cm 0.01 180◦ 0.01◦5.5.1 Outage Probability with Zero Boresight Pointing ErrorsThe outage probability with nonzero boresight pointing errors in (5.15) is evaluated and verifiedthrough Monte Carlo simulation in both coastal and harbor water. Figures 5.14 and 5.15 demon-strate the outage probability with different transmit power for 5m and 10m link distance in costaland harbor water, respectively.By comparing Figure 5.14 and Figure 5.15, we have found that the decrement of transmit poweror increment of wind speed can increase the outage probability. We also observe that, with thesame transmit power, the increase of water turbidity and link distance will also degrade the outageperformance. In Figure 5.15, the outage probability increases from 10−10 to 10−2 approximately asthe wind speed U increases from 2 m/s to 5 m/s when the transmit power is fixed at 15 dBm incoastal water. While in harbor water which is more turbid, the outage probability changes from10−11 to 10−7 as U varies for a fixed transmit power of 18 dBm, which indicates that the outageprobability is less sensitive to the wind speed in harbor water than that in coastal water. Similarphenomena can be observed from Figure 5.14 for a relatively short link range of 5 m. Hence, theincrement of seawater turbidity may diminish the outage performance degradation caused by thewind and enhances the link reliability. This is due to the fact that the multiple scattering processin turbid medium may strongly spread the beam spatially [10].705.5. Numerical Results(a) Coastal water c = 0.3/m.(b) Harbor water c = 2.19/m.Figure 5.14: Outage probability of a vertical buoy-based UWOC system with zero boresight pointingerrors. Link distance L = 5m715.5. Numerical Results(a) Coastal water c = 0.3/m.(b) Harbor water c = 2.19/m.Figure 5.15: Outage probability of a vertical buoy-based UWOC system with zero boresight pointingerrors. Link distance L = 10m725.5. Numerical Results5.5.2 Outage Probability Bounds with Nonzero Boresight Pointing ErrorsIn Figure 5.16(a), we plot the exact outage probability with nonzero boresight pointing errorsin (5.16) and its bounds in (5.22) and (5.26) respectively. In this case, we let N = 15.In order to show the tightness of the bounds can be improved with the increment of N , we alsodemonstrate the bounds with N = 100 in Figure 5.16(b). From Figure 5.16(b), we observe thatthe bounds become much tighter than in Figure 5.16(a). When N approaches infinity, the lowerand upper bounds of outage probability will converge to the exact outage probability with nonzeroboresight pointing errors.By comparing Figure 5.16 with Figure 5.14(a), we have also found that, the UWOC system withimpact of zero boresight pointing errors needs to consume 8dBm transmit power to maintain anoutage probability of 10−5. But for the system with the impact of non-zero pointing errors, 10dBmtransmit power is needed to maintain the same outage probability of 10−5. This phenomenonindicates that the UWOC system requires higher transmit power to maintain the same outageprobability level with the impact of nonzero boresight pointing errors.735.5. Numerical Results(a) N = 15.(b) N = 100.Figure 5.16: Outage probability and its bound with nonzero boresight pointing errors. µx = 0.01,µy = 0.02, U = 2m/s, c = 0.3m−1, L = 5m, D = 5cm, γth = 0.01.745.6. Summary5.6 SummaryIn this chapter, we introduced the pointing error models for the buoy-based UWOC system.Based on this model, we developed a novel method to derive the outage probability of verticalbuoy-based UWOC links using IM/DD OOK. A closed-form expression for the outage probabilitywith zero boresight pointing errors was obtained. We also derived the closed-form bounds of outageprobability with nonzero boresight pointing errors. The bounds were proved to converge to theexact outage probability when N approaches infinity. At the end of this chapter, we demonstratedthe numerical results of the derived outage performance.75Chapter 6ConclusionsIn this chapter, we conclude the thesis by summarizing the contributions of this work andsuggesting some potential further research topics.6.1 Summary of ContributionsIn this thesis, we have provided a comprehensive technical survey for the research of UWOC.This survey covers three comprehensive aspects of the state-of-the-art of UWOC research in theperspective of communication engineering: UWOC channel modeling, modulation and coding tech-nologies, and experimental UWOC discoveries. The summarization that we’ve made can providea comprehensive overview of UWOC as well as potential research directions for the scholars andengineers who are working on this area. In order to conclude the thesis, we will summarize thecontributions as follows:− In Chapter 1, we have introduced the history and current development of UWOC. Severalsignificant discoveries of UWOC have been stated. We have also carried out four link config-urations that are widely implemented in UWOC systems: point-to-point LOS, diffused LOS,retroreflector-based LOS and NLOS configurations. The corresponding characterizations andapplication scenarios of each link configurations have also been explained. In the second partof this chapter, we have introduced the advantages and limitations of UWOC by comparingit with other conventional UWC carriers such as acoustic and RF waves.− Chapter 2 has investigated the channel modeling of UWOC. We have firstly introduced severalbasic properties of light propagation in water which provide a foundation for UWOC channelmodeling. The concept of absorption and scattering coefficients as well as their characteri-zations in different water types have been stated. Secondly, we have presented the modelingwork of aquatic optical attenuation in LOS and NLOS configurations. Several UWOC chan-nel modeling approaches that include analytical and numerical solutions of RTE, stochastics766.2. Suggested Future Workmodeling of aquatic optical attenuation have been introduced. Finally, we have demonstratedthe modeling of link misalignment and turbulence in UWOC. A summary of each UWOCchannel modeling work has been given at the end of this Chapter.− In Chapter 3, we have studied the channel modulation and coding techniques that applied inUWOC. Several conventional intensity modulation schemes such as OOK, PPM, DPIM andtheir applications in UWOC systems have been presented. We have also presented the imple-mentations of both simple and complex error correction codes in UWOC. The characterizationand performance of each codes in UWOC system have also been presented.