Comparing the Performance of Linear Estimators with Massive MIMO Receivers for 5G NR systems

In this work, a comparative analysis is performed to study the bit error rate and error vector magnitude achieved with the least-squares (LS), the minimum mean squared error (MSE), and the Kalman filter (KF) channel estimators when these are applied to the maximum-ratio combining (MRC) and the regularised zero-forcing (RZF) receivers. The MSE achieved with the different channel estimators was also compared by varying the noise and interference power at the receiver. The proposed methodology relies on the characterisation of a massive multiple-input multiple-output (MIMO) channel with a quasi-deterministic radio channel generator and a cyclic prefix orthogonal frequency division multiplexing link-level radio simulation. The fifth-generation (5G) new radio (NR) frame structure was used to perform channel estimation and equalisation for operation frequencies below 6 GHz. Numerical results show that the MRC receiver achieves its maximum performance with the KF estimator, especially at low signal-to-noise ratio scenarios, while the RZF receiver achieves its maximum performance with the LS estimation even in high interference scenarios.