Real-Time Anomaly Detection of Network Traffic Based on CNN

Network traffic anomaly detection mainly detects and analyzes abnormal by extracting the statistical features of network traffic. It is necessary to fully understand concept symmetry in mitigation. However, original information on easily lost, adjustment dynamic configuration becomes gradually complicated. To solve this problem, we designed realized a new online system based software defined networks. The uses convolutional neural directly extract flow for analysis, which can realize real- time packet extraction detection. utilizes SDN flexibly adapt changes network, allowing zero-configuration system. filter used automatically implement mitigation strategies achieve real-time experimental results show that proposed method more accurate warn manager security measures be taken, demonstrates effectively detect problems improve performance edge clustering