An effective approach for improving responsiveness of Total Bandwidth Server

As a result of increasing in diversity and complexity, real-time embedded systems have to deal with both periodic and aperiodic tasks. In such systems, it is expected that response time of aperiodic tasks is as short as possible. Total Bandwidth Server (TBS) is one of effective task scheduling algorithms for a mixture of periodic and aperiodic tasks. As a successor to TBS, Adaptive TBS (ATBS) is an algorithm where TBS is enhanced with prediction of execution times. Virtual release advancing (VRA) is an additional technique for TBS to shorten response time of tasks and Enhanced VRA (EVRA) is the extended version of VRA with lightweight complexity. In this paper, combination of ATBS and EVRA is examined to further improve the responsiveness of TBS. Simulation results show that the combination can reduce response times by up to 25% compared with TBS under heavy loads, while maintaining the schedulability and relatively low time complexity of the system.