Pareto based Multi-objective Mapping IP Cores onto NoC Architectures

This paper presents a multi-objective exploration approach to map the IP cores onto the network on chip (NoC). The NoC mapping problem is treated as a two conflicting objective optimization problem of minimizing the average hop and achieving the thermal balance. Thus a set of Pareto-optimal mapping, instead of a single mapping, is an efficient solution. The proposed approach uses a novel representation and the multi-objective genetic algorithm capable of finding multiple Pareto-optimal solutions simultaneously for different performance requirement. Experimental results show the approach is highly efficient and accurate