| In one-to-many links wireless networks, the forward link capacity may become a "bottleneck" operating under heterogeneous traffic and dynamically changing channel conditions. To meet this challenge, we developed an Adaptive Resource Allocation and Management system that sustains throughput while maintaining the Quality of Service (QoS) by employing adaptive admission and rate control in the presence of congestion and severe channel conditions. Specifically, we attempt to maximize statistical multiplexing gain, under varying channel and traffic conditions, by distributing the rate reduction during congestion evenly among services. Thus, providing graceful degradation during overload intervals. Additionally, by using transmit queue monitoring and end-receiver's QoS reports as a closed loop feedback mechanism, we utilize adaptive control of channel (Forward Error Correction: (FEC)) and traffic source (MPEG compression and data transmission) rates. Since measures of congestion, QoS and channel variations involve observations over different time periods, the adaptive control mechanisms are activated on different time scales as a response to these variations. In particular, short-term time scales relate to transmit queue management issues, moderate-term time scales relate to statistical traffic fluctuations and QoS variations due to channel problems, and long-term time scales relate to the maximization of throughput and persistent QoS violations.; This thesis presents the details of an efficient and comprehensive resource allocation algorithm that is developed based on a centralized topology that maximizes the system resources while maintaining the QoS under dynamic traffic and channel conditions. One major contribution of this research is the integration of admission control and adaptive source/FEC rate control in the design of wireless network system. While most of the research present in the literature addresses admission control and rate control independently, we show that coordination of these control mechanisms is crucial in the optimization of the overall system performance and resource management. For measures of performance, we developed a discrete-event simulator that quantifies throughput, frame loss due to congestion during transmission as well as QoS variations due to channel and source rate changes. The results indicate significant throughput and/or quality gains are possible when the source/FEC pairs are adjusted properly in coordination with admission control. |
