Phased scheduling of continuous media tasks to improve quality of service

To improve the quality of service, a multimedia system needs to accommodate transient overload with minimum deadline misses of time-critical tasks. It also has to minimize the average response time of non-time-critical tasks for smooth, consistent user interactions. The objective of the paper is to present a phased scheduling technique for periodic, continuous media tasks to minimize both the deadline miss ratio and the average response time. Phased scheduling is achieved by varying the phases or points of time at which independent periodic tasks are scheduled for execution. An algorithm is proposed to identify the optimal phase vector. The discussion focuses on the earliest deadline first algorithm, although it can be extended to other scheduling algorithms. Simulation results show that the phase shifting substantially enhances the quality of service.