AI Cognitive Radio: Dynamic Spectrum Access Simulator
Watch a reinforcement-learning agent share radio spectrum with unpredictable primary users: a live 3D waterfall of channel occupancy, Q-learning channel selection, and real-time throughput/collision readouts.
An AI secondary-user agent shares a bank of radio channels with unpredictable primary-user traffic, learning which channels tend to be free using tabular Q-learning and ε-greedy exploration. The 3D spectrum waterfall shows the most recent slots scrolling back in time across every channel — red blocks mark primary-user occupancy, green marks a successful AI transmission, orange marks a collision — while live readouts track throughput, collision rate and the agent's currently best-learned channel as the traffic load, exploration rate and learning rate are adjusted.
An AI secondary-user agent learns via Q-learning which radio channels are free from unpredictable primary-user traffic, shown as a live 3D spectrum waterfall with real-time throughput and collision readouts.
3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install