大规模MIMO系统信道状态信息反馈开销降低方法

Channel State Information Feedback Overhead Reduction Method for Massive MIMO Systems

针对大规模多输入多输出(massive-multiple input multiple output,massive MIMO)系统,结合离散余弦变换(discrete cosine yransform,DCT)和快速傅里叶变换(fast fourier transform,FFT)基,研究了基于压缩感知的信道状态信息(channel state information,CSI)反馈开销降低方法。首先在用户端,该方法基于压缩感知理论对三维(three dimension,3D)CSI采用不同稀疏基组合进行表示,进而形成两种不同的观测矩阵对其进行观测,其中方法 1基于3D CSI直接形成观测矩阵,而方法 2则基于垂直维CSI和水平维CSI分别形成两个中间观测矩阵,进而通过Kronecker积形成最终观测矩阵。最后,将观测值经矢量化后反馈给基站端;基站端则通过正交追踪匹配算法(orthogonal matching pursuit,OMP)重构CSI。仿真结果表明,基于两种稀疏基的组合可以使得CSI反馈开销得到大幅度降低,同时,基于方法 1生成的观测矩阵所重构CSI性能明显优于基于方法2的。

This paper combined with the discrete cosine transform （DCT） and fast Fourier transform （FFT） for Massive-Multiple Input Multiple Output （Massive MIMO）, study of the channel state information （CSI） feedback overhead reduction method based on compression sensing. First, at the receiver, the method use different sparse base combinations for three-dimension CSI to express based on compressed sensing theory. Then form two different observation matrixes and observe it, which the method one is formed observation matrix based on three-dimension CSI, the method two is formed two intermediate observation matrix based on the vertical and horizontal CSI, and form the final observation matrix by Kronecker product. Finally, a small amount of measurements after vectoring were fed back to the base station. In the base station, it reconstructed the CSI through orthogonal matching pursuit algorithm. Simulation results show that based on combination with two sparse base can make the CSI feedback overhead has been greatly reduced, at the same time, the observation matrix was generated by method one which can recover the CSI is superior to the observation matrix was generated by method two.