Intelligent Energy‐Aware Routing in Wireless Sensor Networks: A Survey of Deep Reinforcement Learning Approaches

ABSTRACT Wireless sensor networks (WSNs) are increasingly required to operate efficiently under stringent energy and scalability constraints. Conventional routing and clustering protocols often face limitations in large‐scale and dynamically changing network environments, particularly when traffic patterns and topology vary over time. Deep reinforcement learning (DRL) has emerged as a promising paradigm for intelligent routing optimization. It integrates deep neural network‐based feature representation with adaptive decision‐making mechanisms. This survey provides a comprehensive review of DRL‐based energy‐efficient routing approaches in WSNs, covering prominent learning architectures such as deep Q‐networks (DQNs), double DQN (DDQN), deep deterministic policy gradient (DDPG), proximal policy optimization (PPO), and actor–critic frameworks. Existing methods are systematically categorized based on their learning paradigms, optimization objectives, and network design assumptions. A comparative analysis with classical routing protocols is presented in terms of energy consumption, network lifetime, convergence behavior, and scalability. In addition, a statistical assessment of recent literature highlights key research trends in DRL‐based energy‐efficient routing that provides insight into the growing adoption of specific learning models and application domains. In closing, this survey identifies open challenges and outlines future directions for developing sustainable, adaptive, and energy‐aware routing frameworks for next‐generation IoT sensor networks.