Relayer-Enabled Retransmission Scheduling in 802.15.4e LLDN -- Exploring a Reinforcement Learning Approach
We consider the scheduling of retransmissions in the low-latency deterministic network (LLDN) extension to the IEEE 802.15.4 standard. We propose a number of retransmission schemes with varying degrees of required changes to the LLDN specification. In particular, we propose a retransmission scheme that uses cooperative relayers and where the best relayer for a source node is learned using a reinforcement-learning method. The method allows for adapting relayer selections in the face of time-varying channels. Our results show that the relayer-based methods achieve a much better reliability over the other methods, both over static (but unknown) and over time-varying channels.