− Chapter 4 has demonstrated the recent progress of UWOC experimental research. Firstly, wehave explained the architecture of a typical LOS UWOC experimental system. The recentresearch works based on this architecture have been introduced. We have also presented theUWOC experimental demonstrations with other link configurations such as diffused LOS andNLOS. Secondly, we have discussed several popular topics of experimental UWOC researchwhich include retroreflector, smart transceivers, UWOC for underwater vehicles and hybridUWOC systems. The recent progress in each of these topics has been presented in details.Finally, we have summarized the literatures of experimental UWOC in recent years.− Chapter 5 has studied the outage performance of vertical buoy-based UWOC links usingIM/DD OOK with zero and nonzero boresight pointing errors. Firstly, we have introduced thepointing error models with zero and nonzero boresight. Secondly, we have introduced the beamspread function which describes the propagation of light in underwater environment. Thirdly,we have analyzed the outage probability with zero and nonzero boresight pointing errors. Aclosed-form outage probability with zero boresight pointing errors has been achieved. We havealso derived the closed-form outage probability bounds with nonzero boresight pointing errors.These bounds can be made arbitrarily tight and approach the exact outage probability.6.2 Suggested Future WorkAlthough considerable research work on UWOC have already been proposed during the pastfew years, large scale commercial applications of UWOC systems have not been realized so far.There are still several challenges in this technology that need to be overcome. According to theprevious survey and investigation of UWOC systems, we provide several potential directions for776.2. Suggested Future Workfuture UWOC research as follows:− On the aspect of UWOC channel modeling, although lots of modeling work focusing on thehorizontal LOS configuration have been demonstrated, few channel models have considered thevertical link. Compared with horizontal link configuration, vertical links take into account thevariations of refractive index with the depth and temperature, which is a challenging task forfuture UWOC research. In FSO communications, there are a lot of works considering channelturbulence. However, in UWOC research, turbulence of water has not been fully considered.Another issue is to derive an accurate model for the NLOS UWOC. The modeling of LOSUWOC are relatively mature, several close-to-reality models have already been demonstrated.But for NLOS UWOC, such accurate models have not been proposed so far.− There are also several potential research directions in the aspect of UWOC transceivers. Formost theoretical UWOC research, the impact of transceiver noise to UWOC has not been fullyinvestigated. It’s necessary to study the noise model of UWOC transceivers and use them toevaluate the system performance. As we have presented in the previous chapters, link misalign-ment is an inevitable phenomena. Although a few research works on smart transceivers forovercoming link misalignment have been proposed, considering the rapid growing of UWSNs,there’s still a need for developing highly intelligent UWOC transceivers. The next genera-tion UWSNs require high bandwidth, energy-efficient, and compact UWOC transceivers tobe large-scaly implemented in AUVs, ROVs, and underwater sensor nodes. Thus, there’shuge research potential for developing more advanced and low-cost transmission light sources,receiving devices, as well as energy preservation system for the next-generation UWSNs.− The design of appropriate modulation and coding schemes that can adapt the characterizationsof underwater environment is another potential research direction. In recent years, researchershave implemented almost all the conventional optical modulation and coding techniques inUWOC. These schemes have been proved useful and improved the system performance. How-ever, few implementations have considered to design a modulation or coding scheme that candynamically adapt to the link characterization. Since most UWOC systems are embeddedon a battery-powered platform, the energy efficiency is thus considered to be important. Ifthere is a mechanism can adaptively switch modulation and coding schemes according to theturbidity of water (applying simple weak error coding scheme in clear water, complex powerfulerror coding scheme in turbid water), then the system will save considerable energy and have786.2. Suggested Future Worklonger cruising time.− Suitable network protocols are also needed for the UWOC. To this end, most of the researchwork on UWOC are mainly focusing on the physical layer such as channel modeling, modula-tion and channel coding. Only few studies on UWOC networks have been demonstrated so far[176, 177]. Considering the unique characterization of wireless optical channel in underwaterenvironment, novel efficient network protocols need to be proposed.− In Chapter 5, we have derived the closed form bounds of Beckmann CDF. But a closed-formbounds or approximation of Beckmann PDF have not been proposed. Therefore, a futureresearch topic is to develop accurate closed-form approximation or bounds of Beckmann PDF.These approximation and bounds can be used to analyze BER performance of vertical buoy-based UWOC system with boresight pointing errors.79Bibliography[1] X. Che, I. Wells, G. Dickers, P. Kear, and X. Gong, “Re-evaluation of RF electromagneticcommunication in underwater sensor networks,” IEEE Communications Magazine, vol. 48,pp. 143–151, Dec. 2010. → pages vi, 7, 10[2] N. G. Jerlov, Marine Optics. New York: Elsevier Scientific Publishing, 1976. → pages vi,14, 18[3] T. Jacobson, “Phys104,” 2008. [Online]. Available: http://www.physics.umd.edu/grt/taj/104a/104anotessupps.html → pages vii, 3[4] D. Pompili and I. F. Akyildiz, “Overview of networking protocols for underwater wirelesscommunications,” IEEE Communications Magazine, vol. 47, pp. 97–102, Jan. 2009. → pagesvii, 2, 4, 7, 8[5] C. D. Mobley, Light and Water: Radiative Transfer in Natural Waters. San Diego, CA:Academic press, 1994. → pages vii, 12, 13, 14, 15, 18, 19, 23[6] D. Anguita, D. Brizzolara, G. Parodi, and Q. Hu, “Optical wireless underwater communicationfor auv: Preliminary simulation and experimental results,” in Proc. MTS/IEEE OCEANS(OCEANS’11), Santander, Spain, Jun. 2011, pp. 1–5. → pages vii, 16, 17, 48, 53[7] M. A. Chancey, Short Range Underwater Optical Communication Links. North CarolinaState University: M.S. Thesis, 2005. → pages 41, 53[8] J. A. Simpson, A 1 Mbps Underwater Communications System using LEDs and Photodiodeswith Signal Processing Capability. North Carolina State University: M.S. Thesis, 2007. →pages vii, 16, 17, 41, 53[9] C. Gabriel, M. Khalighi, S. Bourennane, P. Le´on, and V. Rigaud, “Investigation of suitablemodulation techniques for underwater wireless optical communication,” in Proc. International80BibliographyWorkshop on Optical Wireless Communications (IWOW’12), Pisa, Italy, Oct. 2012, pp. 1–3.