Bandwidth/Fault tolerance/Contention Aware Application-Specific NoC Using PSO as a Mapping Generator
Atena Roshan Fekr, Ahmad Khademzadeh, Majid Janidarmian, Vahhab Samadi Bokharaei · 2010
Abstract— This paper presents a novel and efficient method to produce different mappings, based on Particle Swarm Optimization (PSO) evolutionary algorithm. It produces a lot of mappings with different metrics which totally evaluated by a total cost function. Total cost function helps to find the optimal application-specific Network-on-Chip (NoC) based on designer's decisions to how customize and prioritize the impact of three parameters on special mapping. Three mentioned parameters are communication cost, robustness index and contention factor. Communication cost is a common metric in evaluation of different mapping algorithms which have direct impact on power consumption and performance of mapped NoC. Robustness index is used as a criterion for evaluating fault tolerant properties of NoC. Contention Factor is another performance metric, highly affects the latency, throughput and communication energy consumption. The experimental results reveal the power of proposed procedure which is mainly focused in generating different solutions that speared through the explored design space.