Adversarial Signal Processing and Inverse Cognitive Sensing
Abstract
This project is motivated by our long term vision of understanding the interaction between adversarial statistical signal processing and inverse reinforcement learning in cognitive sensing. Cognitive sensing refers to a reconfigurable sensor that dynamically adapts its sensing mechanism using stochastic control to optimize its sensing resources. The fundamental research associated with this project involves the interplay of deep ideas in Bayesian estimation of random measures, revealed preferences (inverse reinforcement learning) with nonlinear budgets and stochastic dynamics, and tracking stochastically evolving equilibria (with stochastic averaging analysis resulting in differential inclusions). These fundamental ideas will be considered in the context of cognitive (adaptive) radars as the main application. Objectives The objectives fall under four inter-related themes: Task 1. Inverse Bayesian Filtering and Parameter Estimation in Adversarial Systems Task 2. Optimal Probing to Estimate EnemyÕs Capabilities Task 3. Detecting utility maximization behavior and estimating the utility function of the adversarial decision system Task 4. Tracking time evolving equilibria when the adversaryÕs utility evolves over time. The problems considered in this project transcend classical statistical signal processing (which deals with extracting signals from noisy measurements) to address the deeper issue of how to infer the strategy of adversarial sensors. The generative models, adversarial inference algorithms and associated mathematical analysis will lead to useful developments in the design and understanding of how sophisticated sensing systems operate.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jun 25, 2021
- Source ID
- W911NF2110093
Entities
People
- Vikram Krishnamurthy
Organizations
- Army Contracting Command
- Cornell University
- United States Army