增广立方体网络的t/k诊断度研究  

Research on t/k-diagnosability of augmented cubes network

作　　者：陈昭蓉[1] 陈秒江 梁家荣[1] 

机构地区：[1]广西大学计算机与电子信息学院,南宁530004

出　　处：《计算机应用研究》2017年第12期3647-3650,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(61364002;61462006);广西自然科学基金资助项目(2014GXNSFAA118391)

摘　　要：在多处理器系统,传统的可诊断算法在处理大规模故障集时有明显的局限性。针对增广立方体诊断度提升的问题,提出了一种可诊断的变形算法,即t/k可诊断算法,在该算法下,可明显提高增广立方体的诊断度。算法核心思想是,在故障节点个数不大于t的情况下,允许故障集中出现k个非故障节点,从而在牺牲少数非故障节点的情况下,达到提高网络诊断度的目的。最终证明,增广立方体在t/k诊断算法下的诊断度明显优于其传统诊断度和条件诊断度。The classical t-diagnosability approach has its limitation when dealing with large fault sets in large multiprocessor systems. Aiming at the problem of increasing the diagnosability of augmented cubes,this paper proposed an alternative approach to systems,called t/k-diagnosability. In this new measure,it was obviously to increase the diagnosability of augmented cubes of multiprocessor systems. The algorithm's pivotal idea was allowing the number of free faulty unites to be diagnosis incorrectly,where the number of faulty unites did not exceed t. It could enhance diagnosability of augmented cubes while lost of a few free faulty unites. The results show that the t/k-diagnosability of augmented cubes,which is bigger than its ordinary t-diagnosability and conditional diagnosability.

关 键 词：增广立方体 故障诊断 t/k-可诊断 系统级诊断 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]