A multi-objective mapping strategy for application specific emesh network-on-chip (noc)

This paper proposes a new optimization model for mapping IP cores onto a new Network-on-Chip (NoC) topology, Emesh. Since competition for one port of the router in Emesh is violent, network competition is considered in our model, in addition to energy and communication cost. To solve this optimization model, the authors present a new application specific multi-objective mapping algorithm onto Emesh topology based on the idea of the crossover and mutation of genetic algorithm. The experimental results show that, compared to traditional heuristic genetic algorithm, the proposed algorithm has lower energy consumption, lower communication cost and less network competition.