Abnormal pattern detection based on visualization
Aoxing Jin, Dae-Dong Hwang, Gye-young Kim, Hoon Chang, Sangjun Lee, Hyung-Il Choi · 2010
Malicious traffics on the internet are difficult to detect among massive network traffic flow. In most existing method about network attack visualization systems, normally, the attacks cannot be automatically detected by the system. In the paper, a new network traffic visualization based on artificial neural network is proposed. The proposed method is capable of detecting attack traffic pattern more easily. In the proposed method, image is firstly made using the source IP, source port, destination IP and destination port, and then patterns are detected by Hough transform. Pattern features are evaluated by artificial neural network, through which abnormal patterns are classified. The performance of the proposed method proved through experiment results.