毫米波NOMA系统的一种用户分簇及预编码算法

A user clustering and precoding algorithm for millimeter wave NOMA systems

针对毫米波大规模多输入多输出非正交多址接入(multiple input multiple output-non-orthogonal multiple access,MIMO-NOMA)系统,给出基于K均值的用户分簇及选择算法和基于信漏噪比最大混合预编码算法。基站根据归一化角度筛除掉部分用户,再利用K均值方法对剩余用户进行分类,将每类中信道增益差最大的一对用户作为一簇。根据信漏噪比最大原则为每簇用户设计模拟预编码矢量,并由基站利用每簇强用户的等效信道生成数字预编码矩阵。所给算法可降低用户受到的簇间干扰,提升系统速率,适用于毫米波大规模MIMONOMA系统。

For millimeter-wave massive multiple input multiple output non-orthogonal multiple access (MIMO-NOMA) systems,a user clustering and selection algorithm based on K-mean and a hybrid precoding algorithm based on maximum signal-to-leakage-and-noise ratio (SLNR) are presented.The base station filters out some users according to their normalized angle and then classifies the remaining users by using the K-means method.A pair of users with the largest channel gain difference in each class is regarded as a cluster.According to the principle of maximum SLNR,analog precoding vectors are designed for each cluster user,and the digital precoding matrix is generated by the base station using the equivalent channel of each cluster strong user.The presented algorithm can reduce the intercluster interference and increase the system speed.It is suitable for millimeter-wave massive MIMO-NOMA systems.