A Preemption Control Technique for System Energy Minimization of Weakly Hard Real-Time Systems

This paper aims to present a general scheduling algorithm which offers lesser energy consumption for weakly hard real time systems modeled with (m, k) constraint. The weakly hard real time system consists of a DVS processor (frequency dependent) and peripheral devices (frequency independent). The energy minimization is done in two phases. In the first phase we suggest new static partitioning strategy that partitions the jobs into mandatory and optional followed by a greedy based speed assignment at the task level. Theorem is being derived to show the feasibility condition of weakly hard real-time system with modified partitioning strategy. The second phase proposes a preemption control technique that can effectively reduce the preemption impact by delaying the higher priority jobs. The simulation results and examples demonstrate that our approach can effectively reduce the overall system energy consumption (especially for systems with higher utilizations) while guaranteeing the (m, k) requirement at the same time.