→ pages vii, 32, 34, 35, 38[10] Y. Dong, S. Tang, and X. Zhang, “Effect of random sea surface on downlink underwaterwireless optical communications,” IEEE Communications Letters, vol. 17, pp. 2164–2167,Nov. 2013. → pages vii, 25, 27, 30, 57, 58, 62, 70[11] B. M. Cochenour, L. J. Mullen, and A. E. Laux, “Characterization of the beam-spread functionfor underwater wireless optical communications links,” IEEE Journal of Oceanic Engineering,vol. 33, pp. 513–521, Oct. 2008. → pages viii, 18, 21, 23, 26, 27, 30, 60, 61, 70[12] A. G. Bell, “On the production and reproduction of sound by light,” American Journal ofSciences, vol. 3, pp. 305–324, Oct. 1880. → pages 1[13] D. Killinger, “Free space optics for laser communication through the air,” Optics & PhotonicsNews, vol. 13, pp. 36–42, Oct. 2002. → pages 1[14] B. A. Lengyel, Lasers: Generation of Light by Stimulated Emission. London, UK: Wiley,1962. → pages 1[15] S. Q. Duntley, “Light in the sea,” Journal of the Optical Society of America A, vol. 53, pp.214–233, Feb. 1963. → pages 1, 8[16] G. Gilbert, T. Stoner, and J. Jernigan, “Underwater experiments on the polarization, co-herence, and scattering properties of a pulsed blue-green laser,” in Proc. Underwater PhotoOptics I, vol. 7, Santa Barbara, USA, Jan. 1966, p. 3. → pages 1[17] S. Karp, “Optical communications between underwater and above surface (satellite) termi-nals,” IEEE Transactions on Communications, vol. 24, pp. 66–81, Jan. 1976. → pages 2[18] M. Callaham, “Submarine communications,” IEEE Communications Magazine, vol. 19, pp.16–25, Nov. 1981. → pages 2[19] J. Puschell, R. Giannaris, and L. Stotts, “The autonomous data optical relay experiment:first two way laser communication between an aircraft and submarine,” in Proc. NationalTelesystems Conference (NTC’92), Washington, DC, May 1992, pp. 14/27–14/30. → pages 281Bibliography[20] T. Wiener and S. Karp, “The role of blue/green laser systems in strategic submarine com-munications,” IEEE Transactions on Communications, vol. 28, pp. 1602–1607, Sep. 1980. →pages 2[21] M.-A. Khalighi, C. Gabriel, T. Hamza, S. Bourennane, P. Leon, and V. Rigaud, “Underwaterwireless optical communication; recent advances and remaining challenges,” in Proc. 16thInternational Conference on Transparent Optical Networks (ICTON’14), Jul. 2014, pp. 1–4.→ pages 2[22] L. J. Johnson, F. Jasman, R. J. Green, and M. S. Leeson, “Recent advances in underwateroptical wireless communications,” Underwater Technology, vol. 32, pp. 167–175, Nov. 2014.→ pages 2, 6, 12, 28[23] S. Arnon, “Underwater optical wireless communication network,” Optical Engineering, vol. 49,pp. 015 001–1, Jan. 2010. → pages 6, 25[24] S. Arnon and D. Kedar, “Non-line-of-sight underwater optical wireless communication net-work,” Journal of Optical Society of America A, vol. 26, pp. 530–539, Jan. 2009. → pages 6,25, 30[25] S. Tang, Y. Dong, and X. Zhang, “On path loss of NLOS underwater wireless optical commu-nication links,” in Proc. MTS/IEEE OCEANS (OCEANS’13), Bergen, Norway, Jun. 2013,pp. 1–3. → pages 6, 25, 30[26] B. Truax, Acoustic Communication. Santa Barbara, CA: Greenwood Publishing Group,2001. → pages 7[27] E. M. Sozer, M. Stojanovic, and J. G. Proakis, “Underwater acoustic networks,” IEEE Journalof Oceanic Engineering, vol. 25, pp. 72–83, Jan. 2000. → pages 7[28] J. Partan, J. Kurose, and B. N. Levine, “A survey of practical issues in underwater networks,”ACM SIGMOBILE Mobile Computing and Communications Review, vol. 11, pp. 23–33, Sep.2007. → pages 7[29] W. W. Au, P. E. Nachtigall, and J. L. Pawloski, “Acoustic effects of the atoc signal (75 hz,195 db) on dolphins and whales,” The Journal of the Acoustical Society of America, vol. 101,pp. 2973–2977, Jan. 1997. → pages 782Bibliography[30] L. J. Johnson, R. J. Green, and M. S. Leeson, “Underwater optical wireless communications:depth dependent variations in attenuation,” Applied Optics, vol. 52, pp. 7867–7873, Oct. 2013.→ pages 8, 19, 26, 27, 30[31] R. W. Spinrad, K. L. Carder, and M. J. Perry, Ocean Optics. Oxford: Clarendon Press,1994. → pages 12[32] A. Morel and H. Loisel, “Apparent optical properties of oceanic water: dependence on themolecular scattering contribution,” Applied Optics, vol. 37, no. 21, pp. 4765–4776, Jul. 1998.→ pages[33] H. R. Gordon, O. B. Brown, and M. M. Jacobs, “Computed relationships between the inherentand apparent optical properties of a flat homogeneous ocean,” Applied Optics, vol. 14, pp.417–427, Feb. 1975. → pages[34] R. E. Green and H. M. Sosik, “Analysis of apparent optical properties and ocean color modelsusing measurements of seawater constituents in new england continental shelf surface waters,”Journal of Geophysical Research: Oceans (1978–2012), vol. 109, Mar. 2004. → pages 12[35] L. Johnson, R. Green, and M. Leeson, “A survey of channel models for underwater opticalwireless communication,” in Proc. 2nd International Workshop on Optical Wireless Commu-nications (IWOW’13), Newcastle upon Tyne, UK, Oct. 2013, pp. 1–5. → pages 13, 21[36] C. Mobley, E. Boss, and C. Roesler, “Ocean optics web book,” 2010. [Online]. Available:http://www.oceanopticsbook.info/ → pages 14, 19, 21, 30[37] C. Gabriel, M. Khalighi, S. Bourennane, P. Leon, and V. Rigaud, “Channel modelingfor underwater optical communication,” in Proc. IEEE Global Communications Conference(GLOBECOM’11), Houston, TX, Dec. 2011, pp. 833–837. → pages 14, 18, 19, 20, 22, 30[38] K. Shifrin, Physical Optics of Ocean Water. New York: American Institute of Physics, 1988.→ pages 14[39] R. C. Smith and K. S. Baker, “Optical properties of the clearest natural waters (200–800nm),” Applied Optics, vol. 20, pp. 177–184, Jan. 1981. → pages 14[40] A. Bricaud, A. Morel, and L. Prieur, “Absorption by dissolved organic matter of the sea(yellow substance) in the uv and visible domains,” Limnology and Oceanography, vol. 26, pp.43–53, Jan. 1981. → pages 1483Bibliography[41] W. Breves and R. Reuter, “Bio-optical properties of gelbstoff in the arabian sea at the onsetof the southwest monsoon,” Journal of Earth System Science, vol. 109, pp. 415–425, Dec.2000. → pages 14[42] V. I. Haltrin, “Chlorophyll-based model of seawater optical properties,” Applied Optics,vol. 38, pp. 6826–6832, Nov. 1999. → pages 15, 30[43] D. A. Hansell and C. A. Carlson, Biogeochemistry of Marine Dissolved Organic Matter. SanDiego, CA: Academic Press, 2014. → pages 15[44] J. R. Apel, Principles of Ocean Physics. London, UK: Academic Press, 1987. → pages 15,19[45] M. Kishino, N. Okami, M. Takahashi, and S.