Secure trajectory planning against undetectable spoofing attacks

Abstract

This paper studies, for the first time, the trajectory planning problem in adversarial environments, where the objective is to design the trajectory of a robot to reach a desired final state despite the unknown and arbitrary action of an attacker. In particular, we consider a robot moving in a two-dimensional space and equipped with two sensors, namely, a Global Navigation Satellite System (GNSS) sensor and a Radio Signal Strength Indicator (RSSI) sensor. The attacker can arbitrarily spoof the readings of the GNSS sensor and the robot control input so as to maximally deviate its trajectory from the nominal precomputed path. We derive explicit and constructive conditions for the existence of undetectable attacks, through which the attacker deviates the robot trajectory in a stealthy way. Conversely, we characterize the existence of secure trajectories, which guarantee that the robot either moves along the nominal trajectory or that the attack remains detectable. We show that secure trajectories can only exist between a subset of states, and provide a numerical mechanism to compute them. We illustrate our findings through several numerical studies, and discuss that our methods are applicable to different models of robot dynamics, including unicycles. More generally, our results show how control design affects security in systems with nonlinear dynamics.

Secure Route Planning Using Dynamic Games with Stopping States

“…Reference [6] addresses the use of game theoretic methods to compute locally optimal solutions in presence of attacks and known bounds on disturbances applied to GPS-spoofing of a linear dynamical system. Reference [7] The authors are with the Department of Electrical and Computer Engineering at Michigan State University, East Lansing, MI, USA. Emails: baniksan@msu.edu; shaunak@egr.msu.edu determine a secure trajectory for a robot (autonomous vehicle) when moving a source to a destination and characterize conditions under which attack remains undetected.…”

“…Theorem 3.1: Consider the dynamic game with stopping states modeled using equation (7), ∀k ∈ {1, 2, . .…”

Reinforcement Learning for Linear Quadratic Control is Vulnerable Under Cost Manipulation

“…The incorporation of learning into control, however, enlarges the attack surface of the underlying system and creates opportunities for the adversaries. Adversarial parties can launch attacks such as Denial-of-Service attacks (DoS) [7], [8], False Data Injection (FDI) attacks [9], [10], spoofing attacks [11], [12] on the communication channels, and data poisoning attacks [13], [14] on the existing dataset to mislead the agent and sabotage the underlying system. Such attacks if not dealt with properly can lead to a catastrophe to the underlying system.…”

Stealthy multiplicative attacks against cyber‐physical systems: A gap metric approach

“…17 Recently, the concepts of zero-dynamic attacks and undetectable attacks have been extended to uncertain cyber-physical systems 18 and applied to the secure trajectory planning in adversarial environments. 19 Compared with strictly stealthy attacks, generalized stealthy attacks are more practical due to the resource limitations of attackers. The Kullback-Leibler divergence (KLD), measuring the distance between two probability distributions, is a widely used attack stealthiness index independent of detection methods.…”