Semi Supervised Learning Model Construction for WSN Network Topology Burst Traffic Unstable State Identification
The topology of wireless sensor networks has a high degree of self-organization and multi hop transmission uncertainty.When subjected to interference such as noise and attack behavior,it is easy to cause unstable network topology,reducing the reliability and se-curity of wireless sensor networks.Therefore,based on the construction of a semi supervised learning model,the identification of unstable states of burst traffic in WSN network topology is realized.Based on the AE model,a semi supervised learning model is estab-lished to process the data,the key features of the data are extracted,k-Means clustering algorithm is used to cluster,the low dimensional feature vectors of the key points of the data are obtained,histogram is used to mark the data feature categories,Euclidean distance is used to calculate the processed distance,and the identification of the unstable state of the burst traffic in the wireless sensor network to-pology is completed.The simulation results show that the proposed method identifies more than 90 abnormal traffic and 50 abnormal IP data,with a recognition false alarm rate of less than 10％,and has good recognition performance.
wireless sensor networkidentification of unstable states of sudden traffic flowsemi supervised learning modelhistogramk-Means clustering algorithm
万方数据
维普
