Smart Traffic Estimation Model based on Deep Conditional Variational Reinforcement learning with Hiking Optimization in 6G

Massive data processing and transmission have been made by the rapid growth of Internet of Things (IoT) applications and mobile broadband networks. The extreme traffic congestion has negatively impacted the fast expanding networks and businesses. Predicting network traffic has turn into a crucial tool to improve network resource allocation and ensure seamless communications. To deliver AI-native solutions in 6G vision, an intelligent edge-native architecture that offers a high-speed traffic prediction based on DL with reinforcement learning (TMSDLRL) is proposed. The model predicts network traffic using deep conditional variational reinforcement learning (DCVRL) in which Q-value approximation is preformed using conditional variational auto-encoder. Further, the optimal policy of DCVRL is selected using hiking optimization algorithm (HikOA). The simulation of TMS-DLRL is carried out using python platform. The experimental results display that the TMS-DLRL model accurately predicts the network traffic against actual observations. The minimum RMSE value obtained by TMS-DLRL model is about 0.057%, which is better than existing methods.