A Reinforcement Learning assisted Backoff Algorithm for LoRa networks

The Long Range (LoRa) networks are able to provide both long-range coverage and low power consumption for sensing devices. These features make it one of the most attractive solutions for Internet of Things (IoT) applications. The default channel access mechanism used by the protocol is pure ALOHA, which due to its simplicity and lack of synchronization creates scalability problems when the number of supported end devices increases. The slotted ALOHA channel access mechanism has been proposed as an alternative, which through the synchronization of transmissions achieves the improvement of the efficiency and scalability of LoRa networks. One of the most important operating parameters of slotted ALOHA is the backoff algorithm. In this paper we present a novel backoff algorithm assisted by reinforcement learning mechanism, which regulates the offered load on the network by adjusting the congestion window. Simulation results show that the proposed algorithm improves efficiency and energy consumption for the end devices compared to other backoff algorithms from the literature.