INTELLIGENT NETWORK SLICING FOR SCALABLE AND RESILIENT 6G VEHICULAR COMMUNICATIONS

Abstract

 The advent of 6G networks provides new possibilities to facilitate autonomous vehicle (AV) ecosystems with the requirement of ultra-reliable, low-latency, and high-throughput communication. The paper suggests a framework of Dynamic QoS-Aware Network Slicing (DQNS) that combines the use of the Long Short-Term Memory (LSTM)-based traffic prediction and Deep Reinforcement Learning (DRL)-based resource allocation to meet the dynamic and heterogeneous needs of various types of services, such as URLLC, eMBB and mMTC slices. The framework uses closed-loop orchestration process in the MEC-core coordination layer to predictive load estimation to support multi-objective optimization to achieve strict compliance with SLA. Results of the simulations indicate that the proposed DQNS technique decreases URLLC latency by more than 51 percent, augments the ratio of packet delivery to 99.3 percent, enhances the eMBB throughput by 29 percent, and efficiency in utilizing resources to 91.6 percent compared to the basic static and reactive slicing techniques. The suggested solution offers an intelligent, AI-powered orchestration approach, which makes it a promising enabler of intelligent transportation systems in the future in 6G-enabled smart mobility networks.