This paper considers three inter-related adversarial inference problems involving cognitive radars. We first discuss inverse tracking of the radar to estimate the adversary's estimate of us based on the radar's actions and calibrate the radar's sensing accuracy. Second, using revealed preference from microeconomics, we formulate a non-parametric test to identify if the cognitive radar is a constrained utility maximizer with signal processing constraints. We consider two radar functionalities, namely, beam allocation and waveform design, with respect to which the cognitive radar is assumed to maximize its utility and construct a set-valued estimator for the radar's utility function. Finally, we discuss how to engineer interference at the physical layer level to confuse the radar which forces it to change its transmit waveform. The levels of abstraction range from smart interference design based on Wiener filters (at the pulse/waveform level), inverse Kalman filters at the tracking level and revealed preferences for identifying utility maximization at the systems level.

We study how a cognitive radar can mask (hide) its cognitive ability from an adversarial jamming device. Specifically, if the radar optimally adapts its waveform based on adversarial target maneuvers (probes), how should the radar choose its waveform parameters (response) so that its utility function cannot be recovered by the adversary. This paper abstracts the radar's cognition masking problem in terms of the spectra (eigenvalues) of the state and observation noise covariance matrices, and embeds the algebraic Riccati equation into an economics-based utility maximization setup. Given an observed sequence of radar responses, the adversary tests for utility maximization behavior of the radar and estimates its utility function that rationalizes the radar's responses. In turn, the radar deliberately chooses sub-optimal responses so that its utility function almost fails the utility maximization test, and hence, its cognitive ability is masked from the adversary. We illustrate the performance of our cognition masking scheme via simple numerical examples. Our approach in this paper is based on revealed preference theory in microeconomics for identifying rationality.

No abstract

Inverse reinforcement learning (IRL) deals with estimating an agent's utility function from its actions. In this paper, we consider how an agent can hide its strategy and mitigate an adversarial IRL attack; we call this inverse IRL (I-IRL). How should the decision maker choose its response to ensure a poor reconstruction of its strategy by an adversary performing IRL to estimate the agent's strategy? This paper comprises four results: First, we present an adversarial IRL algorithm that estimates the agent's strategy while controlling the agent's utility function. Our second result for I-IRL result spoofs the IRL algorithm used by the adversary. Our I-IRL results are based on revealed preference theory in microeconomics. The key idea is for the agent to deliberately choose sub-optimal responses that sufficiently masks its true strategy. Third, we give a sample complexity result for our main I-IRL result when the agent has noisy estimates of the adversary specified utility function. Finally, we illustrate our I-IRL scheme in a radar problem where a meta-cognitive radar is trying to mitigate an adversarial target.

This paper establishes the equivalence between Bayesian revealed preference (Caplin and Dean, 2015) and classical revealed preference with non-linear budget constraints (Forges and Minelli, 2009). Classical revealed preference tests for utility maximization given known budget constraints. Bayesian revealed preference tests for costly information acquisition given a utility function. Our main result shows that the key theorem in Caplin and Dean (2015) on Bayesian revealed preference is equivalent to Afriat-type feasibility inequalities (Afriat, 1967) for general (non-linear) budget sets. Our second result exploits this equivalence of classical and Bayesian revealed preference to construct a monotone convex information acquisition cost from decision maker's decisions in Bayesian revealed preference.