Surveillance Performance of Digital Radars in Non-uniform Target Behavior
Jameson D. Morgan, Brian D. Rigling · 2023
Surveillance radar systems generally do not use knowledge of the operating environment to adapt operation. In situations with non-uniform target behavior, failure to consider such operating environment knowledge will lead to sub-optimal performance. In this work, using a surveillance-inspired task, we demonstrate the feasibility and benefit of adapting digital radar operation using operating environment knowledge. The core idea is to use a simulated environment that has been configured with a priori knowledge about target behavior in order to generate training data that will allow a neural network to accurately predict which area of the environment to illuminate. The simulated environment allows the radar system to explore the most rewarding solution to different possible states and the neural network provides generalization, enabling solutions on previously unseen states. The results obtained show improvement over the state of practice and demonstrate the feasibility of the proposed solution approach.