能量收集通信系统中基于深度Q网络的最大化保密速率功率控制策略

研究由能量收集发射节点、目的节点和窃听节点组成的能量收集通信系统中,以最大化平均保密传输速率为目标的发送功率控制问题.在环境状态信息事先未知,且系统模型中信道系数、电池电量、收集的能量连续取值的场景下,提出一种基于深度Q网络(deep Q network,DQN)的、仅依赖于当前系统状态的在线功率分配算法.将该功率分配...

Power control strategy based on deep Q network for the maximization of secrecy rate in energy harvesting communication systems

This paper studies the transmission power control problem in an energy harvesting communication system composed of energy harvesting transmitting node, destination node and eavesdropping node, aiming at maximizing the average secure transmission rate. In the case that the environmental state information is unknown in advance, the channel coefficients, battery power, and collected energy are continuously taken in the system model, an online power allocation algorithm based on deep Q network that only depends on the current system state is proposed. Firstly, the power allocation problem is modeled as a Markov decision process. Then, the neural network approximate Q value function is used to solve the problem of infinite combinations of system states. Finally, the decision problem is solved through the deep Q network to obtain the power control strategy that only depends on the current channel state and battery state. The simulation results show that compared with the random power selection algorithm, the greedy algorithm and the Q learning algorithm, the proposed algorithm can obtain a higher long-term average secrecy rate.