Using power side-channel to implement anomaly-based intrusion detection on smart grid terminals

Security threats of smart grid terminals continue to increase. In order to realize non-invasive intrusion detection of power distribution terminal devices in smart grid, we propose a novel anomaly-based intrusion detection approach that use power side-channel. The approach monitors programs running in the terminals by collecting and analyzing the power consumption data of terminal devices. We run three abnormal programs and one normal program in the distribution terminal units (DTUs), and then extract feature values of the power consumption data and lastly conducted anomaly-based intrusion detection through LSTM neural network. Our evaluation results demonstrate that a false positive rate is less than 2% and an accuracy rate is more than 90%.