Protocols, Architectures and Applications of Multi-Hop Wireless NetworksPublic Deposited
Multi-hop wireless networks hold promise for increasing network capacity, lowering power requirements, and improving coverage over traditional cellular networks. However, their widespread adoption is hampered by challenges that include: 1) unreliable and complex routing protocols due to the transient nature of wireless nodes, 2) difficulties in guaranteeing quality of service to real-time applications, and 3) inefficiencies of medium access control in a dynamic wireless networking environment. In this dissertation, we tackle these problems in some specific multi-hop network environments. First, we design low-complexity self-organizing and routing protocols for a large multi-hop network. We develop cell cluster-based routing trees and associated novel hierarchical routing and addressing approaches. Near optimum routing is achieved with a complexity of essentially O(1), versus O(n3) for conventional optimal routing, and performance is validated by simulation. We then design a real-time vehicle guidance solution with a dense wireless sensor network utilizing our routing approach. It features a communication subsystem and a vehicle routing subsystem; the former gathers real-time vehicle data and distributes guidance information while the latter processes vehicle data and makes guidance decisions. Very small communication bandwidth is shown to be required to deliver true real-time vehicle guidance. Second, we develop a virtual circuit communication protocol that supports connection-oriented applications such as voice and streaming video that exploits load balancing multiple-routing trees to minimize connection blocking rate. Lower bounds on blocking rate are analyzed and used to evaluate the performance, and simulations verify the results. Finally, we consider a medium access control problem in a two-hop wireless network where the base station, fixed relays, and mobiles within a cell share a reservation-based TDMA channel. We present a scheduling solution that seeks to identify the optimum uplink path from each mobile while allocating time slots in such a manner that queueing delay at the relays is essentially negligible. We quantify conditions whereby relays offer performance advantages and show analytically that throughput gain due to optimum relay increases dramatically as path loss becomes more severe. Simulations also show that optimum relay yields a significant throughput advantage compared with a one-hop approach and un-optimized two-hop relay approach.