DeepADMR: A Deep Learning based Anomaly Detection for MANET Routing
This work addresses safety concerns for tactical networks using DRL-based routing, though it is incremental as it builds on existing DeepCQ+ methods.
The paper tackles the risk of deploying deep reinforcement learning-based MANET routing policies in untested environments by developing DeepADMR, a neural anomaly detector that monitors temporal difference errors in real-time, achieving effective detection in scenarios with channel disruptions, high mobility, and larger network sizes.
We developed DeepADMR, a novel neural anomaly detector for the deep reinforcement learning (DRL)-based DeepCQ+ MANET routing policy. The performance of DRL-based algorithms such as DeepCQ+ is only verified within the trained and tested environments, hence their deployment in the tactical domain induces high risks. DeepADMR monitors unexpected behavior of the DeepCQ+ policy based on the temporal difference errors (TD-errors) in real-time and detects anomaly scenarios with empirical and non-parametric cumulative-sum statistics. The DeepCQ+ design via multi-agent weight-sharing proximal policy optimization (PPO) is slightly modified to enable the real-time estimation of the TD-errors. We report the DeepADMR performance in the presence of channel disruptions, high mobility levels, and network sizes beyond the training environments, which shows its effectiveness.