A multiobjective approach for real time task assignment problem in heterogeneous multiprocessors

Abstract

Effective assignment of real-time tasks in heterogeneous multi-processor systems to achieve high performance is said to be an NP-hard problem. This paper addresses the problem of real-time task assignment in heterogeneous multiprocessor systems with the goal of maximizing the number of task assigned and decreasing the energy consumption. A heuristic-based Multi-objective Hybrid Max-Min Ant Colony Optimization algorithm (MOHMMAS) on the heterogeneous multiprocessor system is proposed to analyze the tradeoffs between resource utilization of all assigned tasks and cumulative energy consumption. Also, we have constructed pareto fronts to illustrate different task allocations, which can cause a heterogeneous multiprocessor system to consume significantly different amounts of energy. The proposed algorithm has been implemented and evaluated using randomly generated problem instances.It was found that the proposed algorithm outperforms the Multi-objective ACO (MO-ACO) in terms of number of the tasks assigned and cumulative energy consumption of all assigned tasks.