Femtocell deployment is one of the key solutions to achieve the high data rate of the fifth generation mobile communication. Nevertheless, dense femtocell networks face several challenging tasks such as interference control and resource management. In this paper, we address the problem of resource allocation for heterogeneous networks (HetNets), namely dense femtocell networks, by forming stable clusters using an evolutionary game where femtocells learn from the environment and make their decisions considering the achieved payoff. In the literature, clustering has been proposed to organize network topologies by joining nodes (e.g. femtocells) with similar behaviors into logical groups. We focus on cluster stability that is important to obtain good network performance but can be difficult to achieve especially in ultra-dense and heterogeneous networks. In order to guarantee the cluster stability, we use the replicator dynamics that find the evolutionary equilibrium of the evolutionary game. Thus, by guaranteeing cluster stability the network performance is improved and the computational complexity is reduced. In addition, Particle Swarm Optimization (PSO) is used for the resource allocation algorithm that runs locally within each cluster owing to the fact that PSO has been proved to find a satisfying near-optimal solution while having the advantage of speeding up the optimization process. We run simulations for non-dense and dense femtocell networks taking into account two scenarios: fixed public users and public users that keep mobility such as pedestrians or cyclists. Simulation results show that the proposed solution is able to enhance the network throughput, to provide higher subscribers satisfaction, and to reduce the co-tier interference in dense femtocell networks.

毫微微蜂窝基站部署是实现第五代移动通信的高数据速率的关键解决方案之一。尽管如此，密集的毫微微小区网络仍面临一些挑战性任务，例如干扰控制和资源管理。在本文中，我们通过使用进化博弈形成稳定的集群来解决异构网络（HetNet）（即密集的毫微微小区网络）的资源分配问题，其中毫微微小区从环境中学习，并根据获得的收益做出决策。在文献中，已经提出了聚类以通过将具有类似行为的节点（例如，毫微微小区）加入逻辑组来组织网络拓扑。我们关注于群集的稳定性，这对于获得良好的网络性能很重要，但可能很难实现，尤其是在超密集和异构网络中。为了保证群集的稳定性，我们使用复制器动力学来找到演化博弈的演化平衡。因此，通过保证群集稳定性，可以提高网络性能并降低计算复杂度。此外，由于粒子群优化（PSO）已被证明可以找到令人满意的近似最佳解决方案，同时具有加快优化过程的优势，因此它用于在每个群集内本地运行的资源分配算法。考虑到两种情况，我们对非密集和密集的毫微微小区网络进行了仿真：固定的公共用户和行人或骑自行车者等保持机动性的公共用户。仿真结果表明，所提出的解决方案能够提高网络吞吐量，提高用户满意度，并减少密集型毫微微小区网络中的co-tier干扰。