The optimal degree of parallel fault diagnosis for photonic switching network

• Main Authors: Hung-Chang Lin, 林宏璋
• Other Authors: I-Shyan Hwang
• Format: Others
• Language: en_US
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/62097947712427953476

碩士 === 元智大學 === 資訊工程研究所 === 88 === Due to the highly development of computer networks, the connection between computers is an important issue. The photonic network, which transmits data via optical signal, is the fastest technology and provides large amount of bandwidth. A topology, Multistage Interconnection Networks (MIN), which has been widely used in high speed networks, such as ATM and SONET, could provide parallel transmission and programmable data path. There are some well-known approaches for diagnosing so-called stuck-at-x faults in the electronic MIN. But the research in diagnosing photonic switching networks is, in contrast to its electronic counterpart, still in its infancy. This paper considers the Dilated Optical Multistage Interconnection Networks (DOMIN), which the basic component of MIN is a 2×2 directional coupler. Due to the aging of the switching element, some undirected signal will cause the output terminal, and called crosstalk. This motivates us to investigate and design an efficient diagnosing algorithm of automatic test generation which achieves to reduce the number of tests required by overlapping the tests with computation, or/and by paralleling the computations for different characteristics of photonic switching networks. In this paper, we derive algorithms and mathematical modules to find the optimal degree of parallelism of faults diagnosis. Taking the advantages of the properties of disjoint faults, the speed of diagnosis can be speeded up considerably because of the optimal degree of parallel fault diagnosis may growing exponentially. To reduce the time of diagnosis we address the methods to find the maximum number of disjoint faults and examine these disjoint faulty outputs parallelly. Instead of requiring up to 4MN tests as a native approach, we propose a two phases diagnosis algorithm that reduces it to 4N. Furthermore, in this paper, we introduce the concept of disjoint faults, which can reduce the number of test to O(N5/6). Moreover, we also develop an algorithm using tabulation method to find optimal disjoint faults cluster(s) which can speed up the diagnosing procedures for both detection and location in the DOMINs.