-e. Ichimura, “Light utilization efficiency andquantum yield of phytoplankton in a thermally stratified sea1,” Limnology and Oceanography,vol. 31, pp. 557–566, Dec. 1986. → pages 15[46] C. D. Mobley, B. Gentili, H. R. Gordon, Z. Jin, G. W. Kattawar, A. Morel, P. Reinersman,K. Stamnes, and R. H. Stavn, “Comparison of numerical models for computing underwaterlight fields,” Applied Optics, vol. 32, pp. 7484–7504, Dec. 1993. → pages 18, 21, 23, 30[47] F. Hanson and S. Radic, “High bandwidth underwater optical communication,” Applied Op-tics, vol. 47, pp. 277–283, Jan. 2008. → pages 18, 22, 30[48] A. L. Alldredge and M. W. Silver, “Characteristics, dynamics and significance of marinesnow,” Progress in Oceanography, vol. 20, pp. 41–82, Apr. 1988. → pages 19[49] L. J. Johnson, R. J. Green, and M. S. Leeson, “Underwater optical wireless communications:depth-dependent beam refraction,” Applied Optics, vol. 53, pp. 7273–7277, Oct. 2014. →pages 19, 27, 30[50] C. Gabriel, M.-A. Khalighi, S. Bourennane, P. Le´on, and V. Rigaud, “Monte-carlo-basedchannel characterization for underwater optical communication systems,” IEEE/OSA Journalof Optical Communications and Networking, vol. 5, pp. 1–12, Jan. 2013. → pages 20, 22, 30,60[51] W. Cox and J. Muth, “Simulating channel losses in an underwater optical communicationsystem,” Journal of the Optical Society of America A, vol. 31, pp. 920–934, May 2014. →pages 23, 3084Bibliography[52] D. Toublanc, “Henyey-greenstein and mie phase functions in monte carlo radiative transfercomputations,” Applied Optics, vol. 35, pp. 3270–3274, Jun. 1996. → pages[53] V. I. Haltrin, “One-parameter two-term henyey-greenstein phase function for light scatteringin seawater,” Applied Optics, vol. 41, pp. 1022–1028, Feb. 2002. → pages 30[54] ——, “Two-term henyey-greenstein light scattering phase function for seawater,” in Proc.IEEE International Geoscience and Remote Sensing Symposium (IGARSS’99), Hamburg,Germany, Jun. 1999, pp. 1423–1425. → pages 20[55] J. Smart, “Underwater optical communications systems part 1: variability of water opticalparameters,” in Proc. IEEE Military Communications Conference (MILCOM’05), AtlanticCity, NJ, Oct. 2005, pp. 1140–1146. → pages 20, 30[56] J. W. Giles and I. N. Bankman, “Underwater optical communications systems. part 2: basicdesign considerations,” in Proc. IEEE Military Communications Conference (MILCOM’05),Atlantic City, NJ, Oct. 2005, pp. 1700–1705. → pages 20, 30[57] H. Tan, R. Doerffer, T. Oishi, A. Tanaka et al., “A new approach to measure the volumescattering function,” Optics Express, vol. 21, pp. 18 697–18 711, Aug. 2013. → pages 20[58] S. Arnon, J. Barry, G. Karagiannidis, R. Schober, and M. Uysal, Advanced Optical WirelessCommunication Systems, New York, 2012. → pages 21, 30[59] C. Li, K.-H. Park, and M.-S. Alouini, “A direct radiative transfer equation solver for pathloss calculation of underwater optical wireless channels,” CEMSE division, KAUST, Thuwal,Saudi Arabia, Tech. Rep. 333892, May 2014. → pages 21, 23, 30[60] ——, “On the use of a direct radiative transfer equation solver for path loss calculationin underwater optical wireless channels,” IEEE Communications Letters, 2015, accepted forpublication. → pages 21, 23[61] H. C. Van de Hulst, Multiple Light Scattering: Tables, Formulas, and Applications. NewYork: Academic Press, 2012. → pages 21, 30[62] S. Jaruwatanadilok, “Underwater wireless optical communication channel modeling and per-formance evaluation using vector radiative transfer theory,” IEEE Journal on Selected Areasin Communications, vol. 26, pp. 1620–1627, Dec. 2008. → pages 21, 33, 3885Bibliography[63] B. Cochenour, L. Mullen, and A. Laux, “Spatial and temporal dispersion in high bandwidthunderwater laser communication links,” in Proc. IEEE Military Communications Conference(MILCOM’08), San Diego, CA, Nov. 2008, pp. 1–7. → pages 21, 30, 42, 53[64] J. Potter et al., “Ucomms: A conference and workshop on underwater communications, chan-nel modeling, and validation,” IEEE Journal of Oceanic Engineering, vol. 38, pp. 603–613,Oct. 2013. → pages 21, 24[65] L. Wang, S. L. Jacques, and L. Zheng, “Monte carlo modeling of light transport in multi-layered tissues,” Computer Methods and Programs in Biomedicine, vol. 47, pp. 131–146, Jul.1995. → pages 21, 30[66] R. M. Lerner and J. D. Summers, “Monte carlo description of time- and space-resolved mul-tiple forward scatter in natural water.” Applied Optics, vol. 21, pp. 861–869, Mar. 1982. →pages 21[67] R. A. Leathers, T. V. Downes, C. O. Davis, and C. D. Mobley, “Monte carlo radiative transfersimulations for ocean optics: a practical guide,” Naval Research Laboratory, Washington, DC,Tech. Rep. NRL/MR/5660–04-8819, Sep. 2004. → pages 21, 30[68] J. Li, Y. Ma, Q. Zhou, B. Zhou, and H. Wang, “Monte carlo study on pulse response ofunderwater optical channel,” Optical Engineering, vol. 51, pp. 066 001–1, Jun. 2012. → pages22, 30[69] J. Li, Y. Ma, Q. Zhou, H. Wang, and B. Zhou, “Channel capacity study of underwater wirelessoptical communications links based on monte carlo simulation,” Journal of Optics, vol. 14, p.015403, Dec. 2011. → pages 22, 30[70] F. R. Dalgleish, F. M. Caimi, A. K. Vuorenkoski, W. B. Britton, B. Ramos, T. E. Giddings,J. J. Shirron, and C. H. Mazel, “Efficient laser pulse dispersion codes for turbid undersea imag-ing and communications applications,” Proceedings of SPIE, vol. 7678, pp. 76 780I–76 780I,Apr. 2010. → pages 22, 30, 41[71] B. Edouard, Multiple Scattering of Light in Optical Diagnostics of Dense Sprays and OtherComplex Turbid Media. Cranfield University: Ph.D Thesis, 2006. → pages 23[72] W. Cox, “Photonator,” 2012. [Online]. Available: https://github.com/gallamine/Photonator→ pages 23, 3086Bibliography[73] H. Zhang, Y. Dong, and X. Zhang, “On stochastic model for underwater wireless opticallinks,” in Proc. IEEE/CIC International Conference on Communications in China (ICCC’14),Shanghai, China, Oct. 2014, pp. 156–160. → pages 23, 24, 30[74] H. Zhang and Y. Dong, “General stochastic channel model and performance evaluation forunderwater wireless optical links,” IEEE Transactions on Communications, 2015, acceptedfor publication. → pages 23, 24[75] H. Zhang, L. Hui, and Y. Dong, “Angle of arrival analysis for underwater wireless opticallinks,” IEEE Communications Letters, 2015, accepted for publication. → pages 24[76] S. Tang, X. Zhang, and Y. Dong, “On impulse response for underwater wireless optical links,”in Proc. MTS/IEEE OCEANS (OCEANS’13), Bergen, Norway, Jun. 2013, pp. 1–4. → pages24, 