An Effective Channel Estimation for Massive MIMO–OFDM System

Abstract

The massive Multiple-Input and Multiple-Output Orthogonal Frequency Division Multiplexing (MIMO–OFDM) system provides a high data transmission for the next generation mobile communication i.e. 4G, 5G, etc. In the practical MIMO–OFDM system, N points of IFFT/FFT is larger than M data subcarriers ( $$N>M$$ ) in each OFDM symbol to reject the aliasing after D/A converter. To demodulate information data, the channel responses for all MIMO channel links need to be estimated so as to employ in MIMO data detection for demodulation at the receiver. The discrete Fourier transform estimator (DFE) was proposed for the system which can estimate the MIMO channels accurately when $$N=M$$ . However, its accuracy will be hugely degraded when $$N>M$$ because of the oversampling of data transmission. To improve the estimation accuracy when $$N>M$$ , the maximum likelihood estimator (MLE) was proposed for the system which can achieve higher estimation accuracy than that of the DFE. However, its accuracy will be degraded a lot in the massive MIMO–OFDM system when $$N>M$$ , due to the estimation error increased in proportion to the increasing of $$N_T$$ transmit antennas. To solve these problems, this paper proposes a direct time-domain estimator (DTE) with preamble symbol with scattered-pilot (preamble-SCP) for the massive MIMO–OFDM system when $$N>M$$ . In the proposed method, it is presented with three salient features; achieving higher estimation accuracy with keeping almost the same computational complexity as the conventional estimators, improving $$Bit-Error-Rate$$ (BER) with low-complexity MIMO data detection, and providing higher transmission data rate compared with the MLE. Using the $$normalized\,MSE$$ , BER and throughput evaluated by computer simulations, it can be verified that the proposed DTE with preamble-SCP obviously provides higher estimation accuracy, better BER with low-complexity MIMO data detection, and much higher transmission data rate which is approximately 32.5 Mbps gain over the MLE at 5 MHz-BW respectively.