This correspondence presents a new matrix inversion approximation (MIA) method for massive multiple-inputmultiple- output (MIMO) systems. In contrast to the existing methods which are mostly derived from the Neumann series expansion framework, additional coefficients have been introduced in our proposed method to enhance the precision of approximation. We propose an efficient algorithm for the coefficient design which consists of an eigenvalue estimation procedure derived from random matrix theory, and a least-squares fitting procedure that solves a low-dimension over-determined system of linear equations. Complexity analysis and simulation results show that our eigen-based MIA method exhibits practically comparable computational complexity while achieving substantial performance enhancement compared to other benchmark methods.
IEEE Transactions on Vehicular Technology 66(6), p.5480-5484