基于多目标免疫算法的NoC映射优化  

No C Mapping Optimization Based on Multi-objective Immune Algorithm

作　　者：吕兴胜[1] 李光顺[1] 吴俊华[1] 

机构地区：[1]曲阜师范大学计算机科学学院,山东日照276826

出　　处：《计算机工程》2015年第4期316-321,共6页Computer Engineering

基　　金：国家自然科学基金资助项目(61373027);山东省优秀中青年科学家奖励基金资助项目(BS2009DX024;BS2010DX013);山东省自然科学基金资助项目(ZR2012FM023);山东省高校科技计划基金资助项目(J09LG34)

摘　　要：片上网络映射算法对系统的功耗、可靠性等性能有重大影响。引入新的抗体初始化算子和抗体变异算子,提出一种多目标映射免疫算法,以降低系统功耗,提高系统可靠性,避免产生额外的资源开销。算法中新的抗体初始化算子利用贪心算法产生初始抗体,新的抗体变异算子通过交换IP核位置减小通信距离,对解进行优化,从而降低由变异随机性产生的退化风险。根据网络的动态特性,提出一种新的功耗模型,使得功耗计算更准确。仿真结果表明,该算法能够有效降低功耗,提高可靠性。Network-on-Chip（ No C） mapping algorithm has significant impact on system power consumption,reliability and other performances. A new M ulti-objective M apping Immune Algorithm based on Energy and Reliability（ M M IAER）is proposed to decrease pow er consumption and increase system reliability w ithout additional overhead added. The M M IAER is constructed by designing a new antibody initialization operator and a new mutation operator based on basic immune algorithm. The new antibody initialization operator of M M IAER generates the initial antibodies by greedy algorithm,and therefore,the quality of the initial solutions is improved. The new antibody mutation operator reduces the communication distance and optimizes the solutions through exchanging positions of Intellectual Property（ IP） cores,and therefore,the new antibody mutation operator decreases the degradation risk due to the randomness. With the dynamic character of the netw ork,a new pow er consumption model is proposed to make the calculation of pow er consumption more accurately. Simulation results indicate that the M M IAER can decrease the pow er consumption and increase the system reliability effectively.

关 键 词：片上网络 片上系统 多目标映射 免疫算法 低功耗 可靠性 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]