Scheduling real-time tasks on battery-operated embedded systems
Jian Lin · 2009
The problems of considering energy usage in real-time system scheduling and assignment are discussed in this dissertation. Issues covered include energy-efficient scheduling for multiple feasible interval jobs on a single processor [44], and assigning frame real-time tasks on multiprocessor systems with rechargeable battery [45]. Both problems are NP-Hard, therefore requiring efficient methods to solve them. For the first problem of multiple feasible interval jobs scheduling, a Simulated Annealing (SA) approach and an on-line greedy heuristic are used to save the energy consumption during execution. In the second problem, the recharge rate of the battery becomes a constraint, and we develop four techniques as solutions, namely Minimum Schedule Length (MSL), Min-min Schedule Length (MmSL), Genetic Algorithm (GA), and Ant Colony Optimization (ACO). The effectiveness of the approaches or techniques for each problem is also shown by experimental results, respectively.