有限回程链路容量下的多小区处理技术研究

In this project, we study the problem of multi-cell processing with finite backhaul capacity. We take a different viewpoint from existing results, and look at the communication model as a multi-destination multiple access diamond channel. Firstly, we plan to extend the achievability of the multiple access diamond channel to incorporate the existence of multiple destinations and propose a novel scheme of correlated codebooks. Secondly, we plan to combine the converse proofs of both the multiple access diamond channel and the multi-antenna broadcast channel, and derive an outer bound on the capacity region that is tighter than the existing outer bound, which is based on the max-flow min-cut theorem. In practical systems, to reduce the complexity of base station cooperation, each base station will only serve a subset of end-users, rather than all end-users. So thirdly, we plan to study the allocation of power, spatial dimensions and backhaul capacity of each base station among the different sets of end-users that it is serving, and also study the optimal matching between the base stations and the end-users. Finally, considering the scenario that base stations may cache some popular contents beforehand to save backhaul capacity, we plan to extend our results to the scenario where the base stations have already cached part of the end-users' data. The aim here is to understand how to use the backhaul capacity to fully take advantage of the already cached data. The goal of the project is to provide a novel downlink transmission scheme in the scenario of multi-cell processing with finite backhaul capacity, and further provide some theoretical results to help understand what is the most efficient way to utilize the finite capacity of the backhaul links.

项目拟研究有限回程链路容量下的下行多小区处理场景。新颖性在于把通信模型看成是多信宿的多址菱形信道，这与已有的研究不同。项目一方面拟将多址菱形信道的编码思想扩展到多信宿中，提出新颖的相关码本方案；另一方面将多址菱形信道和多天线广播信道的容量上界进行结合，拟推导出目前唯一比最大流最小割紧的性能上界，并在某些情况下证明所提的相关码本方案的最优性。由于实际中为降低小区合作的复杂度，每个基站只会服务部分而不是全部用户，因此项目接下来将研究基于相关码本方案的基站回程链路、功率和空间资源在不同服务用户集之间的分配问题以及用户和基站的配对问题。最后，考虑到基站可能缓存热门内容，项目拟将结果拓展到基站缓存用户部分数据的场景，研究回程链路的使用应如何配合基站已缓存的数据。通过项目的研究，希望能为有限回程链路容量下多小区处理场景提供一种新颖有效的下行传输方案，为最有效利用回程链路容量的问题提供理论依据。

项目研究有限回程链路容量下的下行多小区处理场景。在传输方案设计方面, 将多址菱形信道的编码思想扩展到多基站多终端用户场景中，给出基于相关码本的完整、新颖的传输方案。方案强调基站传输信号的相关性，因而能更加高效充分地利用网络的第二跳信道。因此，在回程链路容量中等时，较已有方案可传输更高的速率；在性能上界方面，通过把高斯多址菱形信道和多天线广播信道逆定理方法有机结合，推导出多个基站两个用户下新颖的有限回程容量多小区处理的容量区域外界。在回程链路容量中等时，该外界比目前唯一已知性能外界－－简单割集界紧很多。将此容量外界与已知可达速率相比较，可得知已知可达方案的性能已经相当优异。在选择服务基站及功率、空间资源分配方面，我们给出基站用户配对及预编码矩阵设计的几种算法，这些算法克服用户服务基站选择的整数规划难点，复杂度低，并在降低系统能耗或者增加系统能量效率方面有着优秀的表现。在基站缓存内容设计与回程链路使用方面, 我们首先证明了将相关码字引入基站缓存方案的最优性。其次，我们额外研究两个相关问题：1）在两个基站缓存大小不同并通过三条有限容量回程链路（两条私有链路和一条共享链路）连接中心节点时，给出编码缓存设计及传输方案的最优解，清晰地揭示了最高效利用缓存、私有链路和共享链路的方法；2）针对回程链路容量有限及基站多天线场景，推导随机缓存方案下最优文件数和缓存概率需满足的条件，根据这些条件提出寻求最优文件数和缓存概率的低复杂度算法，揭示最优缓存概率随天线数变化趋势及天线数与缓存大小可互补替换的关系。项目的研究成果为回答如何最有效利用回程链路的有限容量提供了设计指导和理论依据。