QUICK NETWORK TOPOLOGY INFERENCE METHOD BASED ON MULTI-PARAMETER COGNITION

摘要

A quick network topology inference method based on multi-parameter cognition, comprising the following steps: firstly perceiving detection message time delay and detection message loss, conducting difference value analysis on a message time delay parameter, by calculating similarity between leaf nodes, judging a relationship between the leaf nodes and speculating to obtain a leaf node topology, and comparing the leaf node topology with an original topology to judge whether the similarity thereof meets an appropriate threshold value requirement or not, if not, feeding back and adjusting a detection message parameter to complete a cognition circle; and if so, analyzing a father node packet loss parameter, calculating a Hamming distance between father nodes as the similarity thereof, inferring a node relationship according to the similarity, and then inferring a network topology structure. The present invention is based on the perception of two parameters, i.e. detection message time delay and packet loss parameters. The feedback and adjustment of the cognition module can effectively improve the accuracy of topology inference; and the network topology based on multi-parameter perception can deal with a data deviation caused by an unstable network, which leads to a more accurate inference result.