Open Collections

On the performance of connectivity of cellular based IoT network

University of British Columbia

Due to the manifold connected devices future cellular based Internet of Things (IoT) networks require significant attention. Random point process (PP) based networks models are now widely accepted models for designing cellular based IoT networks because of theirs analytical tractability. This thesis focuses on analyzing the performance of cellular networks for massive connectivity of IoT devices using stochastic geometry. In cellular based IoT network, one of the key challenges resides in establishing connection between the devices and the base stations (BSs). An IoT device performs random access channel (RACH) procedure when it needs to establish connection with its intended BS. This thesis specifically focuses on the study of success of RACH access procedure for different BSs' distribution. In particular, we propose a novel generalized approach to calculate connection failure probability of IoT devices in RACH phase of uplink (UL) transmission. The proposed approach uses a calculation of the devices' association probability rather than an approximation of the Voronoi tessellation's cell area distribution. There are some limitations of using Voronoi tessellation based approach. This approach can only be used when the BSs are distributed according to Poisson point process (PPP). The merit of the proposed approach, besides its exact analysis, is that it can be applied for general scenarios of BSs' distributions. To adopt this approach, we derive the void probability for different Poisson cluster processes (PCP), in particular, Matérn cluster process (MCP) and Thomas cluster process (TCP), which is defined as the probability of having no children point of PCP in a given distance. First, we use our proposed mathematical approach to single-tier network. Next, we extend our mathematical framework in multi-tier cellular based IoT network. For both types of networks, we investigate the performance metrics using PPP and PCP distributions for BSs' deployment. We validate our approach with the numerical simulations via MATLAB. We vary different network parameters to see the effect in failure probability which can help network designers to set values of various design parameters. Numerical results demonstrate that our mathematical framework is highly accurate and adaptable for general scenarios.

Text

2019-05-22

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/70245

Electrical Engineering

University of British Columbia

UBCO

http://creativecommons.org/licenses/by-nc-nd/4.0/