Anomaly Extraction in Networks
Mr. Naushad Mujawar, Mr. Sohan Patil, Mr. Amit Kanase, Mr. Ravindra Jagadale, Prof. Gajanan Arsalwad · International Journal of Computer Trends and Technology · 2014
The application detects anomaly in network using techniques like histogram, cloning voting, filtering. To extract anomalous flows, one could build a model describing normal flow characteristics and use the model to identify deviating flows. We can compare flows of packets on network with previous flows, like new flows that were not previously observed or flows with significant increase/decrease in their volume. Identify an anomalous flow that combines and consolidates information from multiple histogram-based anomaly detectors (1) (4) (8). Compared to other possible approaches. Build a histogram based detector that (i) applies histogram cloning(1)(4), i.e., maintains multiple randomized histograms to obtain additional views of network traffic(3); and (ii) uses the Kullback-Leibler (KL) distance to detect anomalies.