超密集网络中最大化网络吞吐量的预测资源分配

在超密集网络中，小区间干扰严重制约了小区边缘用户的性能体验以及网络吞吐量。无线大数据分析的飞速发展，使得人们有可能通过预测未来的信道状态来分配资源，在无干扰网络中可达到很大的性能增益。但是在干扰网络中如何利用预测信息，在分配资源的同时有效协调干扰还是一个尚未研究的问题。本文分析了干扰网络中预测资源分配的设计难点和存在的问题，针对该问题提出了相应的解决方法，将资源分配建模成一个凸优化问题，通过求解优化问题得到最优的资源分配方法。仿真表明，与未知预测信息的最大化网络吞吐量方法相比，所提方法能够有效提高用户的成功传输率、平均传输进度和网络的吞吐量。当用户数据需求较大时，所提方法可以提供较大的网络性能增益。 

Predictive Resource Allocation to Maximize the Throughput of Ultra-Dense Networks (UDNs)

In ultra-dense networks (UDNs), the inter-cell interference (ICI) severely limits the performance experience of cell edge users and the network throughput. The rapid development of wireless big data analysis, makes it possible to predict the future state of the channel to allocate resources, achieving considerable gain in the interference-free network. However, in the interference network, how to allocate resources and coordinate the interference with predictive information is still an open problem. This paper analyzes the difficulties and issues of designing predictive resource allocation, and proposes the solving methods. By formulating the resource allocation as a convex optimization problem, we obtain the optimal resource allocation method. Simulation results indicates that, compared with the existing methods maximizing the network throughput without prediction information, the proposed method can effectively improve users’ transmission success rate, average transmission progress and the network throughput. When the users’ data demands are high, the proposed method can provide considerable network performance gain.