杨伟民^1，， 赵新胜^1，*， 邹素玲^2，

（  1、东南大学 移动通信国家重点实验室，南京 210096；       2、大唐移动通信设备有限公司移动通信事业部，北京 100083；  ）

摘要： 移动通信网络不断演进为高密集、无定型的异构网络，异构网络资源管理的挑战之一是业务需求与资源供给的匹配问题，即负载均衡问题。负载均衡作为LTE-A系统自优化网络中的关键技术，应具有自适应、自优化的特点。该文将智能控制中的Q-Learning 思想引入到LTE-A HetNet系统中，用于解决LTE-A宏基站与微基站混合组网的复杂通信环境下，存在网络拥塞和负载不均衡等问题，给出了一种基于Q-Learning的LTE-A HetNet系统负载均衡算法。该算法将各个家庭基站看作一个智能体Agent，通过不断地与环境进行交互和学习来自主获取最优的参考信号功率控制策略。仿真表明，该算法具有分布式、低复杂度等特点，同时有效提高系统吞吐量，降低系统阻话率和服务中断率，使系统达到负载均衡，可作为未来移动通信系统走向智能化的一种尝试。
关键词： LTE-A HetNet；自优化网络；负载均衡；Q-Learning

Yang Weimin^1,， Zhao Xinsheng^1,*， Zou Suling^2,

（  1、National Mobile Communications Research Lab., Southeast University, Nanjing 210096；       2、Datang Mobile Communications Equipment Co mobile communications division, Beijing 100083；  ）

Abstract： The mobile communication network is constantly evolving into high-density and non-shaped heterogeneous network. One of the challenges in the management of heterogeneous network resources is the matching of traffic demand and rnetwork resources. Load balancing is a key technique in LTE-A system self-organizing network, which should be adaptive and self optimized.In this paper, the Q-Learning idea of intelligent control is introduced into the HetNet LTE-A system, which is used to solve the problem of network congestion and load imbalance in the complex communication environment, and a load balancing algorithm based on LTE-A HetNet Q-Learning system is presented.In this algorithm, each femto is considered as an intelligent Agent, and the power control strategy is obtained by interacting with the environment and continuous learning. Simulation results show that the algorithm has the characteristics of distributed, low complexity and can effectively improve the system throughput, reduce the call blocking Rate and service disruption rate. So that the system can achieve load balance. The idea can be an attempt in the mobile communication system in the future.

Tag：

点此返回栏目查看更多>>>参考论文