Unslotted CSMA/CA mechanism with reinforcement learning of Wi-SUN MAC layer

Abstract

Wi-SUN, a protocol necessary for realizing a smart city, uses IEEE 802.15.4 Unslotted CSMA/CA for MAC Layer to build a large-scale network. However, as the scale increases, the contention becomes more severe, and the performance of Unslotted CSMA/CA deteriorates rapidly. Also, there is a problem of wasting resources because there is no scheduling. Therefore, in this paper, the performance of the MAC layer is improved by 20% by applying reinforcement learning to each node and learning to select the optimal backoff delay value.