Abstract: The sharp increase of data center network traffic leads to frequent abnormal traffic attacks, which seriously threatens the user data security. Therefore, a data center network abnormal traffic detection algorithm based on SDN( Software Defined Networking) technology is proposed. The data flow transmission process is constructed according to the SDN technology network framework and the time and frequency set method, and then the data flow characteristics are extracted using fuzzy C-means clustering, quadruple, BP(Back Propagation) neural network and other algorithms. The traffic feature subspace is established using the principal component analysis algorithm, and the matrix method is used to project to the subspace. Finally, the set threshold and projection period data vector are used to judge whether there is abnormal traffic in the data center network. The experimental results show that the proposed algorithm is an simple and ensures the accuracy of abnormal traffic detection results, and effectively maintains the stability and security of the data center network.

Key words: software defined networking ( SDN) technology;  , data center network;  , abnormal traffic;  , traffic feature extraction;  , abnormal traffic detection