30[77] S. Tang, Y. Dong, and X. Zhang, “Impulse response modeling for underwater wireless opticalcommunication links,” IEEE Transactions on Communications, vol. 62, pp. 226–234, Jan.2014. → pages 24, 30[78] Y. Dong, H. Zhang, and X. Zhang, “On impulse response modeling for underwater wirelessoptical mimo links,” in Proc. 2014 IEEE/CIC International Conference on Communicationsin China (ICCC’14), Shanghai, China, Oct. 2014, pp. 151–155. → pages 24, 30[79] M. Doniec, M. Angermann, and D. Rus, “An end-to-end signal strength model for underwateroptical communications,” IEEE Journal of Oceanic Engineering, vol. 38, pp. 743–757, Oct2013. → pages 24, 25, 30, 36, 38, 42, 53[80] C. Cox and W. Munk, “Slopes of the sea surface deduced from photographs of sun glitter,”Tech. Rep., Sep. 1956. → pages 25, 30[81] W. J. Plant, “A new interpretation of sea-surface slope probability density functions,” Journalof Geophysical Research: Oceans, vol. 108, pp. 111–113, Sep. 2003. → pages 25, 30[82] A. Choudhary, V. Jagadeesh, and P. Muthuchidambaranathan, “Pathloss analysis of nlosunderwater wireless optical communication channel,” in Proc. International Conference onElectronics and Communication Systems (ICECS’14), Coimbatore, India, Feb. 2014, pp. 1–4.→ pages 2587Bibliography[83] A. Choudhary, F. M. Bui, and P. Muthuchidambaranathan, “Characterization of channel im-pulse responses for nlos underwater wireless optical communications,” in Proc. Fourth Inter-national Conference on Advances in Computing and Communications (ICACC’2014), Kochi,India, Aug. 2014, pp. 77–79. → pages 25, 30[84] S. Tang, Y. Dong, and X. Zhang, “On link misalignment for underwater wireless opticalcommunications,” IEEE Communications Letters, vol. 16, pp. 1688–1690, Oct. 2012. →pages 26, 27, 30[85] B. Cochenour, L. Mullen, and J. Muth, “Temporal response of the underwater optical channelfor high-bandwidth wireless laser communications,” IEEE Journal of Oceanic Engineering,vol. 38, pp. 730–742, Oct. 2013. → pages 27, 30, 42, 53[86] H. Zhang, L. Hui, and Y. Dong, “On capacity of downlink underwater wireless optical mi-mo systems with random sea surface,” IEEE Communications Letters, 2015, accepted forpublication. → pages 27[87] C. Gabriel, M.-A. Khalighi, S. Bourennane, P. Le´on, and V. Rigaud, “Misalignment consid-erations in point-to-point underwater wireless optical links,” in Proc. MTS/IEEE OCEANS(OCEANS’13), Bergen, Norway, Jun. 2013, pp. 1–5. → pages 27, 30[88] L. J. Johnson, R. J. Green, and M. S. Leeson, “The impact of link orientation in underwateroptical wireless communication systems,” in Proc. MTS/IEEE OCEANS (OCEANS’14), St.John’s, NL, Sep. 2014, pp. 1–8. → pages 28, 30[89] F. Hanson and M. Lasher, “Effects of underwater turbulence on laser beam propagation andcoupling into single-mode optical fiber,” Applied Optics, vol. 49, pp. 3224–3230, May 2010.→ pages 28, 30[90] W. Liu, Z. Xu, and L. Yang, “Simo detection schemes for underwater optical wireless com-munication under turbulence,” Photonics Research, vol. 3, pp. 48–53, Jun. 2015. → pages28[91] S. Tang, X. Zhang, and Y. Dong, “Temporal statistics of irradiance in moving turbulentocean,” in Proc. MTS/IEEE OCEANS (OCEANS’13), Bergen, Norway, Jun. 2013, pp. 1–4.→ pages 28, 2988Bibliography[92] V. Nikishov and V. Nikishov, “Spectrum of turbulent fluctuations of the sea-water refractionindex,” International Journal of Fluid Mechanics Research, vol. 27, pp. 82–98, Nov. 2000. →pages 28[93] O. Korotkova, N. Farwell, and E. Shchepakina, “Light scintillation in oceanic turbulence,”Waves in Random and Complex Media, vol. 22, pp. 260–266, Feb. 2012. → pages 28, 30[94] A. Laux, R. Billmers, L. Mullen, B. Concannon, J. Davis, J. Prentice, and V. Contarino, “Thea, b, cs of oceanographic lidar predictions: a significant step toward closing the loop betweentheory and experiment,” Journal of Modern Optics, vol. 49, pp. 439–451, Mar. 2002. → pages30, 40[95] N. Fair, A. Chave, L. Freitag, J. Preisig, S. White, D. Yoerger, and F. Sonnichsen, “Opticalmodem technology for seafloor observatories,” in Proc. MTS/IEEE OCEANS (OCEANS’06),Boston, MA, Sep. 2006, pp. 1–6. → pages 30, 42, 53[96] W. Liu, D. Zou, P. Wang, Z. Xu, and L. Yang, “Wavelength dependent channel characteriza-tion for underwater optical wireless communications,” in IEEE International Conference onSignal Processing, Communications and Computing (ICSPCC’14), Guilin, China, Aug. 2014,pp. 895–899. → pages 30[97] W. Liu, D. Zou, Z. Xu, and J. Yu, “Non-line-of-sight scattering channel modeling for underwa-ter optical wireless communication,” in IEEE International Conference on Cyber Technologyin Automation, Control, and Intelligent Systems (CYBER’2015), Shenyang, China, Jun. 2015,pp. 1265–1268. → pages 30[98] M. Khalighi and M. Uysal, “Survey on free space optical communication: A communicationtheory perspective,” IEEE Communications Surveys and Tutorials, vol. 16, pp. 2231–2258,Apr. 2014. → pages 33, 34, 35, 37, 39, 40, 41, 48[99] F. Akhoundi, J. Salehi, A. Tashakori et al., “Cellular underwater wireless optical cdma net-work: Performance analysis and implementation concepts,” IEEE Transactions on Commu-nications, vol. 63, pp. 882–891, Feb. 2015. → pages 33, 38, 42, 53[100] Z. Ahmad and R. Green, “Link design for multi-hop underwater optical wireless sensor net-work,” in Proc. 7th International Conference on Systems and Networks Communications (IC-SNC’12), Lisbon, Portugal, Nov. 2012, pp. 65–70. → pages 33, 38, 42, 5389Bibliography[101] X. He and J. Yan, “Study on performance of M-ary PPM underwater optical communicationsystems using vector radiative transfer theory,” in Proc. 10th International Symposium onAntennas, Propagation & EM Theory (ISAPE’12), Xian, China, Oct. 2012, pp. 566–570. →pages 33, 38[102] S. Meihong, Y. Xinsheng, and Z. Zhangguo, “The modified ppm modulation for underwa-ter wireless optical communication,” in Proc. International Conference on CommunicationSoftware and Networks, (ICCSN’09), Macau, China, Feb. 2009, pp. 173–177. → pages 33[103] H. Sari and B. Woodward, “Underwater voice communications using a modulated laser beam,”in Proc. MTS/IEEE OCEANS (OCEANS’98), Nice, France, Oct. 1998, pp. 1183–1188. →pages 33, 42, 53[104] M. Chen, S. Zhou, and T. Li, “The implementation of ppm in underwater laser communica-tion system,” in Proc. International Conference on Communications, Circuits and Systems(ICCCAS’06), Guilin, China, Jun. 2006, pp. 1901–1903. → pages[105] D. Anguita, D. Brizzolara, and G. Parodi, “Vhdl modeling of phy and mac layer modules forunderwater optical wireless communication,” in Proc. 5th European Conference on Circuitsand Systems for Communications (ECCSC’2010), Belgrade, Serbia, Nov. 2010, pp. 185–188.