| Summary: | 碩士 === 國立成功大學 === 電腦與通信工程研究所 === 97 === For the video delivery services, system capacity is one of important issues because of high resource need, especially during heavier workload. Efficiency of service or resource allocation plays key factor of system's capacity. Coordinating user demands and service supplies is an effective method for efficient resource allocation. This thesis proposed a Coordinated Channel Allocation (COCA) strategy to negotiate with the users who bring higher workload. COCA makes the heavier workload requests wait for a specific duration with access promising after waiting. By delaying partial requests, COCA scheduler shifts the workload to non-peak period. The shift in workload reduces the renege rate during heavier workload,
and improves the utilization during low workload. The Cooperated Scheduler (CoS) comprises four major components: 1) Coordinated Scheduling Engine, 2) Resource Requirement Evaluator, 3) Deadline Assigner, 4) Channel Allocation Predictor. The experiments are provided to evaluate performance of following four algorithms: On-demand, PRC, enhanced On-Demand with COCA strategy, enhanced PRC with COCA strategy. The experiment results show that COCA is effective for decreasing the
renege rate without reducing scheduling optimization.
|
|---|
