| During the real-time system design, power management and reliability requirement are two important aspects. In our research, we focus on designing real-time scheduling algorithms to save energy consumption and to balance the reliability of the underlying real-time systems simultaneously.;First of all, power management issue of real-time systems will be addressed. With the size of modern processors getting smaller and smaller, leakage power (static power consumption) becomes a big component in total power consumption. Secondly, it has been shown in the literature that the reliability of the systems is significantly reduced when slowing down the processor. As a preliminary study, we try to find execution speeds beyond any fixed reliable speed for a periodic task set that can minimize energy consumption. Then in order to meet higher reliability requirement, we adopt recovery job during scheduling to guarantee system reliability. We also integrate leakage control method, shared-recovery technique and slack-reclaim technique flexibly to improve energy consumption. Several algorithms have been proposed in the thesis. The simulation results show that, compared with previous work, our algorithms can reach better results in both power management and reliability aspects. |
