Multi-Radar Tracking Optimization for Collaborative Combat

TL;DR

This paper introduces two novel reward-based learning methods, including reinforcement learning, for decentralized coordination of collaborative radars, demonstrating their effectiveness in a simulated multi-target tracking scenario.

Contribution

It presents new reward-based learning approaches for radar coordination, simplifying the RL problem while maintaining equivalence, and validates them through simulation.

Findings

RL approaches outperform greedy baseline in tracking accuracy

Radars learn implicit cooperation through the proposed methods

Simulation results confirm effectiveness of the approaches

Abstract

Smart Grids of collaborative netted radars accelerate kill chains through more efficient cross-cueing over centralized command and control. In this paper, we propose two novel reward-based learning approaches to decentralized netted radar coordination based on black-box optimization and Reinforcement Learning (RL). To make the RL approach tractable, we use a simplification of the problem that we proved to be equivalent to the initial formulation. We apply these techniques on a simulation where radars can follow multiple targets at the same time and show they can learn implicit cooperation by comparing them to a greedy baseline.