Cooperative Radar and Communications Coexistence Using Reinforcement Learning

The growing population of radar and communications systems poses a challenge to spectral coexistence. In this preliminary study, we investigate the utility of reinforcement learning in facilitating cooperative radar and communications coexistence. We also highlight subtleties and challenges associated with employing reinforcement learning in this context. We simulate a small joint radar-communications network whose members communicate while tracking a moving target’s range. A network management system must learn how to distribute allotted bandwidth to jointly maximize communications and radar performance. Considering several constraints and assumptions in the simulation, the agent sensibly chooses actions that accomplish the required performance behavior.