Abstract
This thesis presents a novel wireless scheduling mechanism, called Adaptive Service Curve (ASC). The proposed mechanism increases the flexibility with which network operators can adjust the allocation of resources. We first point out that a good scheduler should exhibit three kinds of flexibility, and then we show the ASC can achieve this goal. First, ASC should be able to differentiate the error resilience requirements due to the impact of location-dependent channel errors in wireless networks. Specifically, ASC can employ link adaptation, enabling a choice to be made between maximizing system throughput and making/wasting more link effort on error-prone channels. Second, the above flexibility is not system-wide. Rather, users can subscribe to different error resilience requirements. ASC utilizes the service curve model that can best meet QoS requirements to provide the third kind of flexibility, and prevents an unacceptable over-allocation of scarce radio resources. Accordingly, a framework is proposed to make the existing service curve model operate effectively in wireless environments. The ASC scheduler combines in a single-framework the three aspects of scheduling, namely, link adaptation, flexibility and a mature traffic characterization model. This design represents a complete solution for wireless resource management.