Multi-objective mapping for network-on-chip based on bio-inspired optimization algorithms

The network-on-chip (NoC) mapping confronts a balance between energy consumption, area and the performance of system-on-chip (SoC) in order to achieve optimum performance. In this article, we want to solve a two-objective optimization problem in terms of NoC energy consumption and communication latency by optimizing distribution of link load. Traditional heuristic approaches such as genetic algorithm are likely to trap into local optimal solutions. We present a new application specific multi-objective mapping algorithm based on bio-inspired optimization algorithms combining membrane computing and conventional genetic algorithms. Then this approach is applied to two real applications with different numbers of cores. Experimental results demonstrate that the represented mapping is a feasible way to obtain better NoC architecture in term of energy consumption, latency and traffic balance than mapping based on genetic algorithm.