微蜂窝是实现5G网络容量提升1 000倍和提供无处不在宽带覆盖最具潜力的方法,其中如何实现动态资源分配是一个关键问题。文章基于微蜂窝负载和用户通信需求,提出了一种新的动态资源分配算法。该动态资源分配问题划分为两个子问题:基于负载量的信道选择和基于用户需求的功率控制。两个子问题建模为博弈模型,并分别提出相应的在线学习算法。仿真结果验证了算法的收敛性,通过信道选择和功率控制降低了系统干扰,提高用户的数据传输速率。 Micro-cellular is the most promising way to achieve 1 000 times the capacity increase of 5G networks and to provide ubiquitous broadband coverage. How to realize dynamic resource allocation is a key issue. Based on the requirements of micro-cell load and user communication, a new dynamic resource allocation algorithm is proposed. The dynamic resource allocation problem is divided into two sub-problems: load-based channel selection and user-based power control. The two subproblems are modeled as game models, and the corresponding online learning algorithms are proposed respectively. Simulation results verify the convergence of the algorithm, reduce system interference through channel selection and power control, and improve user’s data transmission rate.