Spatial modeling and performance analysis for the large-scale 5G wireless networks

Abstract

The fifth generation (5G) wireless networks are on the horizon with the missions of delivering the ubiquitous connectivity fabric for massive Internet-of-Things (IoT) and providing high-quality communication and computation services with ultra-low latency, ultra-high data rate and capacity. This will usher in a new era of mobile user experiences. 5G is envisioned to incorporate a wide range of cutting-edge technologies, including wirelessly powered backscatter communications (WP-BackCom) for massive IoT, mobile edge computing (MEC), and millimeter-wave (mmWave) communications, to achieve a new height of performance and efficiency. These enabling technologies call for novel networks spatial modeling and comprehensive performance analysis to understand performance bottlenecks and development solutions. Therefore, in this dissertation, we have proposed the spatial models for the large-scale WP-BackCom, MEC, urban and air-to-everything (A2X) mmWave networks, and built on the models to investigate their network performance by jointly applying stochastic geometry, queueing, and convex optimization theories. As a result, new fundamental design insights have been obtained for deploying practical 5G wireless networks.

First, to power massive number of small computing or sensing devices, we have proposed a novel network architecture that enables large-scale passive IoT deployment by seamlessly integrating wireless power transfer (WPT) and backscatter communication, called the WP-BackCom network. The power beacons (PBs) that are stations dedicated for WPT are deployed for wirelessly powering dense backscatter sensors and each sensor transmits data by reflecting and modulating the carrier signals sent by PBs. Applying stochastic geometry, the WP-BackCom network is modeled as a Poisson cluster process and its coverage probability and transmission capacity are derived, analyzed, and optimized as functions of backscatter parameters.

Second, to investigate the performance bottleneck of MEC network, we first modeled a large-scale MEC network which can be decomposed as a radio-access-network (RAN) cascaded with a computer-server-network (CSN). Then, we have analyzed its network-constrained latency performance, namely communication-latency and computation-latency, under the constraints of RAN-connectivity and CSN-stability. The scaling laws of latency are derived with respect to various network parameters including the densities of mobiles and access points, RAN communication resources, and CSN computation capabilities. The results provide design guidelines for MEC network provisioning and planning to avoid either RAN or CSN becoming the performance bottleneck.

Third, we have studied the blockage effect on the connectivity of a mmWave network in a Manhattan-type urban region by modeling buildings as a random lattice process. Different lower bounds on the connectivity probability are derived as functions of buildings' sizes, density, as well as base-station density and transmission range. Both the single-tier and K-tier heterogeneous networks are considered to derive guidelines for practical mmWave network deployment and performance evaluation.

Last, we have proposed an analytical framework to define and characterize the connectivity for an aerial access point (AAP) of an A2X mmWave network. The blockage area for an arbitrary building is calculated and minimized by optimizing the AAP's altitude. A lower bound on the connectivity probability is obtained as a function of AAP's altitude and parameters of users and buildings, yielding design insights on A2X mmWave network deployment.