Deep Reinforcement Learning-Based Energy Consumption Optimization for Peer-to-Peer (P2P) Communication in Wireless...

Area of Science:

• Wireless Sensor Networks (WSN)
• Communication Systems
• Artificial Intelligence

Background:

• Wireless sensor networks (WSN) face significant energy consumption challenges due to rapid sensor development.
• Peer-to-peer (P2P) communication is crucial for overcoming WSN energy bottlenecks.
• Spectrum interference and power limitations hinder WSN performance.

Purpose of the Study:

• To propose a deep reinforcement learning-based energy consumption optimization strategy for P2P communication in WSN.
• To enable P2P sensors (PUs) to intelligently manage spectrum sharing with authorized sensors (AUs).
• To reduce interference and conserve energy in WSN environments.

Main Methods:

• Utilized P2P sensors as agents capable of controlling power and selecting resources to mitigate interference.
• Employed a double deep Q network (DDQN) algorithm for agents to learn detailed interference features.
• Simulated the proposed algorithm against traditional methods and a standard deep Q network (DQN).

Main Results:

• The DDQN-based approach demonstrated superior performance compared to DQN and traditional algorithms.
• The proposed method effectively reduced energy consumption for P2P communication in WSN.
• Intelligent resource allocation by agents minimized interference between PUs and AUs.

Conclusions:

• Deep reinforcement learning, specifically DDQN, offers an effective solution for energy optimization in WSN P2P communication.
• The proposed strategy successfully addresses interference and power constraints in WSN.
• This approach enhances the efficiency and longevity of wireless sensor networks.