High Performance Wireless Networking: Adaptation of the Networking Stack to Radio Conditions

As systems and protocols have evolved to facilitate fast and reliable communication, wireless networks have deservedly been the focus of research for ubiquitous access to communication infrastructure. The network stack design for wireless access technologies differs from the wire-line case in that the unreliability of communication over the physical medium necessitates adaptation to radio conditions. In this work, we identify key areas in wireless networking and investigate the benefits of cross-layer interactions among the networking layers. The use of TCP for several Internet applications today mandates its adoption as a transport layer protocol in heterogeneous networks comprised of both wire-line and wireless links. However, the congestion control mechanisms of TCP cause the losses over the wireless channel to be perceived as an indication of congestion and the data transfer rate of TCP to be adversely impacted. The solutions that have been proposed over the years to troubleshoot the problem have their share of drawbacks. The use of adaptive link layer techniques in emerging wireless technologies opens up the possibility of a cross-layer approach for TCP performance enhancement. In our work we outline an optimization framework based on the congestion control dynamics of a bulk-transfer TCP flow. By identifying TCP window evolution patterns in the dynamics, we develop a generic optimization framework for throughput enhancement via adaptive link layer measures. We demonstrate the performance increase on application of the optimization approach to EGPRS and IEEE 802.11a networks. We next investigate a cross-layer multi-hop routing approach based on link connectivity assessment in mobile ad-hoc networks. Ad-hoc networking has conventionally been associated with communication in combat fields and sites of disaster, but has now come to include a multitude of applications like inter-vehicle communication. Routing, however,