A Survey of Methods and Architectures for Strategy Design for Energy-Efficient Data Offloading in 6G-Enabled Vehicular Edge Computing Networks Using Double Deep Q-Network

Abstract

The rapid evolution of intelligent transportation systems and 6G communication networks has significantly increased the demand for efficient data processing in vehicular environments. Vehicular Edge Computing (VEC) has emerged as a promising paradigm to support latency-sensitive and computation-intensive applications by offloading tasks from vehicles to nearby edge servers. However, designing energy-efficient and adaptive offloading strategies remains a major challenge due to high mobility, dynamic network conditions, and resource constraints. Recent advancements in deep reinforcement learning, particularly Double Deep Q-Network (DDQN), have provided effective solutions to these challenges by enabling intelligent and stable decision-making. DDQN addresses the overestimation problem of traditional DQN and improves convergence stability in dynamic environments. This survey reviews recent methods and architectures for energy-efficient data offloading in 6G-enabled vehicular edge computing networks, focusing on DDQN-based approaches. The study analyses key techniques, including single-agent DRL, multi-agent reinforcement learning, and hybrid optimization frameworks. It also highlights emerging trends such as hierarchical architectures and mobility-aware strategies. The findings reveal that DDQN-based models significantly enhance energy efficiency and reduce latency compared to conventional approaches. Finally, open challenges and future research directions are discussed to guide the development of scalable and intelligent offloading strategies in next-generation vehicular networks.