Channel Noise Estimation for MIMO-OFDM using Mask Function

Abstract

In wireless communication systems, the estimated noise variance is a significant parameter for SNR (Signal to Noise Ratio) estimation, estimation of channel impulse response, and MMSE (Minimum Mean Square Error) calculation. More accurately estimated noise variance provides better signal detection performance of systems. However the conventional noise estimation methods result in degradation of performance for the estimation when the noise power becomes higher than threshold value, especially with MMSE filter. In this paper, for more accurate noise estimation, a new noise estimation scheme is proposed by using mask function for FFT (Fast Fourier Transform) block in frequency domain at receiver. The mask function can reduce magnitude of side-lobes of channel response in time domain then it can estimate more accurate reference symbols. Through estimated reference symbols for channel estimation step, more accurate noise component can be estimated so we can obtain better performance of noise variance estimation. The accurately estimated noise variance value is used to calculate weighting matrix for MMSE filter. To prove that the systems using a new noise estimation scheme have better BER performance, we could analyze under several channel variations and compare BER performances for MIMO-OFDM (Multiple Input Multiple Output – Orthogonal Frequency Division Multiplexing) systems. BER performances using the proposed scheme are better than those using conventional noise estimation scheme under all channel situations.