Stochastic Geometric Analysis of Cellular Networks Enhanced with D2D and M2M Communication

Cellular networks have to undergo a complete transformation to meet the formidable capacity demands from the ever-increasing number of smart devices. A close look at the requested traffic profile of the devices reveals a deeper underlying challenge for 5G wireless networks. This is because, there is no "one-size fits all" solution and different categories of devices have contrasting requirements. For example, smart phones expect anytime, anywhere connectivity and because of the data hungry applications, they also expect a certain quality of experience. On the other hand, the massive amount of other smart things deployed in the near future will primarily require to stay connected with only a small payload to transmit/receive. Thus, the network operators have an arduous task ahead to design ultra-flexible networks that can easily accommodate a large number of devices with unique requirements and specifications. Device-to-device (D2D) communication has been recently proposed as a promising solution to enhance the capacity of cellular networks by enabling direct communication between user equipments (UEs) located in close proximity without the intervention of the base station (BS). In this thesis, we borrow tools from stochastic geometry to analyze the gains in throughput achieved by offloading UEs to communicate via D2D. We study distance-based and and content-based mode selection strategies. In the distance based mode selection, it is assumed that the D2D pair already exists and D2D mode is selected if the distance separation between them is below a certain threshold. On the other hand, in content based mode selection, the D2D pairs are created subject to requested content availability. We also study how the concept of D2D communication could be extended to establish D2D connections between UEs and the machine-type-devices (MTDs) located in their close proximity to aggregate M2M data. This could potentially reduce the burden of massive access of MTDs on the BS. We employ a novel Poisson hard sphere model for the association between UEs and the MTDs. We quantify the number of MTDs from which a UE can successfully aggregate data without compromising its quality of service (QoS) requirements. Finally, another contribution of this dissertation is comprehensive statistical modeling of the coverage of M2M networks operating in the same spectrum as the cellular networks. The MTDs employ cognition to satisfy the strict QoS constraints of the primary cellular networks. We consider that the MTDs are energy limited as they harvest energy from the sun. The transmit power of the MTDs is therefore shown to play a critical role to efficiently utilize the harvested energy and also maximize the spectrum access opportunities.