Abstract - The Massive-MIMO method has become popular to increase the data rates for communications based on the channel hardening phenomenon, by reducing statistical fluctuations in fading channels. However, complications occur when not considering the traditional i.i.d. Rayleigh channel. A Massive-MIMO OFDM system is tested comparing least square channel estimation for Rayleigh, Rician and Correlation channel models for Matched Filter, Zero Forcing and Minimum Mean Square Error linear detection schemes. 16, 64 and 128-QAM is compared with 64, 256 and 1024 OFDM subcarriers on Massive-MIMO channel of 50 single transmitter antennas to 50, 100, 200 and 300 receive antennas. Results show that Massive-MIMO linear detection methods have significant performance improvement at higher OFDM subcarriers but difficult to overcome performance gains in Rician and Correlation models.
This project provides a good tutorial with MATLAB code for performing Massive-MIMO-OFDM channel modeling, least squares channel estimation and MIMO detection in a wireless downlink system.