On-line leakage-aware energy minimization scheduling for hard real-time systems

As the semiconductor technology proceeds into the deep sub-micron era, leakage and its dependency with the temperature become critical in dealing with the power/energy minimization problem. In this paper, we develop an analytical method to estimate energy consumption on-line with the leakage/temperature dependency taken into consideration. Based on this method, we develop an on-line scheduling algorithm to reduce the overall energy consumption for a hard real-time system scheduled according to the Earliest Deadline First (EDF) policy. Our experimental results show that the proposed energy estimation method can achieve up to 210X speedup compared with an existing approach while still maintaining high accuracy. In addition, with a large number of different test cases, the proposed energy saving scheduling method consistently outperforms two closely related researches in average by 10% and 14% respectively.