→ pages 41, 53[106] D. Anguita, D. Brizzolora, and G. Parodi, “Optical wireless communication for underwaterwireless sensor networks: Hardware modules and circuits design and implementation,” inProc. MTS/IEEE OCEANS (OCEANS’10), Seattle, WA, Sep. 2010, pp. 1–8. → pages 36,38, 41, 53[107] S. Tang, Y. Dong, and X. Zhang, “Receiver design for underwater wireless optical commu-nication link based on apd,” in Proc. 8th International Conference on Communications andNetworking in China (CHINACOM’13), Kunming, China, Aug. 2012, pp. 301–305. → pages46, 53[108] P. Swathi and S. Prince, “Designing issues in design of underwater wireless optical communi-cation system,” in Proc. International Conference on Communications and Signal Processing(ICCSP’14), Melmaruvathur, India, Apr. 2014, pp. 1440–1445. → pages 34, 42, 5390Bibliography[109] R. M. Hagem, D. V. Thiel, S. G. O’Keefe, and T. Fickenscher, “Optical wireless commu-nication for real time swimmers feedback: A review,” in Proc. International Symposium onCommunications and Information Technologies (ISCIT’12), Gold Coast, QLD, Oct. 2012, pp.1080–1085. → pages 33, 38, 41[110] M. Doniec, I. Vasilescu, M. Chitre, C. Detweiler, M. Hoffmann-Kuhnt, and D. Rus, “Aquaopti-cal: A lightweight device for high-rate long-range underwater point-to-point communication,”in Proc. MTS/IEEE OCEANS (OCEANS’09), Biloxi, MS, Oct. 2009, pp. 1–6. → pages 34,38, 42[111] M. Doniec and D. Rus, “Bidirectional optical communication with aquaoptical ii,” in Proc.IEEE International Conference on Communication Systems (ICCS’10), Singapor, Nov. 2010,pp. 390–394. → pages 42, 53[112] M. Doniec, C. Detweiler, I. Vasilescu, and D. Rus, “Using optical communication for re-mote underwater robot operation,” in Proc. IEEE/RSJ International Conference onIntelli-gent Robots and Systems (IROS’10), Taipei, China, Oct. 2010, pp. 4017–4022. → pages 34,38, 48, 53[113] Z. Wang, Y. Dong, X. Zhang, and S. Tang, “Adaptive modulation schemes for underwaterwireless optical communication systems,” in Proc. 7th ACM International Conference onUnderwater Networks and Systems (WUWNet’12). → pages 34, 38[114] B. Cochenour, L. Mullen, and A. Laux, “Phase coherent digital communications for wirelessoptical links in turbid underwater environments,” in Proc. MTS/IEEE OCEANS (OCEAN-S’07), Vancouver, BC, Canada, Sep. 2007, pp. 1–5. → pages 34, 38[115] M. Sui, X. Yu, and F. Zhang, “The evaluation of modulation techniques for underwater wire-less optical communications,” in Proc. International Conference on Communication Softwareand Networks (ICCSN’09), Macau, China, Feb. 2009, pp. 138–142. → pages 34, 38[116] W. C. Cox, B. L. Hughes, and J. F. Muth, “A polarization shift-keying system for underwateroptical communications,” in Proc. MTS/IEEE OCEANS (OCEANS’09), Biloxi, MS, Oct.2009, pp. 1–4. → pages 35, 38[117] Y. Dong, T. Zhang, and X. Zhang, “Polarized pulse position modulation for wireless optical91Bibliographycommunications,” in Proc. 47th Annual Conference on Information Sciences and Systems(CISS’13), Baltimore, MD, Mar. 2013, pp. 1–5. → pages 35[118] X. Zhang, Y. Dong, and S. Tang, “Polarization differential pulse position modulation,” in Proc.7th ACM International Conference on Underwater Networks and Systems (WUWNet’12), LosAngeles, CA, Nov. 2012, pp. 41:1–41:2. → pages 35, 38[119] W. C. Cox, J. A. Simpson, and J. F. Muth, “Underwater optical communication using soft-ware defined radio over led and laser based links,” in Proc. IEEE Military CommunicationsConference (MILCOM’11), Baltimore, MD, Nov. 2011, pp. 2057–2062. → pages 35, 38, 42,53[120] G. Cossu, R. Corsini, A. Khalid, S. Balestrino, A. Coppelli, A. Caiti, and E. Ciaramella, “Ex-perimental demonstration of high speed underwater visible light communications,” in Proc.2nd International Workshop on Optical Wireless Communications (IWOW’13), Newcastleupon Tyne, UK, Oct. 2013, pp. 11–15. → pages 35, 36, 38, 41, 53[121] J. G. Proakis and M. Salehi, Digital Communications, 5th ed. New York: McGraw-Hill,2007. → pages 36[122] W. C. Cox, J. A. Simpson, C. P. Domizioli, J. F. Muth, and B. L. Hughes, “An underwateroptical communication system implementing reed-solomon channel coding,” in Proc. MT-S/IEEE OCEANS (OCEANS’08), Quebec City, QC, Canaad, Sep. 2008, pp. 1–6. → pages36[123] J. A. Simpson, W. C. Cox, J. R. Krier, B. Cochenour, B. Hughes, and J. Muth, “5 mbpsoptical wireless communication with error correction coding for underwater sensor nodes,” inProc. MTS/IEEE OCEANS (OCEANS’10), Seattle, WA, pp. 6–9. → pages 36[124] W.-P. Wang and B. Zheng, “The simulation design of led-based close-range underwater op-tical communication system,” in Proc. 10th International Computer Conference on WaveletActive Media Technology and Information Processing (ICCWAMTIP’13), Chengdu, China,Dec. 2013, pp. 283–285. → pages 36, 38, 41, 53[125] M. Doniec, A. Xu, and D. Rus, “Robust real-time underwater digital video streaming usingoptical communication,” in Proc. IEEE International Conference on Robotics and Automation(ICRA)’13, Karlsruhe, Germany, May 2013, pp. 5117–5124. → pages 36, 38, 42, 5392Bibliography[126] J. Everett, Forward-Error Correction Coding for Underwater Free-space Optical Communica-tion. North Carolina State University: M.S. Thesis, 2009. → pages 37, 38[127] W. C. Cox, K. F. Gray, J. A. Simpson, B. Cochenour, B. L. Hughes, and J. F. Muth, “Amems blue/green retroreflecting modulator for underwater optical communications,” in Proc.MTS/IEEE OCEANS (OCEANS’10), Seattle, WA, USA, Sep. 2010, pp. 1–4. → pages 38,40, 45, 53[128] M. Sun, B. Zheng, L. Zhao, X. Zhao, and F. Kong, “A design of the video transmission basedon the underwater laser communication,” in Proc. MTS/IEEE OCEANS (OCEANS’14), St.John’s, NL, Sep. 2014, pp. 1–4. → pages 41, 43, 53[129] L. Mullen, A. Laux, and B. Cochenour, “Time-dependent underwater optical propagationmeasurements using modulated light fields,” in SPIE Defense, Security, and Sensing. Inter-national Society for Optics and Photonics, 2009, pp. 73 170D–73 170D. → pages[130] ——, “Propagation of modulated light in water: implications for imaging and communicationssystems,” Applied Optics, vol. 48, pp. 2607–2612, Apr. 2009. → pages[131] F. Schill, U. R. Zimmer, and J. Trumpf, “Visible spectrum optical communication and dis-tance sensing for underwater applications,” in Proc. Australasian Conference on Robotics andAutomation (ACRA’04), Canberry, Australia, Dec. 2004, pp. 1–8. → pages 41, 53[132] R. M. Hagem, S. G. O’Keefe, T. Fickenscher, and D. V. Thiel, “Self contained adaptableoptical wireless communications system for stroke rate during swimming,” IEEE SensorsJournal, vol. 13, pp. 