An efficient network detection approach based on correlation relationship

Network detection in large distributed systems is a challenging task which requires accurate detect whether there exist fault, there are several ways of network detection; Active probing is widely used in this domain recently. Exist works in active probing mainly use entropy-criterion; however, they have limitations to large-scale net. In this paper, we present an approach, which can resist network noise, using correlation relationship to solve the problem of network detection, we decrease the size of detection probe-set, prevent shortcoming that a node of system may be detected excessively. We simulate our algorithm and MI, the result shows that our algorithm chooses less probes when obtained the same fault detection probability, our algorithm is an efficient network detect approach that suit for complexity network.