Tasks mapping in multi-core based system: hybrid ACO&GA approach

In the recent years, multi-core based distributed systems are rapidly gaining its popularity among consumer electronics, telecommunication, military electronics etc. System level synthesis on such systems is totally NP-complete/NP-hard, and traditional approaches always result in low efficiency or local optimality. In this paper, novel memory models, which emphasize the local memory capacity constraint, memory occupation and variable lifetime, are presented; moreover, we present a novel mapping method. For a given architecture and a set of task graphs, tasks are assigned to different PEs using genetic algorithm (GA), and then these partitioned tasks sets are transformed to a special graph, so that tasks scheduling is solved by a new meta-heuristic namely Cooperative Ant Colony Optimization.