Research on High-Efficiency Routing Protocols for HWSNs Based on Deep Reinforcement Learning
Yu Song,Zhigui Liu,Kunran Li,Xiaoli He,Weizhuo Zhu
DOI: https://doi.org/10.3390/electronics13234746
IF: 2.9
2024-11-30
Electronics
Abstract:In heterogeneous wireless sensor networks (HWSNs), optimizing energy efficiency presents significant challenges due to variations in node energy levels and the complexity of the network environment. This paper introduces an energy efficiency optimization algorithm for HWSNs based on the Deep Q-Network (HDQN). The algorithm aims to address these challenges by selecting the optimal information transmission path. The HDQN leverages energy differences between nodes and real-time environmental data to enhance network efficiency. Its reward function takes into account node distance, remaining energy, and relay node count to balance node participation and minimize overall energy consumption. The Deep Q-Network (DQN) uses the mean squared error for precise reward estimation, and an improved packet header structure supports effective routing decisions. Simulation results show that the HDQN significantly outperforms existing algorithms—EEHCHR, 2L-HMGEAR, NCOGA, DEEC, and SEP—in terms of energy efficiency, network lifetime, and robustness, demonstrating its potential to advance the performance of HWSNs. The research results of the paper provide a theoretical basis for future energy efficiency research in wireless communication and contribute to the study of the new generation of wireless networks.
engineering, electrical & electronic,computer science, information systems,physics, applied