A New Slack Reclaiming Algorithm for Real-time Systems

Abstract

Real-time applications are ubiquitous in generalpurpose computing environments, while the real-time systems are growing in complexity. Thus in these hybrid real-time systems, schedulers must guarantee that all hard real-time jobs be completed before their deadlines and improve QoS of soft real-time tasks as much as possible. Towards this goal we have proposed a new slack reclaiming algorithm for server-based real-time systems, and have also implemented it in a real time emulator (RTSIM). This algorithm, named HBASH, which enhances the Constant Bandwidth Server (CBS) by slack reclaiming, allocates slack generated from the running process to the task that needs the slack most, and then this selected task will be scheduled immediately. Hence the algorithm is able to make full use of slack and reduce the response time of soft realtime tasks as much as possible. In this paper, we proved that our algorithm does not violate the schedulability of tasks, and we also evaluated the performance of this algorithm. The experimental results demonstrate that HBASH outperforms other slack reclaiming algorithms and improves soft real-time performance significantly.