Abstract: Recent routing algorithms show the potential of deep reinforcement learning (DRL) for optimal routing in mobile ad-hoc networks (MANETs). However, existing DRL-based methods still have ...