On the design of next-generation wireless networks

Wireless technology has gained great attention from both academia and industry in the past decades. It is believed to be the essential part to achieve the ultimate goal of communications: communication happens whenever, wherever, and to whomever. With the proliferation of different wireless technologies and their vast deployment, we will soon enter the age of ubiquitous computing. Compared to its wireline counterpart, wireless network faces more technical challenges. The wireless channel is less reliable and more variable than fibers and cables. Wireless devices are usually mobile and energy-constraint. Wireless transceiver has limited coverage area. All these characteristics make the resource allocation and algorithm design in wireless networks challenging tasks. This dissertation investigates both issues in different types of next-generation wireless networks.This dissertation first explores the problem of providing a traffic-oblivious routing and scheduling for wireless mesh networks. We present an optimization model which generates a static routing and scheduling policy with worst-case performance guarantees under a range of traffic conditions. Packet-level simulations demonstrate good average performance for the generated routing and scheduling in addition to its worse-case guarantees.Next, we investigate the problem of identifying the maximum capacity of multichannel multi-radio wireless networks, given the number of available channels. We provide an Integer Linear Program (ILP) framework to compute the maximum capacity. We also discuss how to extend the framework to consider heterogeneous network. Then the framework will output the minimum number of radios necessary for each node to fully utilize the available channels, which can be used to identify the bottleneck in the topology, and also indicates the “goodness” of the topology.We also study the integration gain of heterogenous WiFi and WiMAX networks. We propose a generic framework to identify the integration gain, the gain comes from better utilization of resources rather than the addition of resources. For the performance objective, we focus on the max-min throughput fairness, and also briefly cover the proportional fairness. We propose approximation algorithms to achieve different objectives and compute the integration gain based on the framework. The proposed framework can be used to evaluate the potential benefit of combining different heterogeneous wireless networks. The approximation algorithms can be directly used to perform intelligent user association in the integrated network.Last, we study the unique characteristics of cognitive radio networks. Based on our observations, we propose distributed algorithms to perform channel allocation among secondary users. We consider both throughput and fairness, and study their tradeoff.This dissertation makes important contributions by introducing novel models, algorithms and architectures that will help improve the network analysis, planning, and deployment of next-generation wireless networks.