UBC Theses and Dissertations
Optical Code Division Multiplexing for sub-wavelength switching systems Khattab, Tamer
Optical Code Division Multiplexing (OCDMA) is a method used to enable simultaneous transmission of multiple optical data flows over the same fiber using the same wavelength. In OCDMA, isolation between different data flows is achieved using a set of uncorrelated, or loosely correlated, spreading codes to encode the transmitted signal and decode it at the receiver side. The process of encoding and decoding is performed entirely in the optical domain without the need for optical-to-electrical-to-optical conversion. This increases the granularity of traffic isolation in the optical domain while maintaining higher speed switching because of the all-optical encoding/decoding capability. Although code division multiplexing is a well established technique in wireless transmission where all processing of data and switching are performed electronically, there are many challenges in applying this scheme in the optical domain mainly due to the different characteristics of the medium and the fact that negative-valued signals are not easy to produce. This thesis has three main objectives: to deploy OCDMA as a switching mechanism at the sub-wavelength level in order to increase the granularity of traffic isolation in all-optical core switching, to design new mechanisms that enhance the performance of OCDMA as a multiplexing method over long-haul optical fiber transmissions, and to model the performance of OCDMA based switching and multiplexing mechanisms. All-optical switching at the core of the network provides very high speed switching. However, it suffers from low utilization or lack of quality of service guarantees due to lack of fine granularity traffic isolation. This thesis presents an optical network architecture called Optical Code Labeled Generalized Multi-Protocol Label Switching (OC-GMPLS), which utilizes OCDMA as a switching mechanism in backbone GMPLS networks. OC-GMPLS uses OCDMA as an all-optical labeling space in GMPLS switching in order to achieve finer granularity switching at the all-optical network core. The deployment of OC-GMPLS networks mandates performance modeling to show its advantages and to enable tuning of the new network parameters so that performance can be optimized. In this thesis we present an analytical model for the throughput and switching capacity of OC-GMPLS networks. Using our model, we show how to find optimal operating points for OC-GMPLS networks based on physical layer and network layer parameters. The performance of OC-GMPLS networks depends on the performance of OCDMA transmission, which is affected by the modulation method and the optical spreading codes properties. In order to enhance the performance of OC-GMPLS networks, we take two different approaches. The first approach is based on proposing a modulation mechanism that enhances the communication reliability while maintaining low bit error rate for OCDMA transmissions. Our Chip-Level Modulated Binary Pulse Position Modulation (CLM-BPPM) scheme provides a simple to implement (in the all-optical domain) yet a very powerful physical layer method for sending multiple optical flows using OCDMA while maintaining the Bit Error Rate (BER) due to Multiple Access Interference (MAI) effects between these flows at a low level of about 10⁻¹² for 10 simultaneous users. Our method provides a better capability in terms of clock recovery and user activity detection while achieving error rates in the range of those provided by On-Off Keying (OOK). Performance of OCDMA transmission depends to a great extent on the efficiency of the codes used to perform the multiplexing. In order to tackle this side, we investigate the problem of Optical Orthogonal Code (OOC) design by proposing a method called Rejected Delays Reuse (RDR) for constructing OOCs using an element-by-element based greedy algorithm. We show that our method provides a computationally less complex algorithm for designing OOCs, which makes it more practical. Our analysis and simulation results show that OOCs designed using the RDR greedy method are also higher in multiplexing efficiency than OOCs designed using classical element-by-element constructions. This is because RDR designed OOCs possesses smaller code lengths for the same code cardinality and weight than their counterpart classical element-by-element greedy designed codes.
Item Citations and Data