超大规模MIMO下行系统1位传输信号设计

Construction of 1-Bit Transmit Signal Vectors for Downlink Ultra-Massive MIMO System

在超大规模MIMO系统中,采用1位数模转换器可降低硬件成本和功耗,但增加了预编码器设计的难度.文章研究PSK调制下超大规模MU-MIMO系统的1位预编码问题.首先利用相长干涉的思想,对接收信号进行坐标变换,将1位预编码问题建模成多目标优化问题.然后通过降维的方法,把多目标优化问题转换成双目标优化问题,提出一种基于相长干涉的交替优化算法来获得双目标优化问题的次佳解.该算法包括初始分配阶段和局部搜索优化阶段,按逐个元素的方式确定1位量化信号.结果表明,该算法可获得与现有的非线性预编码方法相媲美的性能,同时具有更好的性能-复杂性平衡.

In an ultra-massive MIMO system,1-bit digital-to-analog converters(DACs)can reduce the hardware cost and power consumption,but also bring the challenge of precoding design.We study the 1-bit precoding problem in a downlink ultra-massive MIMO system with PSK modulation,where the base station employs 1-bit DACs.This study first exploited the concept of constructive interference to express the received PSK signal in terms of the coordinate vector.With the coordinate vector,a multi-objective optimization problem is formulated and is then transformed into a two-objective optimization problem by means of dimension reduction.Afterward,we proposed a CI-based alternating optimization algorithm to obtain the sub-optimal solution through iterations.The proposed algorithm consisted of the assignment and refinement stages and determined the transmitted signal sequentially.The numerical results showed the proposed algorithm achieved comparable performance to the existing nonlinear precoding approaches while yielding a more elegant performance-complexity tradeoff.