基于大规模天线阵的多长度导频机制研究

A massive MIMO system based multi-length pilot scheme

在基于时分双工的大规模天线阵系统中,导频污染是阻碍实现高频谱效率、高吞吐率的最主要瓶颈之一.传统的单长度导频机制使得用户受到所有相邻小区非正交导频序列用户的干扰,进而加重导频污染问题.因此,提出一种新的多长度导频机制(MLPS),使得目标小区各用户的导频长度可以更加接近其最优导频长度,从而减弱导频污染对系统的影响进而提升吞吐率的目的.首先,利用Monte-Carlo仿真观察出系统内各用户最优导频长度的分布情况.然后,基于多长度导频机制带来的不同长度导频间干扰与导频-数据干扰,给出一种可行的导频设计准则和两种时分双工时序策略,分别用于消除以上两类干扰.实验结果表明,相比于传统单长度导频机制,多长度导频机制在上行、下行吞吐率上分别有18%与20%的提升.

In massive MIMO system,pilot contamination is considered as a challenging problem,which deteriorates channel estimation at base station（ BS） and therefore reduces the total throughput. In the conventional single-length pilot schemes,such contamination is severe due to the full reuse of pilot sequences in adjacent cells. Therefore,in this paper,we propose a novel multilength pilot scheme（ MLPS） so that each user equaipment（ UE） in the target cell can approach respective optimal pilot length and the total throughput can be increased. First,we illustrate the fact that the optimal pilot lengths of UEs in the same cell are distributed among several values. Then,to eliminate the interferences caused by nonorthogonality of different-length pilots and simultaneous transmission of pilots and data,a feasible pilot design criterion is presented and two time-division strategies are developed. Finally,we evaluate our schemes by extensive simulations,MLPS obtains a gain of up to 18% for uplink throughput and up to 20% for downlink throughput.