面向C-RAN的群稀疏线性预编码

云无线接入网络（cloud radio access network，C-RAN）是一种能够集中处理信号的网络架构。C-RAN能够通过算法动态选择无线电单元（remote radio head，RRH）来调整用户通信速率。而通信速率作为用户服务质量（quality of service，QoS）的关键部分，当参与服务的RRH越多时，用户的通信速率更大且体验更好，但同时所带来的能源损耗越大，因此本文研究通信速率和功率消耗二者之间的权衡关系。提出一种优化算法，将权衡问题建模成一个单目标优化模型，通过权衡系数来协调速率和RRH激活个数之间的矛盾。为了解决$\ell_0$-范数的非凸问题，本文使用重复加权$\ell_1$-范数去近似$\ell_0$-范数，同时使用加权最小均方误差（weighted minimum mean square error，WMMSE）的方法将通信速率从非凸问题转换成一个凸问题，最后使用改进的次梯度法对预编码矩阵进行更新。仿真结果证明该算法减少了时间复杂度，同时达到了与穷举法相近的性能。

Group sparse linear precoding for C-RAN

Cloud radio access network (C-RAN) is a network architecture that centralized signal processing. C-RAN can dynamically adjust the user communication rate by dynamically selecting remote radio heads (RRHs). Rate performance is a key part of user quality of service (QoS). As the number of RRHs increases, the rate performance can be improved at the expense of higher energy consumption. Therefore, the tradeoff between rate performance and power consumption is focused. In this paper, an optimization algorithm is proposed. In this algorithm, the contradiction between the rate maximization and the number of RRHs minimum is coordinated through the trade-off coefficient. To solve the non-convex problem, the re-weighted $\ell_1$ norm to approximate the $\ell_0$ norm is used, and the weighted minimum mean square error (WMMSE) method is used to convert the non-convex sum rate express to a convex function. Finally, a modified subgradient method is used to update the precoding matrix. The simulation results show that the method reduces the complexity and achieves near-optimal performance.