A learning-based MAC for energy efficient wireless sensor networks

Designing energy-efficient communication protocols is one of the main challenges in wireless sensor networks. This work presents an adaptive radio scheduling schema employing a reinforcement learning algorithm for reducing the energy consumption while preserving the other network performances. By means of a decentralized on-line approach, each nodes determines the most beneficial radio schedule by dynamically adapting to its own traffic load and to the neighbors’ communication activities. We compare our approach with other learning-based MAC protocols as well as conventional MAC approaches and show that, under different simulating scenarios and traffic conditions, our protocol achieves better trade-offs in terms of energy consumption, latency and throughput.