Gaussian Approximation Based Interpolation for Channel Matrix Inversion in MIMO-OFDM Systems

Channel matrix inversion, which requires significant hardware resource and computational power, is a very challenging problem in MIMO-OFDM systems. Casting the frequency-domain channel matrix into a polynomial matrix, interpolation-based matrix inversion provides a promising solution to this problem. In this paper, we propose novel algorithms for interpolation based matrix inversion, which require little prior information of the channel matrix and enable the use of simple low-complexity interpolators such as spline and low pass filter interpolators. By invoking the central limit theorem, we show that a Gaussian approximation function well characterizes the power of the polynomial coefficients. Some low-complexity and efficient schemes are then proposed to estimate the parameters of the Gaussian function. With these estimated parameters, we introduce phase shifted interpolation and propose two algorithms which can achieve good interpolation accuracy using general low-complexity interpolators. Simulation results show that up to 85% complexity saving can be achieved with small performance degradation.