quotation:		[Copy]
		[Copy]

This Paper:Browse 2416 Download 48
异构网络中用户关联和功率控制的协同优化
樊雯,陈腾,菅迎宾
0 Fontlarge +\|Default\|Small
(石家庄铁路职业技术学院信息工程系，石家庄 050041)

摘要:

针对正交频分多址(Orthogonal Frequency Division Multiplexing Access,OFDMA)异构网络中用户关联和功率控制协同优化不佳的问题，提出了一种多智能体深度Q学习网络(Deep Qlearning Network,DQN)方法。首先，基于用户关联和功率控制最优化问题，构建了正交频分多址的双层异构网络系统模型，以实现智能决策；其次，根据应用场景和多智能体DQN框架的动作空间，对状态空间和奖励函数进行重构；最后，通过选取具有宏基站(Base Station,BS)和小型BS的两层异构网络，对多智能体DQN算法的性能进行仿真实验。仿真结果表明，相较于传统学习算法，多智能体DQN算法具有更好的收敛性，且能够有效提升用户设备（User Equipment,UE)的服务质量与能效，并可获得最大的长期总体网络实用性。

关键词: 异构网络用户关联功率控制强化学习深度Q学习网络（DQN）

DOI：

基金项目:河北省高等学校青年基金项目(SQ201042)

Cooperative optimization of user association and power control in heterogeneous networks

FAN Wen,CHEN Teng,JIAN Yingbin

(Department of Information Engineering,Shijiazhuang Railway Vocational and Technical College,Shijiazhuang 050041，China)

Abstract:

To solve the problem of poor joint optimization of user association and power control in orthogonal frequency division multiplexing access(OFDMA) heterogeneous networks,a multiagent deep Qlearning network(DQN) method is proposed.First,a twolayer heterogeneous network system model of OFDMA is constructed to realize intelligent decisionmaking based on user association and power control optimization.Secondly,the state space and reward function are reconstructed according to the application scenario and the action space of the multiagent DQN framework.Finally,the performance of the multiagent DQN algorithm is simulated by selecting a twolayer heterogeneous network with macro base station(BS) and small BS.The simulation results show that the multiagent DQN algorithm has better convergence,and can effectively improve the service quality and energy efficiency of user equipment(UE).At the same time,the maximum longterm overall network utility can be obtained compared with traditional learning algorithms.

Key words: heterogeneous network user association power control reinforcement learning deep Q learning network(DQN)