Poster: Learning to Shine – Optimizing Glossy at Runtime with Reinforcement Learning
Valentin Poirot, Olaf Landsiedel · Chalmers Research (Chalmers University of Technology) · 2019
Glossy is a dissemination protocol that allows a node to propagate information to the entire network through constructive interference. We present GLossAI, a new artificial intelligence-based version of Glossy. We use reinforcement learning to determine and update Glossy’s parameters at runtime. Each node individually learns the best strategy to minimize energy consumption while maintaining high reliability. Furthermore, nodes can dynamically adapt their parameters to follow the dynamics of the medium.