Anomaly Detection and Traffic Shaping under Self-Similar Aggregated Traffic in Optical Switched Networks

Recent work in traffic analysis has shown that modern network produces traffic streams that are self-similar over several time scales from microseconds to minutes. Simulation studies have demonstrated that self-similarity leads to larger queueing delays and higher drop rates than the Markovian Short Range Dependence (SRD) traffic. At the same time, the dramatic expansion of applications on modern network gives rise to a fundamental challenge for network monitoring and security. Therefore, it is critical to reduce the degree of second order scaling for better network performance and detect traffic anomalies efficiently. In this paper, we propose an approach which can capture the traffic anomalies and decrease the degree of Long Range Dependence at the conjunction of the optical packet switching backbone network. In this method, a traffic shaping technique is proposed and a reference model is generated based on the well-behaving traffic for anomaly detection. Further, we apply the compensation bursty parameter for smoothing the deviation error caused by burstiness difference existing in the traffic data sets. The simulation results show that our work can decrease the degree of self-similarity and detect the anomaly-behaving traffic efficiently.