Adaptive Physical Access Detection against Ambient Temperature Variation

With the rapid development of the new power system dominated by renewable energy, distributed energy re-source (DER) has been widely deployed. However, as DERs are geographically dispersed and lack of strict border supervision, adversaries can eavesdrop or inject false commands via physically attaching devices into the DER. This physical access attack is difficult to detect by traffic-based intrusion detection methods due to its reticence. The physical-characteristics-based method is promising for detecting this attack. However, its applicabil-ity in complex scenarios is limited. Especially, the change of ambient temperature will lead to the drift of devices' physical characteristics, thus reducing the accuracy rate. In this paper, we clarify that the temperature and physical access attack will both transform the impedance of underlying communication networks, which further changes the transmission signal's voltage. To distinguish the attack from temperature variation, we propose an adaptive physical access detection method based on sliding-window cumulative sum (CUSUM) algorithm. First, we extract the time-domain feature of the signals by segments. Then we use CUSUM to cumulate feature deviations at every time window. When the cumulative value exceeds the preset threshold, it indicates possible physical access attacks. Simulation results demonstrate that our proposed method could effectively detect physical access attacks under different ambient temperatures.