【摘要】 在即将来临的5G时代,随着数据量和接入设备量的激增,异构蜂窝网络(Heterogeneous Networks,HetNets)成为满足业务需求的重要手段。资源分配问题是异构蜂窝网络的重点研究领域。在综合考虑资源分配和用户关联时,该问题是一个NP-hard问题。基于自动微分(Automatic Differentiation,AD)方法可获得其可行解,即将该问题的目标函数与约束条件用动态计算图描述,利用自动微分方法迭代计算该计算图的损失函数获得其可行解。最后,仿真验证了该算法的有效性,与对比方案相比,可以实现更高的网络吞吐量和更高的频率重用。更多还原

【Abstract】 In the coming 5 G era, heterogeneous cellular networks are adopted to meet the rapid increasing of user demand and the amount of access devices. The resource allocation problem is a key research field of heterogeneous networks. In comprehensive consideration of resource allocation and user association, this problem is but a NP-hard problem, and its feasible solution can be obtained based on AD(automatic differentiation) method, that is, the constraints and objective function of this problem are described by with a dynamic calculation chart, and by using automatic differentiation to iteratively calculated the loss function of this calculation chart, its feasible solution can be obtained. Finally, the simulation results indicate the effectiveness of this algorithm. Even higher network throughput and higher frequency reuse can be achieved as compared to the max SINR and SINR bias schemes.更多还原