Data Integrity Attack on Consensus-based Load Shedding Algorithm for Power Systems

摘要

The paper presents a novel data integrity attack on consensus-based load shedding algorithm. In particular, we show that by sending out elaborately falsified information during the consensus iterations, attackers could manipulate the system operating point to achieve selfish goals, e.g. loads remain being served under contingencies while shedding other loads. More importantly, still maintain the system stability. This data integrity attack has three major features: 1) no additional information about system topology or other devices are required to launch the attack; 2) the attacking effect is accumulative which enables attackers to complete the attacks in either single or multiple iterations; 3) attackers could fulfill their goal using the local agent alone, no other agents need to be compromised during the attack. By revealing such potential risks, this paper conveys the message that besides the efforts of designing novel consensus-based applications, it is equally important to protect distributed smart grid applications from possible malicious cyber-attacks. The potential impact of the data integrity attacks is illustrated on simulation examples.