Malicious Code Detection on Smart Meters -A Side-Channel Based Approach

With the continuous development of smart grids, smart meter, one of the typical embedded equipment, is also becoming more intellectualized. Smart meters nowadays are capable of not only recording power consumption, but also uploading information to the electricity supplier for monitoring and billing, which makes the electric power dispatching and usage more efficient and convenient. However, the enhanced functions and computation power of the smart meter renders it more valuable and vulnerable for the adversary to conduct a cyber attack. One of the most common and important security threats of the smart meter is the malicious code due to its widespread and the easy access for adversaries. Therefore, it is of great criticalness to detect malicious code in the smart meters to safeguard the security of smart terminals. In this paper, we propose a novel approach that detects malicious code in embedded terminals, e.g., smart meters, with a Long Short Term Memory (LSTM) network based on the power consumption of the CPU or MCU of the device. The evaluation results demonstrate that the proposed method can effectively detect the malicious code in smart meters with an accuracy as high as 92%.