| 摘要: |
| 针对正交时频空(Orthogonal Time Frequency Space，OTFS)调制系统中均衡器性能不佳及线性滤波器复杂度较高等问题，提出了一种LU(Lower-Upper)分解与迭代最小均方误差(Iterative Minimum Mean Square Error，IMMSE)均衡器结合的OTFS系统信号检测算法(LU-IMMSE)。该算法依据时延多普勒域稀疏信道矩阵的特征，采用一种低复杂度的LU分解方法，以避免MMSE均衡器求解矩阵逆的过程，在保证均衡器性能的前提下降低了均衡器复杂度。在OTFS系统中引入一种IMMSE均衡器，通过不断迭代更新发送符号均值和方差这些先验信息来逼近MMSE均衡器最优估计值。LU-IMMSE算法通过调节迭代次数可以有效降低误比特率。在比特信噪比为8 dB时，5次迭代后的LU-IMMSE均衡器误比特率相比传统的MMSE均衡器降低了约11 dB。随着迭代次数的增大，较传统IMMSE算法降低了计算复杂度。在最大时延系数为4、符号数为16的情况下，与直接求逆相比，所提出的低复杂度LU分解方法降低了约91.72%的矩阵求逆计算复杂度。 |
| 关键词: 正交时频空(OTFS) 信号检测 最小均方误差均衡 三角分解 |
| DOI：10.20079/j.issn.1001-893x.240112001 |
|
| 基金项目:重庆市自然科学基金创新发展联合基金（中国星网）(CSTB2023NSCQ-LZX0114) |
|
| A Low Complexity Signal Detection Algorithm for OTFS Systems |
| CHEN Fatang,CHEN Jiajie,XIA Qiyu,HUANG Liang |
| (School of Communications and Information Engineering，Chongqing University of Posts and Telecommunications，Chongqing 400065,China) |
| Abstract: |
| For the problem of poor equalizer performance and high complexity of the linear filter in orthogonal time frequency space(OTFS) modulation systems,a signal detection algorithm for OTFS systems combining lower-upper(LU) decomposition and iterative Minimum Mean Square Error(IMMSE) equalizer(LU-IMMSE) is proposed.According to the characteristics of the sparse channel matrix in the Doppler-delay domain,a low-complexity LU decomposition method is proposed to avoid matrix inversion calculation of the MMSE equalizer.On the premise of guaranteeing the performance of the equalizer,the complexity of the equalizer is reduced.Moreover,an iterative MMSE equalizer is introduced into the OTFS systems,which approximates the optimal estimate of the MMSE equalizer by iterative updating the prior information such as the mean and variance.The proposed LU-IMMSE algorithm can effectively reduce the bit error rate(BER) by adjusting the iterative times.When the energy per bit ratio noise power spectral density is at 8 dB,the LU-IMMSE equalizer after five iterations reduces the BER by about 11 dB compared with the traditional MMSE equalizer.With the increase of iterative times,the computational complexity is reduced compared with the traditional iterative MMSE algorithm.When the maximum delay coefficient is four and the number of signs is sixteen,the proposed low-complexity LU decomposition method reduces the computational complexity of matrix inversion by about 91.72% compared with the direct inversion method. |
| Key words: orthogonal time frequency space(OTFS) signal detection minimum mean square error equalizer LU decomposition |
