时间：2013年7月12日(星期五)上午10:00
地点：仓山校区成功楼603室
主讲：西安电子科技大学 朱强副教授
主办：数学与计算机科学学院
专家简介：朱强，西安电子科技大学数学系副教授，硕士生导师。2005年获中国科学技术大学应用数学专业博士学位。2007年晋升为副教授。2009年8月至2010年12月在西安电子科技大学计算机系从事博士后研究。2011年9月至2013年3月在美国西弗吉尼亚大学访问，进行合作研究。研究方向包括并行与分布式系统，多处理器互连网络，算法设计与分析，图论等。担任IEEE TC、IEEE TPDS、JPDC、 Parallel Computing、Info. Sci.等多个国际著名学术期刊的审稿人。在IEEE TC、Info. Sci.、J.Supercomputing等国外学术期刊上发表论文多篇。目前主持国家自然科学基金项目1项。
报告摘要：In large multiprocessor systems which may include hundreds of thousands of processors, processor failures may occur. Fault diagnosis is thus important to the design and Maintenance of such multiprocessor systems. System level diagnosis is an important approach for the fault diagnosis of multiprocessor systems. PMC model is the most well-known and widely studied model for systems level diagnosis. The main contributions of this paper includes:
1. Propose a new parameter called restricted diagnosability to better measure the diagnosis capability of interconnection networks and determine the restricted diagnosability of hypercubes under the PMC model.
2. Present an efficient diagnosis algorithm based on a graph-coloring model. Given a syndrome, the algorithm can polynomially locate all the 1-step diagnosable faulty processors. If not all faulty processors can be 1-step diagnosed, by transferring the problem of determining suspicious faulty processors of a multiprocessor system into finding the maximal independence sets of its contracted diagnosis graph, the suspicious faulty sets are presented together with a weight which can measure their occurring probability. The algorithm is shown to be correct, complete, not based on any conjecture and can be applied to the fault identification for any multiprocessor system. Simulation Results also show that the algorithm is quite efficient.