3144–3151, Aug. 2013. → pages 41[133] H. Brundage, Designing a Wireless Underwater Optical Communication System. Mas-sachusetts Institute of Technology: M.S. Thesis, 2010. → pages 41[134] A. Destrez, Z. Toffano, and P. Leon, “Underwater high bit-rate optical free-space commu-nication system,” in Proc. International Workshop on Optical Wireless Communications (I-WOW’12), Pisa, Italy, Oct. 2012, pp. 1–3. → pages 42, 53[135] J. B. Snow, J. P. Flatley, D. E. Freeman, M. A. Landry, C. E. Lindstrom, J. R. Longacre,and J. A. Schwartz, “Underwater propagation of high-data-rate laser communications pulses,”Proceedings of SPIE, vol. 1750, pp. 419–427, Jul. 1992. → pages 42, 5393Bibliography[136] W. C. Cox Jr, A 1 Mbps Underwater Communication System Using a 405 nm Laser Diodeand Photomultiplier Tube. North Carolina State University: M.S. Thesis, 2008. → pages 42[137] P. A. Hiskett and R. A. Lamb, “Underwater optical communications with a single photon-counting system,” Proceedings of SPIE, vol. 9114, pp. 91 140P–91 140P–15, May 2014. →pages 43[138] I. Mizukoshi, N. Kazuhiko, and M. Hanawa, “Underwater optical wireless transmission of405nm, 968mbit/s optical im/dd-ofdm signals,” in Proc. OptoElectronics and Communica-tion Conference and Australian Conference on Optical Fibre Technology (OECC/ACOFT’14),Melbourne, VIC, Australia, Jul. 2014, pp. 216–217. → pages 43[139] J. A. Simpson, B. L. Hughes, J. F. Muth et al., “A spatial diversity system to measureoptical fading in an underwater communications channel,” in Proc. MTS/IEEE OCEANS(OCEANS’09), Biloxi, MS, Oct. 2009, pp. 1–6. → pages[140] G. Baiden, Y. Bissiri, and A. Masoti, “Paving the way for a future underwater omni-directionalwireless optical communication systems,” Ocean Engineering, vol. 36, pp. 633–640, Jul. 2009.→ pages[141] B. M. Cochenour, Experimental Measurements of Temporal Dispersion for Underwater LaserCommunications and Imaging. North Carolina State University: Ph.D. Thesis, 2013. →pages 43, 53[142] C. Pontbriand, N. Farr, J. Ware, J. Preisig, and H. Popenoe, “Diffuse high-bandwidth opticalcommunications,” in Proc. MTS/IEEE OCEANS (OCEANS’08), Quebec City, QC, Canada,Sep. 2008, pp. 1–4. → pages 43, 46, 53[143] B. Cochenour and L. Mullen, “Channel response measurements for diffuse non-line-of-sight (n-los) optical communication links underwater,” in Proc. MTS/IEEE OCEANS (OCEANS’11),Waikoloa, HI, Sep. 2011, pp. 1–5. → pages 43, 53[144] D. Alley, L. Mullen, and A. Laux, “Compact, dual-wavelength, non-line-of-sight (nlos) under-water imager,” in Proc. MTS/IEEE OCEANS (OCEANS’11), Santander, Spain, Jun. 2011,pp. 1–5. → pages 43, 5394Bibliography[145] B. Cochenour, L. Mullen, and J. Muth, “A modulated pulse laser for underwater detec-tion, ranging, imaging, and communications,” Proceedings of SPIE, vol. 8372, pp. 83 720S–1–83 720S–10, Jun. 2012. → pages 43, 53[146] P. G. Goetz, W. S. Rabinovich, R. Mahon, M. S. Ferraro, J. L. Murphy, H. Burris, M. Stell,C. I. Moore, M. R. Suite, W. Freeman et al., “Modulating retro-reflector devices and currentlink performance at the naval research laboratory,” in Proc. IEEE Military CommunicationsConference (MILCOM’07), Orlando, FL, USA, Oct. 2007, pp. 1–7. → pages 43, 45[147] W. S. Rabinovich, R. Mahon, H. R. Burris, G. C. Gilbreath, P. G. Goetz, C. I. Moore, M. Stell,M. J. Vilcheck, J. L. Witkowsky, L. Swingen et al., “Free-space optical communications link at1550 nm using multiple-quantum-well modulating retroreflectors in a marine environment,”Optical Engineering, vol. 44, pp. 056 001–056 001, May 2005. → pages 44[148] P. G. Goetz, W. S. Rabinovich, R. Mahon, J. L. Murphy, M. S. Ferraro, M. Suite, W. R.Smith, H. R. Burris, C. I. Moore, W. W. Schultz et al., “Modulating retro-reflector lasercomsystems for small unmanned vehicles,” IEEE Journal on Selected Areas in Communications,vol. 30, pp. 986–992, Jun. 2012. → pages 45[149] P. G. Goetz, W. S. Rabinovich, R. Mahon, J. L. Murphy, M. S. Ferraro, M. Suite, W. Smith,B. Xu, H. Burris, C. Moore et al., “Modulating retro-reflector lasercom systems at the navalresearch laboratory,” in Proc. IEEE Military Communications Conference (MILCOM’10),San Jose, CA, Oct. 2010, pp. 1601–1606. → pages 45[150] L. Mullen, B. Cochenour, W. Rabinovich, R. Mahon, and J. Muth, “Backscatter suppressionfor underwater modulating retroreflector links using polarization discrimination,” AppliedOptics, vol. 48, pp. 328–337, Jan. 2009. → pages 45, 53[151] B. Cochenour, L. Mullen, W. Rabinovich, and R. Mahon, “Underwater optical communica-tions with a modulating retro-reflector,” Proceedings of SPIE, vol. 7317, pp. 73 170G–73 170G,Apr. 2009. → pages 45, 53[152] J. Simpson, B. L. Hughes, J. F. Muth et al., “Smart transmitters and receivers for underwa-ter free-space optical communication,” IEEE Journal on Selected Areas in Communications,vol. 30, pp. 964–974, Jun. 2012. → pages 46, 5395Bibliography[153] H. Yoshida, T. Aoki, H. Osawa, S. Tsukioka, S. Ishibashi, Y. Watanabe, J. Tahara, T. Miyaza-ki, T. Hyakudome, T. Sawa et al., “Newly-developed devices for the two types of underwatervehicles,” in Proc. MTS/IEEE OCEANS (OCEANS’07), Aberdeen, UK, Jun. 2007, pp. 1–6.→ pages 47[154] A. Bowen, D. Yoerger, C. Taylor, R. McCabe, J. Howland, D. Gomez-Ibanez, J. Kinsey,M. Heintz, G. McDonald, D. Peters et al., “The nereus hybrid underwater robotic vehicle,”Underwater Technology, vol. 28, pp. 79–89, Jul. 2009. → pages 47[155] A. Bowen, D. Yoerger, C. Taylor, R. McCabe, J. Howland, D. Gomez-Ibanez, J. Kinsey,M. Heintz, G. McDonald, D. Peters, J. Bailey, E. Bors, T. Shank, L. Whitcomb, S. Martin,S. Webster, M. Jakuba, B. Fletcher, C. Young, J. Buescher, P. Fryer, and S. Hulme, “Fieldtrials of the nereus hybrid underwater robotic vehicle in the challenger deep of the marianatrench,” in Proc. MTS/IEEE OCEANS (OCEANS’09), Biloxi, MS, Oct 2009, pp. 1–10. →pages 47[156] I. Vasilescu, P. Varshavskaya, K. Kotay, and D. Rus, “Autonomous modular optical under-water robot (amour) design, prototype and feasibility study,” in Proc. IEEE InternationalConference on Robotics and Automation (ICRA’05), Barcelona, Spain, Apr. 2005, pp. 1603–1609. → pages 47, 53[157] P. Corke, C. Detweiler, M. Dunbabin, M. Hamilton, D. Rus, and I. Vasilescu, “Experimentswith underwater robot localization and tracking,” in Proc. IEEE International Conference onRobotics and Automation (ICRA’07), Roma, Italy, Apr. 2007, pp. 4556–4561. → pages 47[158] I. Vasilescu, C. Detweiler, M. Doniec, D. Gurdan, S. Sosnowski, J. Stumpf, and D. Rus,“Amour v: A hovering energy efficient underwater robot capable of dynamic payloads,” TheInternational Journal of Robotics Research, vol. 29, pp. 547–570, Apr. 2010. → pages 