Watermark-Based Replay Attack Detection for Unmanned Marine Vehicles

Guangrui Bian, Tieshan Li, Yue Long, Hanqing Yang

TL;DR: Abstract

Abstract: This paper is concerned with the problem of attack detection for un- manned marine vehicles (UMVs) suffering from replay attacks. As a highly au- tomated equipment, UMVs play the pivotal role in maritime operations. How- ever, due to the complexity of UMVs working environments and the openness and vulnerability of networks, they are susceptible to external disturbances, com- munication interference, and network attacks. Among various forms of attacks, replay attacks, due to not requiring knowledge of the targeted system's model by the attacker, are easy to execute. Sourced from normal system operation, the data replayed inherently possesses a certain level of stealthiness. The danger lies in attackers not needing to deeply understand the specific operational patterns and objectives of UMVs, nor having to crack the system monitoring devices of UMVs. Attacks can be launched simply by modifying transmitted information, thereby disrupting the normal operational state of UMVs. Therefore, it is of great significance to detect replay attacks promptly and efficiently. To this end, an at- tack detection scheme based on watermark signals is proposed. Firstly, a Takagi- Sugeno (T-S) fuzzy system model for the UMV are established. Using multiple fuzzy subsystems to approximate the nonlinear UMV system facilitates subse- quent research. Then, encoded and decoded signals based on the watermark sig- nal are constructed, which are added to terminals of the sensors and the state estimators to detect replay attacks. The signals can effectively affect the residual values of the system when it is under attack, thus enabling the detection of replay attacks. Moreover, the method contributes to less system performance loss is demonstrated. Finally, the simulation results verify the effectiveness of this scheme.

Submission Number: 94