Adaptive Task Mapping and Scheduling for Reconfigurable Distributed Embedded Energy Harvesting Systems

摘要

As with the requirements of high-performance and flexible computing, reconfigurable computing systems have become a subject of a great deal of research. Reconfigurable computing systems have to deal with unpredictable events from the environment such as arrival of new tasks, and hardware or software failures, by adapting the task allocation and scheduling, in order to maintain the system feasibility and performance. This paper presents a novel adaptive approach to address the real-time task scheduling issue in reconfigurable multiprocessor embedded systems based on energy harvesting with the consideration of system performance optimization. An energy efficient offline task mapping and scheduling algorithm is proposed to balance the workload in the multiprocessor embedded system. In order to optimize the system performance and to guarantee the system correctness in the presence of reconfiguration scenarios, a distributed control system is built to process a task migration. A cost function to evaluate the energy and time overheads of tasks migration is proposed. Extensive performance evaluations show the effectiveness of the proposed approach in terms of deadline miss ratio, and the energy gain compared with the related state-of-the-art techniques.