一种基于码本聚类的低复杂度用户选择算法

CodeBook cluster user selection with block diagonalization

现有块对角化(BD)预编码系统用户选择算法较少考虑利用已选用户与剩余用户之间的关系来排除不可选用户。针对这点不足,给出了一种采用码本聚类的低复杂度用户选择算法(Code-Greedy算法)。该算法采用弦距离刻画用户信道的相关性,以此为依据将用户划分到不同码本空间中,聚集在同一码本空间的用户信道具有较强的相关性,形成互斥用户组,压缩了算法搜索空间,降低了复杂度。复杂度分析和仿真结果表明,与传统贪婪算法相比,该算法在系统和容量损失非常小的情况下,计算复杂度降低了约17%。

In the present algorithms of user selection for BD precoding system, the relationship between the selected user and the left user is rarely utilized to eliminate user. With consideration of this problem, a low complexity user selection algorithm with codebook cluster dedicated is proposed in this paper. The algorithm divides users into different codebook spaces according to their correlation value,which is scaled by the chordal distance,users in the same space are defined as mutual users and their channels arc correlative highly, this compresses the search range, so reduces the computational complexity distinctly. Complexity analysis and simulation results show that the algorithm reduces computational complexity by about 17% with little system capacity＇ losses compar- ing with that of traditional greedy algorithm.