PHYSICAL-LAYER AWARE CONTROL AND OPTIMIZATION IN STOCHASTIC WIRELESS NETWORKS by

Under the common theme of physical-layer aware optimization and design in stochasticwireless networks, this dissertation can be broadly organized into the following two parts:In Part I, based on optimal stopping theory, a framework for channel-aware distributedscheduling is developed for exploiting rich diversities at the physical-layer and mediumaccess control (MAC) layer; In Part II, a stochastic network utility maximization (NUM)approach is taken to study random access and flow control.Specifically, Part I is centered around devising distributed opportunistic scheduling(DOS) for throughput maximization in wireless ad-hoc networks based on random access. Insuch networks, DOS involves a process of joint channel probing and distributed scheduling.The desired tradeoff, between the throughput gain from better channel conditions and thecost for further channel probing, boils down to judiciously choosing the optimal stoppingrule for channel probing and the rate threshold. Using optimal stopping theory, it is shownthat the optimal DOS is a pure threshold policy, where the rate threshold can be obtainedby solving a fixed point equation. Along a more practical avenue, an opportunistic multi-channel MAC protocol is devised based on the IEEE 802.11 standards, to exploit channelvariations across multiple channels.Part II focuses on cross-layer rate control using stochastic network utility maximiza-tion. First, a two-phase utility maximization approach is taken to study fair MAC design,and stochastic approximation is used to investigate the MAC design using persistent con-tention. Then, joint optimization of flow control and random access is investigated in depthfor end-to-end QoS provisioning. Since the implementation of distributed NUM algorithmshinges heavily on information feedback through message passing among network elements,