基于矩阵和求逆及MRC的低复杂度MIMO检测器

Low Complexity MIMO Detector Based on Inversion of Sum of Matrices and MRC

信道编码MIMO系统需要检测器具有软输入软输出特性,而常规的检测算法通常具有很高的计算复杂度,阻碍了其在实际中的应用。提出一种低复杂度MIMO检测方案。首次迭代中,利用低复杂度快速矩阵和分解方案来获得MMSE检测输出,避免了常规矩阵和求逆中的Jordan标准型化简;其余迭代中,利用信道解码器提供的软信息将MIMO系统转化成多个并行的SIMO系统,进而利用MRC进行检测,其仅需线性运算,克服了MMSE检测器中矩阵求逆的缺点,适合硬件的并行处理特性,大大降低了检测复杂度。仿真结果表明,该检测器和MMSE检测器具有相同的误比特性能和收敛性能,但是前者同后者相比,计算复杂度在首次迭代由O(n2.376)降到了O(n2),在其余迭代由O(n2.376)降到了O(n),因而为实际的MIMO系统提供了一种可行的低复杂度检测方案。

The MIMO detector with the character of soft-input soft-output（SISO） is needed for channel coded MIMO systems,however,conventional detecting algorithms commonly have so high computational complexity,which is prohibitive to be used in practical applications.A low complexity MIMO detecting algorithm is proposed.In the first iteration,a fast decomposition of sum of matrices with low complexity is used to get MMSE output,which avoids the reduction of Jordan canonical form in traditional inversion of sum of matrices.In the rest iterations,soft information from channel decoder is used to transform a MIMO system into multiple parallel SIMO systems.Then MRC is readily utilized as MIMO detector,which only needs linear computation and overcomes the drawback of matrix inversion associated with MMSE detector,and suits parallel processing of hardware,and hence dramatically reduces computational complexity of detection.The simulation results show that the proposed detector has the same performance in BER and convergence as MMSE detector,however,the computational complexity of the former one,compared to the latter one,is reduced form O（n2.376） to O（n2） at the first iteration and from O（n2.376） to O（n） at rest iterations.Consequently,the proposed detector provides a feasible low-complexity solution to practical applications.