大规模MIMO-NOMA下行系统可达吞吐量研究

Achievable throughput in massive MIMO-NOMA downlink system

非正交多址系统的可达吞吐量与用户分簇策略、预编码方法、功率分配算法等有关.为了提高系统可达吞吐量,提出了一种改进的k-means用户分簇算法,该算法利用空间相关性对用户进行分簇,从而降低簇间干扰;然后利用块对角化预编码对发送给各簇的信息进行预处理,使得每簇的预编码向量与其它簇等效信道向量近似正交,进一步消除簇间干扰;同时构建最优功率分配优化问题,并利用Karush-Kuhn-Tucker(KKT)条件求得簇内每个用户的最佳功率分配系数,降低簇内用户间的干扰.实验结果表明:所提出的方法能有效提高系统的可达吞吐量.

The achievable throughput in non-orthogonal multiple access system is mainly related to the user clustering strategy, precoding method and power allocation algorithm. In order to improve the achievable throughput, improved k-means algorithm for user clustering based on spatial correlation is proposed to mitigate the inter-cluster interference. Meanwhile, the block diagonal precoding is utilized to preprocess the user signals, which makes the precoding vector for each cluster approximatively orthogonal to the equivalent channel vector of other clusters. As a result,the inter-cluster interference is eliminated. Moreover, power allocation is optimized to mitigate intra-cluster interference, and the optimal power allocation coefficient of every user is found by solving the optimal problem with Karush-Kuhn-Tucker. Numerical simulation shows that the proposed method can improve the system throughput.