An efficient resource scheduling mechanism in LoRaWAN environment using coati optimal Q-reinforcement learning

Abstract

It is estimated that there will be over two dozen billion Internet of Things (IoT) connections in the future as the number of connected IoT devices grows rapidly. Due to characteristics like low power consumption and extensive coverage, low-power wide area networks (LPWANs) have become particularly relevant for the new paradigm. Long range wide area network (LoRaWAN) is one of the most alluring technological advances in these networks. Although it is one of the most developed LPWAN platforms, there are still unresolved issues, such as capacity limitations. Hence, this research introduces a novel resource scheduling technique for the LoRAWAN network using deep reinforcement learning. Here, the information on the LoRaWAN nodes is learned by the reinforcement technique, and the knowledge is utilized to allocate resources to improve the packet delivery ratio (PDR) performance through a proposed coati optimal Q-reinforcement learning (CO_QRL) model. Here, Q-reinforcement learning is utilized to learn the information about nodes, and the coati optimization algorithm (COA) helps to choose the optimal action for enhancing the reward. In the proposed scheduling algorithm, the weighted sum of successfully received packets is treated as a reward, and the server allocates resources to maximize this Q-reward. The evaluation of the proposed method based on PDR, packet success ratio (PSR), packet collision rate (PCR), time, delay, and energy accomplished the values of 0.917, 0.759, 0.253, 85, 0.029, 7.89, and 10.08, respectively.