Explainable Congestion Control: Opening the Black Box in AI-Driven QoS

While ML-based congestion control and anomaly prediction (as in Zhao et al., 2021; Xu, 2024) achieve impressive results, they often act as black boxes. Network operators are left with little insight into why certain actions were taken or what specifically went wrong. This research direction aims to integrate explainable AI (XAI) techniques into congestion control frameworks, so that every control action (e.g., rate reduction, path switch) is accompanied by an interpretable explanation—such as highlighting which features or anomaly patterns triggered the adaptation. This is particularly novel in networking, where interpretability is seldom prioritized. Such transparency could dramatically ease debugging, compliance, and trust in automated control, especially in mission-critical networks.

References:

1. Predicting Performance Anomalies in Software Systems at Run-time. Guoliang Zhao, Safwat Hassan, Ying Zou, Derek Truong, Toby Corbin (2021). ACM Transactions on Software Engineering and Methodology.
2. Machine Learning Based Traffic Prediction and Congestion Control Algorithms in Software Defined Networks. Yanying Xu (2024). 2024 International Conference on Interactive Intelligent Systems and Techniques (IIST).
3. Congestion Control as a Service: Towards Low Latency Mobile Uploading. Heng Xu, Letian Li, Liang Wang, Fei Song (2024). International Workshop on Quality of Service.