Core Concepts
Design a radar controller using Fisher information to mask the sensing plan from adversaries.
Abstract
The content discusses the application of Fisher Information Approach in designing a Markov Decision Process (MDP) based radar controller to conceal its sensing plan. It explores how perturbations in the MDP's transition kernel and operation cost can reduce Fisher Information, hindering adversary's estimation accuracy.
Introduction to Markov Decision Processes (MDPs) for radar optimization.
Analytical derivation of Fisher Information Matrix (FIM) determinant for MDP-based controllers.
Application of FIM as a design constraint to mask sensing plans.
Perturbation strategies in MDPs to reduce adversary's estimation accuracy.
Comparison with traditional radar strategies and related works.
Stats
The radar controller purposefully minimizes the Fisher information of its emissions so that an adversary cannot identify the controller’s model parameters accurately.
Numerical results show that minor perturbations can reduce the Fisher Information of the emissions, amplifying variability in policy and transition kernel estimation errors.
Quotes
"The primary objective is to decrease the determinant of the FIM concerning the adversary’s estimation of the sensing plan."
"Lower curvature of likelihood function correlates with higher variance in parameter estimate for adversaries."