One Protocol to Rule Them All: Wireless Network-on-Chip using Deep Reinforcement Learning
summary
Neural-MCP builds adaptive medium-access protocols for Wireless Network-on-Chip using deep reinforcement learning to learn traffic structure, correlations, and timing; the system generates and deploys new MAC policies and evaluates them against four baselines on many-core NoC workloads. Results show faster adaptation and measurable system wins: lower latency, higher throughput, and 37%–275% reduction in execution time for highly dynamic traffic. The paper details the DRL model, training/online decision loop, simulation/emulation setup, and which workloads and metrics produce the biggest gains.
AI-written summary · about
chapters (6)
- 0:00Introduction and motivation
- 1:30Wireless NoC with deep RL
- 4:50Protocol challenges and solutions
- 7:10Model training and simulation
- 9:32Deployment and real-world impact
- 12:10Future directions and conclusion
related talks
17mADR-X: ANN-Assisted Wireless Link Rate Adaptation for Compute-Constrained Embedded...
16mCloud-LoRa: Enabling Cloud Radio Access LoRa Networks Using Reinforcement Learning...
01h 01mMachine Learning, Networks, and IoT
14mBrain-on-Switch: Towards Advanced Intelligent Network Data Plane via NN-Driven Traffic...
32mMan or Machine?: Developing a "Turing Test" for Radio Intelligence
14mRe-architecting Traffic Analysis with Neural Network Interface Cards
17mMutant: Learning Congestion Control from Existing Protocols via Online Reinforcement...
20mKeynote: ML for Systems and Chip Design