OPTICAL NETWORK TRAFFIC DETECTIONAL GORITHM BASED ON PRINCIPAL COMPENT ANALYSIS

摘要

The PCA(principal component analysis)can recognize the key business in optical network and arrange the priority. At the same time, it also can divide abnormal space and recognize the abnormalflow or attack in the network flow to filter and alarm in time. Except that, this paper is also specially directing at the abnormal flow with similarity, further put forward the detection method which is based on the global flow abnormal related analysis at the basic of PCA. According to the abnormal attack flow causing the change between the flows correlation, using the PCA extracts the correlation between the potential abnormal parts of multiple flows and make the correlation change degree as detection measure. The experimental result proves the availability of detection measure, which can overcome the difficulty of the low relative amplitude and not easy detection abnormal flow in the backbone network.