Abstract
The Internet of Things (IoT), owing to its ability to support sustainability in various fields, has recently been considered one of the most important components of the information and communications technology (ICT) for sustainable smart cities. To achieve the required quality of the IoT communication system, performance prediction is necessary; it is beneficial for fault avoidance through the dynamic and continuous adaptation of network behavior, which helps achieve sustainable improvement of IoT communication systems in smart cities. Herein, a deep-learning (DL) model to evaluate and predict the performance of an IoT communication system is proposed. In the proposed model, the number of dynamic neural networks is equal to the number of networks in the IoT system. Each neural network predicts the performance of a given IoT network. Subsequently, the performance of the entire IoT communication system can be predicted from the outputs of individual networks, which will be considered the inputs for another dynamic neural network. The proposed IoT DL model was tested using network simulator (ns-3) to construct a simulated IoT environment. The simulation results verify that the proposed DL model predicted and improved the performance of an entire simulated IoT system highly accurately.