GBMAP: An Evolutionary Approach to Mapping Cores Onto a Mesh-based NoC Architecture
Misagh Tavanpour, Ahmad Khademzadeh, Somayyeh Pourkiani, Mehdi Yaghobi · Journal of Communication and Computer · 2010
In this paper a new approach based on genetic algorithm for mapping cores onto Mesh-based Network-on-Chip (NoC) architecture, Genetic-based Mapping (GBMAP), is discussed. This algorithm's main aim is to reduce the requirement bandwidth of NoC design. In this algorithm, structure of chromosomes is like switches array in Mesh topology. Therefore, we have a two-dimensional structure for chromosomes. Proposed algorithm uses genetic operators based on this two-dimensional structure. Our algorithm is compared in to parts with other algorithms. At First we compare our method with some prominent algorithms for required bandwidth onto real applications (i.e., Video Object Plane Decoder (VOPD) and MPEG-4 decoder). According to simulation results, our algorithm is better than the others in both applications. In second part, we have used NS-2 to simulate GBMAP on VOPD application with network standpoint and measured energy consumption of communications.