Aiming at solving the problem of multi-user interference caused by channel correlation in time reversal division multiple access system,eliminating inter-use interference and reducing computational complexity,a low complexity approximation algorithm of matrix decomposition of tridiagonal matrices based on minimum mean square error is proposed.Firstly, the tridiagonal matrix that include the main diagonal line in the detection matrix(Gram-N) is extracted,and the tridiagonal matrix into two double diagonal lines matrices is divided.Then,by using the inverse rule of double diagonal lines matrix, the inverse of these two double diagonal lines matrices is obtained, the inverse of the tridiagonal lines matrix is applied in the algorithm.Finally,according to the Neumann series approximation,the goal of approximating Gram-N's inverse is realized by using the inverse of tridiagonal matrix.Simulations show that the proposed algorithm has obvious performance advantages in terms of bit error rate,spectral efficiency,and complexity.The algorithm can obtain a near-optimal performance gain when the complexity is low.