Energy efficient scheduling for real-time systems

The goal of this dissertation is to extend the state of the art in real-time scheduling algorithms to achieve energy efficiency. Currently, Pfair scheduling is one of the few scheduling frameworks which can optimally schedule a periodic real-time taskset on a multiprocessor platform. Despite the theoretical optimality, there exist large concerns about efficiency and applicability of Pfair scheduling in practical situations. This dissertation studies and proposes solutions to such efficiency and applicability concerns. This dissertation also explores temperature aware energy management in the domain of real-time scheduling. The thesis of this dissertation is: the implementation efficiency of Pfair scheduling algorithms can be improved. Further, temperature awareness of a real-time system can be improved while considering variation of task execution times to reduce energy consumption.;This thesis is established through research in a number of directions. First, we explore the applicability of Dynamic Voltage and Frequency Scaling (DVFS) feature of the underlying platform, within Pfair scheduled systems. We propose techniques to reduce energy consumption in Pfair scheduling by integrating DVFS into the optimal Pfair scheduling algorithm. The integration was achieved by modifying the original Pfair scheduling algorithm to dynamically vary the weight of a task. Our experimental evaluation with synthetic and real benchmarks shows up to 66% savings in energy consumption compared to the basic Pfair scheduling algorithm. Next, we explore the problem of quantum size selection in Pfair scheduled systems so that runtime overheads are minimized. We study the system overhead as a function of quantum size and present quotient search (QS)---a quantum size selection heuristic to reduce the overall scheduling overhead of Pfair scheduling. Our results show that there is a notable difference in the runtime overhead (3% on the average), between QS and other quantum size selection strategies. We also propose a hardware design for a central Pfair scheduler core in a multiprocessor system to minimize the overheads and energy consumption of Pfair scheduling. Three different implementation schemes for the Pfair scheduling algorithm were considered: replicated software scheduler running on each processor, single software scheduler running on a dedicated processor and the proposed hardware scheduler. Experimental evaluation shows that the hardware scheduler outperforms the other two implementation schemes by orders of magnitude in terms of scheduling delay and energy consumption. Finally, we propose a temperature aware energy management scheme for tasks with varying execution times. The proposed scheme, TA-DVS, reduces temperature constraint violations by 18.9% on the average, compared to existing schemes without adversely affecting energy consumption.