Abstract
The Internet of Things (IoT) has developed a well-defined infrastructure due to commercializing novel technologies. IoT networks enable smart devices to compile environmental information and transmit it to demanding users through an IoT gateway. The explosive increase of IoT users and sensors causes network bottlenecks, leading to significant energy depletion in IoT devices. The wireless network is a robust, empirically significant, and IoT layer based on progressive characteristics. The development of energy-efficient routing protocols for learning purposes is critical due to environmental volatility, unpredictability, and randomness in the wireless network’s weight distribution. To achieve this critical need, learning-based routing systems are emerging as potential candidates due to their high degree of flexibility and accuracy. However, routing becomes more challenging in dynamic IoT networks due to the time-varying characteristics of link connections and access status. Hence, modern learning-based routing systems must be capable of adapting in real-time to network changes. This research presents an intelligent fault detection, energy-efficient, quality-of-service routing technique based on reinforcement learning to find the optimum route with the least amount of end-to-end latency. However, the cluster head selection is dependent on residual energy from the cluster nodes that reduce the entire network’s existence. Consequently, it extends the network’s lifetime, overcomes the data transmission’s energy usage, and improves network robustness. The experimental results indicate that network efficiency has been successfully enhanced by fault-tolerance strategies that include highly trusted computing capabilities, thus decreasing the risk of network failure.