Access security testing for wireless power terminals based on DNN

Abstract

The external wireless terminals of the new power system are susceptible to be attacked through internal network penetration triggered by physical contact. The traditional device security test has little effect on improving the security performance of the access devices and is prone to produce a high false positive rate （FPR）. A wireless access security testing system based on deep neural networks （DNN） is proposed. The system adopts a stacked sparse autoencoder （SSAE） to realize the feature dimensionality reduction of the test dataset and selects the appropriate feature dimensions for training. The selected features are used as the input layer of the DNN to construct highly efficient test cases， as well as to monitor and discover the abnormal states. The experiment results show that the system has an accuracy rate of 90% and can efficiently detect anomalies in the access environment of wireless power terminals.

Keywords

• wireless access
• security testing
• machine learning