Delay Optimization in LoRaWAN by Employing Adaptive Scheduling Algorithm with Unsupervised Learning

Low Power Wide Area Network (LPWAN) technologies have been exponentially growing because of the tremendous growth of the Internet of Things (IoT) devices across the globe. Several LPWAN technologies have been utilized by the researchers to address certain issues like increased number of collisions, retransmissions, delay, and energy consumption. However, Long Range Wide Area Network (LoRaWAN) is the most suitable and attractive technology in terms of delay optimization, low cost and efficient energy consumption. The main issue which arises in LoRaWAN is because of its high packet drop rate due to collision. The reason behind this packet drop rate is the MAC scheme known as Pure Aloha used by LoRaWAN for the transmission of the frames. Long Range (LoRa) End Devices (EDs) initiate communication with Pure Aloha that leads to a high number of retransmissions. These retransmissions further enhance the delay in LoRa networks. This paper aims to optimize the delay in LoRaWAN by using an Adaptive Scheduling Algorithm (ASA) with an unsupervised probabilistic approach called Gaussian Mixture Model (GMM). By using ASA with GMM, the retransmissions are reduced which optimizes the delay in LoRaWAN. The results show that in our approach, Packet Collision Rate (PCR) is reduced by 39% as compared to conventional LoRaWAN. In addition, the Packet Success Ratio (PSR) is also increased by 39% as compared to the conventional LoRaWAN and Dynamic Priority Scheduling Technique (PST). Further, the delay is optimized by 91% and 79%. This research could be effective for the environments where the critical data of patients need to be sent with optimised retransmissions and minimum delay towards gateways.