Set-based particle swarm optimization for mapping and scheduling tasks on heterogeneous embedded systems

Modern heterogeneous multiprocessor embedded platforms is important for the high volume markets that have strict performance. However, it presents many challenges that need to be addressed in order to be efficiently utilized for multitask applications. Since mapping and scheduling problems for multi processors belong to the classic of NP-Complete problems, common methods used to solve this kind of problem usually fail. In this paper, we present an algorithm based on the meta-heuristic optimization technique, set-based discrete particle swarm optimization (S-PSO), which efficiently solves scheduling and mapping problems on the target platform. This algorithm can simultaneously addressed the mapping and scheduling problems on a complex and heterogeneous MPSoC and it has better performance than other algorithms in dealing with large scale problems. This algorithm also reduces the execution time of the application by exploring various solutions for mapping and scheduling of tasks and communications. We compare our approach with other heuristics, Ant Colony Optimization (ACO), on the performance to reach the optimum value and on the potential to explore the target platform. The results show that our approach performs better than other heuristics.