Cluster Based Application Mapping Strategy for 2D NoC

Application mapping is one among the most important dimensions of Network-on-Chip (NoC). Integer Linear programming (ILP) is one of the static mapping methods, to find optimum communication cost with larger computational time. To address this issue, cluster based mapping using KL algorithm has been introduced which performs poorly at partitioning cut degree. Based on these studies, we propose modified cluster mapping technique using tailor made algorithm for MPEG4 benchmark and Depth First Search (DFS) for PIP benchmark to optimize the communication cost based on ILP method. The performance and efficiency of proposed methodology was verified with the experiments conducted in different benchmarks in NoC. Experimental results show a 3.8% communication cost reduction for MPEG4 and 3.0% communication cost reduction for PIP benchmark.