Adaptive resource allocation to enhance the quality of service in wireless networks

Most of today's applications and network protocols were optimized to work in a wireline environment where low bit-error-rates, high speed and high reliability are customary. Today's users are increasingly demanding mobility while accessing information on the new generation networks. To provide the needed ubiquity, network solutions must include wireline and wireless links. In a wireless environment, communication links have variable high bit-error-rates, intermittent connectivity and limited bandwidth. Due to these characteristics, most of the applications and network protocols perform poorly. This dissertation presents solutions to enhance the quality of service (QoS) in wireless networks. Specifically, the dissertation develops adaptive solutions at the link, transport and application layers to enhance the QoS in a wireless environment. At the link layer, two solutions are developed to dynamically allocate resources based on the QoS requirements of the application and the state of the wireless channel. The first solution optimally selects an error-correcting code to encode each packet prior to its transmission. This scheme enhances the bandwidth utilization without degrading the QoS. The second solution selects an optimal target signal to interference and noise ratio for each user's transmission. This scheme increases the system capacity and reduces the user's power without impacting the QoS. At the transport layer, the dissertation describes an adaptive scheme to improve the end-to-end throughput of TCP connections passing over a wireless channel. Specifically, an agent is introduced at the base station to perform local retransmissions over the wireless channel of all corrupted packets. In addition, when a loss of a wireless connection is detected, the agent forces the TCP sender into persist mode. At the application layer, the dissertation proposes an adaptive solution to increase the throughput of HTTP servers. The solution adjusts the lifetime of an HTTP session to the server's load based on the history of client behavior. The scheme increases the throughput as seen by the client without jeopardizing the availability of server's resources. The end result is a set of adaptive solutions which combat the key wireless challenges and enhance the QoS received by a mobile user.