针对密集异构网络自回程场景中带宽分配不合理引起的负载不均衡问题,提出一种基于self-backhaul感知的用户接入负载均衡方案.首先根据密集异构网络下各个小基站接入与回程资源的负载状态提出一种用户接入负载均衡策略;其次利用Q-Learning算法对各个小基站带内无线接入与回程带宽分配进行学习,用户在不同带宽分配因子下,根据用户接入负载均衡策略进行重新接入,得到不同接入情况下的系统效用,进而得到最优带宽分配策略,保证负载均衡性的同时实现系统效用最大化.仿真结果表明,该方案在密集异构网络自回程场景中提高了网络负载均衡性,同时提升了用户速率体验. Aiming at the problem of unbalanced load caused by unreasonable bandwidth allocation in dense self-backhauling scenarios, this paper proposes a self-backhaul-based user access load balancing scheme.At first, according to the access and backhaul of each small cell in dense heterogeneous networks, Resource load status, we propose a user access load balancing strategy. Secondly, we use the Q-Learning algorithm to learn the bandwidth allocation of each small cell in-band wireless access and backhaul. Under different bandwidth allocation factors, Policy to re-access, access to the system under different access conditions, and then get the optimal bandwidth allocation strategy to ensure load balancing while maximizing system utility.The simulation results show that the program in dense self-backhauling scenarios Improves network load balancing and improves user experience.