Deep Learning and Optimization Approaches in Strategy Design for Energy-Efficient Data Offloading in 6G-Enabled Vehicular Edge Computing Networks Using Double Deep Q-Network: A Review

Abstract

The rapid evolution of sixth-generation (6G) communication networks has significantly accelerated the development of intelligent vehicular systems, necessitating efficient computation offloading strategies in Vehicular Edge Computing (VEC) environments. With the exponential growth of data-intensive and latency-sensitive applications such as autonomous driving, augmented reality, and intelligent transportation systems, traditional offloading mechanisms fail to meet stringent energy and delay constraints. Recently, deep learning-based optimization approaches, particularly Double Deep Q-Networks (DDQN), have emerged as promising solutions for dynamic and adaptive decision-making in complex vehicular environments. This review explores the integration of deep reinforcement learning and optimization techniques for energy-efficient data offloading in 6G-enabled VEC systems. It systematically analyses recent advancements between 2020 and 2023, focusing on algorithm design, system architectures, and optimization objectives such as latency minimization, energy efficiency, and resource utilization. Furthermore, this paper highlights the advantages of DDQN over conventional methods in addressing overestimation bias and improving convergence stability. Key challenges, including scalability, dynamic mobility, and security concerns, are also discussed. The study provides a comprehensive comparative analysis of existing approaches and identifies future research directions toward sustainable and intelligent vehicular edge computing systems.