MAT4PM: Machine Learning-Guided Adaptive Threshold Control for P4-based Monitoring in SDNs
Henghua Zhang, Jue Chen, Jue Chen, Haidong Peng, Junru Chen, Junru Chen · 2025
This paper presents MAT4PM, a P4-based proactive monitoring framework designed for Software-Defined Networking (SDN). To the best of our knowledge, this is the first monitoring framework to jointly leverage programmable data plane capabilities for event-driven data collection and control plane intelligence for real-time threshold optimization. The system architecture comprises a lightweight P4-based monitoring module deployed at the switch, a machine learning inference engine operating at the controller, and a P4Runtime feedback channel for real-time threshold updates. The framework leverages traffic features to predict optimal monitoring thresholds, which are subsequently synchronized to the data plane. A composite cost function is introduced, jointly considering monitoring error and communication overhead, to guide the model toward a balanced trade-off between accuracy and efficiency. Experimental evaluation on BMv2 software switches demonstrates that, compared to static threshold strategies, MAT4PM reduces monitoring error to 7.0% and achieves a 5.6% reduction in overall cost, while maintaining sub-millisecond inference latency and minimal resource consumption. These results indicate strong practical viability and scalability for real-world SDN environments.