Stochastic Geometry Analysis of Energy Harvesting, Edge Computing, and Device-to-Device Communication in Heterogeneous Wireless Networks

Abstract

THE future of wireless networks will evolve through inter-layer innovations to support new-age applications and use cases such as augmented reality, internet-of-everything (IoE), automated unmanned vehicular systems, remote surgery, and many more. Major requirements are enhanced data rate, higher system capacity, ultra-low end-to-end latency, and improved energy efficiency. The co-existence of large-scale machine-type devices with human-centric and content-centric communication will demand a more complex scalable heterogeneous wireless network environment. Wireless networks can be self-sustaining by harvesting energy from ambient radio-frequency (RF) signals. With ever-increasing data demand, caching popular content in the memory of user devices is seen as a promising alternative to offload demand from macro base stations (BS) and reduce backhaul loads. Considering these necessities, the integration of energy harvesting (EH), edge computing, and device-to-device (D2D) communication can be a promising technology for next-generation heterogeneous wireless networks. Further, stochastic geometry-based mathematical modeling has given a major paradigm shift in the design and performance analysis of such networks. In order to evaluate the performance of heterogeneous wireless networks analytically, stochastic geometry’s primary goal is to endow probability distributions on the positions of network nodes. This dissertation proposes stochastic geometry-based approaches and new frameworks for modeling, optimization, and performance analysis of energy harvesting and cache-enabled cognitive D2D communication. The first contribution of this dissertation is the development of a new stochastic geometry-based comprehensive framework for EH-enabled adaptive mode selection policy for cognitive D2D communication underlying cellular networks. In the proposed peer discovery policy, the status of the energy queue of all potential D2D users is proposed to be shared among themselves before communication starts. The spectrum optimization problem for efficient network functioning is formulated and a solution to choose the optimum spectrum access factor is provided as the second contribution of this dissertation. Also, a channel inversion power control policy and a distance-based correlation-aware mode selection policy are proposed for the optimum utilization of resources. The expressions for the temporal correlation of link outages are provided. Incorporating this in the medium access control (MAC) protocol will reduce the number of retransmissions and improves energy and spectrum efficiency. The impact of different traffic patterns on network throughput is also analyzed. The third direction of this dissertation is the integration of edge computing in terms of caching and energy harvesting capabilities into D2D communication to propose an effective and energy-efficient content distribution network (CDN). Here the Markov chain model is used to analyze the transmission probability of a typical D2D user operating in EH mode. The fourth and final direction of this dissertation is the incorporation of an independent marked Poisson point process (imPPP) to probabilistically model the aggregate interference and signal-to-interference-plus-noise ratio (SINR) distribution of the randomly distributed nodes in the spatial domain. The efficacy of the proposed unified analytical framework is evaluated in terms of the overall network coverage probability improvement due to the minimization of congestion in the base station-oriented core network. All the developed D2D framework, simulation results, and the outlined remarks are utilized to provide significant design insights and specifications for the deployment strategies of EH and cache-enabled cognitive D2D communication in heterogeneous wireless networks.