| Summary: | Attack detection algorithms of large scale mobile internet network need the prior information of attack behaviors or supervised learning to attack behaviors, so these algorithms is not real time and applicable, a real-time attacks blind detection and analysis algorithm of mobile internet network is proposed to handle that problems. Firstly, the largest eigenvalues for all time frames are extracted, the attack behaviors of each time frame are detected by analysis combined largest eigenvalues with model order. Then, the types of detections are analyzed by eigenvalues analysis technique, and the variations details of the eigenvalues are identified. Lastly, similarity analysis schema are designed to analyze the detail information, such as port count and time. Simulation results based on the real experiment and public network traffic dataset show that the proposed algorithm realizes a good attack detection accuracy.
|
|---|