针对带宽和时延约束的低能耗片上网络(NoC)映射问题,提出了一种自适应的混沌遗传退火映射算法.该算法利用Boltzmann更新机制选择遗传个体,引入自适应混沌方法优化适应度较差个体,采用多邻域的退火策略优化较优个体.实验结果表明,所提算法有效地避免了早熟收敛,提高了算法收敛速度,与标准遗传算法和混沌遗传算法相比,平均节能分别为45%和22.6%,有效地降低了NoC系统通信能耗.
An adaptive chaos genetic annealing algorithm is proposed to solve the mapping problem of low-energy consumption network-on-chip (NoC) subject to the constraints of bandwidth and communication latency. Based on Bohzmann update mechanism, the algorithm makes a selection of genetic individuals. Individuals with poor fitness are optimized by the adaptive chaos method, while the optimum individuals are optimized by the strategy of multi-neighborhood annealing. Experiments suggest that the proposed algorithm is able to avoid premature convergence and increase the convergence speed. Compared with the standard genetic algorithm and the chaos genetic algorithm, 45% and 22.6% energy savings are achieved on average, respectively, thus effectively reducing the energy consumption of NoC system communications.