An Efficient and Adaptive Configuration of IEEE 802.15.4 MAC for Communication Delay Optimisation

The IEEE 802.15.4 media access control (MAC) protocol stands today as one of the largely used protocol in IoT low power networks applications. This protocol allows IoT power low devices such as sensors to access to the wireless communication medium in order to transmit the MAC frames. The performance of IoT low power networks using 802.15.4 MAC relies heavily on the proper configuration of MAC parameters. For this reason, various solutions have been proposed to set the IEEE 802.15.4 MAC parameters in order to reduce the network energy consumption. However due to the dynamic nature of most of IoT low power networks, determining the optimal value for MAC parameters remains a challenge. Traditional solutions for configuring IEEE 802.15.4 MAC parameters are limited to uses cases where the network traffic is highly variable but do not consider highly dynamic IoT low power networks. In this paper we propose an efficient approach for configuring the IEEE 802.15.4 MAC parameters. The proposed solution uses the predictive feature of machine learning algorithms in order to determine the IEEE 802.15.4 MAC parameters according to the network traffic state and the network characteristic, and hence supports the dynamicity of the devices/communication. Simulation results show that our proposal improves of the end-to-end delay compared to the usage of the standard IEEE 802.15.4 MAC.