
Research Article
Deep Q Network for Wiretap Channel Model with Energy Harvesting
@INPROCEEDINGS{10.1007/978-3-030-41114-5_32, author={Zhaohui Li and Weijia Lei}, title={Deep Q Network for Wiretap Channel Model with Energy Harvesting}, proceedings={Communications and Networking. 14th EAI International Conference, ChinaCom 2019, Shanghai, China, November 29 -- December 1, 2019, Proceedings, Part I}, proceedings_a={CHINACOM}, year={2020}, month={2}, keywords={Energy harvesting Deep Q network Online power allocation Secrecy rate}, doi={10.1007/978-3-030-41114-5_32} }
- Zhaohui Li
Weijia Lei
Year: 2020
Deep Q Network for Wiretap Channel Model with Energy Harvesting
CHINACOM
Springer
DOI: 10.1007/978-3-030-41114-5_32
Abstract
An energy harvesting wiretap channel model is considered in which the sender is an energy harvesting node. It is assumed that at each time slot only information about the current state of the sending node is available. In order to find an effective power allocation strategy to maximize secrecy rate, we put forward a deep Q network (DQN) scheme. First, we analyze the constraints of the system and the issue of maximizing the secrecy rate. Next, the power allocation problem is formulated as a Markov Decision Process (MDP) with unknown transition probabilities. In order to solve the continuous state space problem that traditional Q learning algorithms cannot handle, we apply neural networks to approximate the value function. Finally, an online joint resource power allocation algorithm based on DQN is presented. Simulation results show that the proposed algorithm can effectively improve the secrecy rate of the model.