Detecting Deception Attacks on Autonomous Vehicles via Linear Time-Varying Dynamic Watermarking

Abstract

Cyber-physical systems (CPS) such as autonomous vehicles rely on both on-board sensors and external communications to estimate their state. Unfortunately, these communications render the system vulnerable to cyber-attacks. While many attack detection methods have begun to address these concerns, they are limited to linear time-invariant (LTI) systems. Though LTI system models provide accurate approximations for CPS such as autonomous vehicles at constant speed and turning radii, they are inaccurate for more complex motions such as lane changes, turns, and changes in velocity. Since these more complex motions are more suitably described by linear time-varying (LTV) system models rather than LTI models, Dynamic Watermarking, which adds a private excitation to the input signal to validate measurements, has recently been extended to LTV systems. However, this extension does not allow for LTV systems that require several steps before the effect of a given control input can be seen in the measurement signal. Additionally, there is no consideration for the timevarying effects of auto-correlation. Furthermore, a proof of concept was only provided using simulations of a simplified model.This paper relaxes the requirement for inputs to be visible in a single step and constructs an auto-correlation normalizing factor to remove the effects of auto-correlation. In addition, Dynamic Watermarking is applied to a high-fidelity vehicle model in CarSim ™ and a 1/10 scale autonomous rover to further reinforce the proof of concept for realistic systems. In each case, the vehicle follows a predefined path with time-varying velocity and turning radii. A replay attack, which replays previously recorded measurements, is shown to be detectable using LTV Dynamic Watermarking in a quick and repeatable manner.

Machine learning attacks and defenses for vehicular cyber physical systems

“…This approach is considered a more intelligent adversary in cybersecurity. The results from [29] demonstrate that dynamic water marking is effective and with dynamic watermarking, replay attacks on vehicular CPS are quicky detectable.…”

A Dynamic Watermarking Algorithm for Finite Markov Decision Problems

“…While the dynamic watermarking has been studied for linear time invariant system with Gaussian disturbance, it has also been analyzed in the system with arbitrary non-Gaussian noise [36], linear time-varying systems [37], [38], and nonlinear systems [34], [39]. For example, [40] considered the linear system case when the covariance matrix of the measurement noise is unknown or slowly varying over the time.…”

Resilient Control of Platooning Networked Robotic Systems via Dynamic Watermarking

“…Originally, dynamic watermarking was derived for single input single output (SISO) linear time-invariant (LTI) systems [42], but extensions to multiple input multiple output (MIMO) LTI systems [43], [44], networked LTI systems [35], switching LTI systems [45], linear time-varying (LTV) systems [46], [47], and simple non-linear systems have also been derived [48], [49]. Furthermore, detecting attacks on connected vehicle platoons using dynamic watermarking has also been previously studied [35], [49].…”

“…While the linearization of non-linear systems, generally does not result in Gaussian distributed noise, this assumption allows us to derive our proposed method using a statistical basis. Furthermore, this assumption has not hindered the efficacy of LTV dynamic watermarking in non-networked settings [47].…”