大规模MIMO预编码中降低反馈开销的方法

研究大规模多输入多输出(MIMO)系统下,天线数增多反馈开销随之显著增大的问题,提出一种分割码本的预编码方法。该方法将有限反馈的大码本等间隔地分成k个码本,信道变化一次k个码本中的多个码本就被循环一次,选出一个最优码字Fi。Fi放入新开辟的储存空间,方便与下次选出的最优码字比较,循环一次Fi更新一次。同时将注水算法与之结合,使系统性能大大提高,而不会增加反馈量。仿真表明：正交移相键控(QPSK)调制、离散傅立叶变换(DFT)码本的预编码、反馈比特数为4的条件下,系统误码率与单一码本下预编码误码率对比,约提升3dB；用较少反馈量便可取得与大码本预编码几乎相同的误码率。

Large Scale MIMO pre-coding method to reduce feedback overhead

with the number of antennas increasing, feedback overhead increases rapidly in massive multiple input multiple output (MIMO) system. To solve the problem, a new pre-coding which makes the big code-book divide into k code-books uniformly-spaced is proposed. When channel changes one time, multiple code-books in k code-books will be cycled once, then an optimal code-word Fi is chosen, which will be updated once multiple code-books are cycled, is put into the new storage space to compare with the next optimal code-word conveniently. It combines with the water-filling algorithm, making the system performance improved greatly, without increasing the number of bits for each feedback. Simulation results show that under the condition of quadrature phase shift keying (QPSK) modulation, discrete Fourier transform (DFT) code-book and feedback bits being 4, its bit error ratio (BER) is about 3dB lower than that of single code-book limited feedback pre-coding; it can obtain almost the same BER as that of the large code-book pre-coding with fewer feedback bits.