47[159] M. Dunbabin, P. Corke, I. Vasilescu, and D. Rus, “Experiments with cooperative control ofunderwater robots,” The International Journal of Robotics Research, vol. 28, pp. 815–833,Jun. 2009. → pages 47[160] M. Dunbabin, J. Roberts, K. Usher, G. Winstanley, and P. Corke, “A hybrid auv designfor shallow water reef navigation,” in Proc. IEEE International Conference on Robotics andAutomation (ICRA’05), Barcelona, Spain, Apr. 2005, pp. 2105–2110. → pages 4896Bibliography[161] M. Doniec, I. Vasilescu, C. Detweiler, and D. Rus, “Complete se 3 underwater robot controlwith arbitrary thruster configurations,” in Proc. IEEE International Conference on Roboticsand Automation (ICRA’10), Anchorage, AK, May 2010, pp. 5295–5301. → pages 48[162] B. Tian, F. Zhang, and X. Tan, “Design and development of an led-based optical communica-tion system for autonomous underwater robots,” in Proc. IEEE/ASME International Confer-ence on Advanced Intelligent Mechatronics (AIM’13), Wollongong, Australia, Jul. 2013, pp.1558–1563. → pages 48[163] M. Tabacchiera, S. Betti, and S. Persia, “Underwater optical communications for swarmunmanned vehicle network,” in Proc. Fotonica AEIT Italian Conference on Photonics Tech-nologies, Naples, Italy, May 2014, pp. 1–3. → pages 48, 53[164] I. Vasilescu, K. Kotay, D. Rus, M. Dunbabin, and P. Corke, “Data collection, storage, andretrieval with an underwater sensor network,” in Proc. 3rd International Conference on Em-bedded Networked Sensor Systems, San Diego, CA, Nov. 2005, pp. 154–165. → pages 48, 49,53[165] M. Dunbabin, P. Corke, I. Vasilescu, and D. Rus, “Data muling over underwater wireless sensornetworks using an autonomous underwater vehicle,” in Proc. IEEE International Conferenceon Robotics and Automation (ICRA’06), Orlando, FL, May 2006, pp. 2091–2098. → pages 48[166] I. Vasilescu, C. Detweiler, and D. Rus, “Aquanodes: an underwater sensor network,” in Proc.The second workshop on Underwater networks (WuWNet’07), Montral, Qubec, Canada., Sep.2007, pp. 85–88. → pages 49[167] N. Farr, A. Bowen, J. Ware, C. Pontbriand, and M. Tivey, “An integrated, underwater opti-cal/acoustic communications system,” in Proc. MTS/IEEE OCEANS (OCEANS’10), Sydney,NSW, Australia, May 2010, pp. 1–6. → pages 51[168] N. Farr, J. Ware, C. Pontbriand, and M. Tivey, “Demonstration of wireless data harvestingfrom a subsea node using a ship of opportunity,” in Proc. MTS/IEEE OCEANS (OCEAN-S’13), San Diego, CA, Sep. 2013, pp. 1–5. → pages 51[169] L. J. Johnson, R. J. Green, and M. S. Leeson, “Hybrid underwater optical/acoustic link de-sign,” in Proc. 16th International Conference on Transparent Optical Networks (ICTON’14),Graz, Austria, Jul. 2014, pp. 1–4. → pages 5197Bibliography[170] S. Han, Y. Noh, R. Liang, R. S. Chen, Y.-J. Cheng, and M. Gerla, “Evaluation of underwateroptical-acoustic hybrid network,” China Communications, vol. 11, pp. 49–59, May 2014. →pages 51, 53[171] A. Lin, W. Lu, J. Xu, H. Song, F. Qu, J. Han, X. Gu, and J. Leng, “Underwater wirelessoptical communication using a directly modulated semiconductor laser,” in Proc. MTS/IEEEOCEANS (OCEANS’15), Genova, Italy, May 2015, pp. 1–4. → pages 53[172] F. Yang, J. Cheng, T. Tsiftsis et al., “Free-space optical communication with nonzero boresightpointing errors,” IEEE Transactions on Communications, vol. 62, pp. 713–725, Feb. 2014. →pages 54, 59, 62[173] P. Cheah, D. Fraser, and N. Reid, “Some alternatives to edgeworth,” The Canadian Journalof Statistics, vol. 5, pp. 131–138, Feb. 1993. → pages 55[174] C. Cox and W. Munk, “Measurement of the roughness of the sea surface from photographs ofthe suns glitter,” Journal of the Optical Society of America, vol. 44, pp. 838–850, Nov. 1954.→ pages 55[175] W. Gappmair, S. Hranilovic, and E. Leitgeb, “Ook performance for terrestrial fso links inturbulent atmosphere with pointing errors modeled by hoyt distributions,” IEEE Communi-cations Letters, vol. 15, pp. 875–877, Aug. 2011. → pages 58[176] A. Mora, D. Ganger, G. Wells, J. Zhang, X. Hu, C. Zhou, A. Richa, and C. Youngbull, “Ad-hoc multi-hop underwater optical network for deep ocean monitoring,” in Proc. MTS/IEEEOCEANS (OCEANS’13), San Diego, CA, 2013, pp. 1–5. → pages 79[177] A. Vavoulas, H. Sandalidis, and D. Varoutas, “Underwater optical wireless networks: A k -connectivity analysis,” IEEE Journal of Oceanic Engineering, vol. 39, pp. 801–809, Oct. 2014.→ pages 79[178] I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals, Series, and Products, 7th ed. SanDiego: Academic Press, 2007. → pages 101[179] J. F. Paris, “Nakagami-q (hoyt) distribution function with applications,” Electronics Letters,vol. 45, pp. 210–211, Feb. 2009. → pages 10298Bibliography[180] J. F. Paris and D. Morales-Jimenez, “Outage probability analysis for nakagami-q (hoyt) fad-ing channels under rayleigh interference,” IEEE Transactions on Wireless Communications,vol. 9, pp. 1272–1276, Apr. 2010. → pages 10299Appendix100Appendix AWe convert xa and ya in (5.5) into polar coordinates as xa = r cos θ and ya = sin θ. Then (5.5)can be expressed asPr(r, θ) =r2piσuσcL2exp(−σ2cr2 cos2 θ + σ2ur2 sin2 θ2L2σ2uσ2c). (A.1)Let sin2 θ = 1− cos2 θ and qH = σc/σu, we havePr(r, θ) =r2piqHσ2uL2exp(r2(1− q2H)cos2 θ − r22q2Hσ2uL2). (A.2)Integrating (A.2) with respect to θ, the PDF of r can be expressed asfr(r) =∫ 2pi0r2piqHσ2uL2exp(r2(1− q2H)cos2 θ − r22q2Hσ2uL2)dθ. (A.3)Applying an integral identity [178, Eq. (3.339)] and double-angle formula cos2 θ = cos 2θ+12 to (A.3),we havefr(r) =rqHσ2uL2exp[−(1 + q2H)r24q2Hσ2uL2]I0((1− q2H)r24q2Hσ2uL2). (A.4)101Appendix BIn [179] and [180], the PDF of Hoyt fading channel are expressed asfx(x) =(1 + q2H)xqHΩxexp[−(1 + q2H)2x24q2HΩx]I0((1− q4H)x24q2HΩx). (B.1)Based on (B.1), the closed-form CDF of Hoyt distribution has also been given in [179] and [180] asFx(x) =Q1√1+qH1−qH√1− q4Hx2qH√Ωx,√1−qH1+qH√1− q4Hx2qH√Ωx−Q1√1−qH1+qH√1− q4Hx2qH√Ωx,√1+qH1−qH√1− q4Hx2qH√Ωx .(B.2)We substitute r =√1 + q2Hx in (5.6) and convert (5.6) asfr(r) =1√1 + q2H(1 + q2H)xqHσ2uL2exp[−(1 + q2H)2x24q2Hσ2uL2]I0((1− q4H)r24q2Hσ2uL2)(B.3)which has the similar form of (B.1). Substitute Ωx = σ2uL2 and x = r√1+q2Hin (B.2), we have theCDF of (5.6) as:Fr(r) =1√1 + q2HQ1√1+qH1−qH√1− q4Hr2qH√1 + q2HσuL,√1−qH1+qH√1− q4Hr2qH√1 + q2HσuL−Q1√1−qH1+qH√1− q4Hr2qH√1 + q2HσuL,√1+qH1−qH√1− q4Hr2qH√1 + q2HσuL .(B.4)Thus, outage probability with zero boresight pointing errors isPout,HT =1− 1√1 + q2HQ1√1+qH1−qH√1− q4HP−1r (γth)2qH√1 + q2HσuL,√1−qH1+qH√1− q4HP−1r (γth)2qH√1 + q2HσuL−Q1√1−qH1+qH√1− q4HP−1r (γth)2qH√1 + q2HσuL,√1+qH1−qH√1− q4HP−1r (γth)2qH√1 + q2HσuL .(B.5).102

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0220823/manifest

Comment

Related Items