A Low-Energy Mapping Optimization Algorithm for Fat-tree Based Network-on-Chip

This paper copes with the problem of mapping for fat-tree based network-on-chip.Considering the difference of the link length between routers in different layers of the network, an energy-aware mapping optimization model is proposed.Differential evolution algorithm based on Predatory Search strategy (PSDE) is designed to solve the mapping model.Differential evolution algorithm is easy to fall into local optimum.To make up for this deficiency, PSDE is proposed to strengthen the concentrated local search ability of predatory search strategy.The experimental results show that, compared with genetic algorithm and simulated annealing algorithm, the PSDE can obtain lower energy mapping results and save running time.