A Secure Data Aggregation Scheme Enabling Abnormal Smart Meters Traceback for Smart Grid

Abstract

In order to prevent smart meter data from being stolen by attackers during transmission, it is common practice to securely aggregate the data and report it to the power company. Although the existing aggregation scheme can protect users' electricity consumption privacy, it cannot distinguish the data that has been attacked by false data injection (FDI), meaning it is difficult to trace and exclude abnormal data sources. To solve this problem, the study proposes a smart meter data aggregation scheme that can trace abnormal nodes. The aggregation center (AC) divides the smart meter (SM) into multiple groups, and the SM in the same group detect the abnormal behavior with each other by calculating whether the Hellinger distance of the power consumption of two adjacent timespan of their counterparts exceeds the set threshold, then feedback to the AC. Through multiple “grouping-detection” iterations, AC locates the groups which contains abnormal SMs. Then AC excludes the abnormal nodes and calculates the normal SMs’ power consumption aggregate value in the group by employing EC-EIGamal homomorphic encryption. Experimental results show that the detection accuracy is 73.3%∼100% under multiple FDI attacks, and attacked SMs can be effectively traced and excluded.