Progressive improvements have been made to the offset-quadratic-amplitude-modulation-based filter-bank multicarrier (FBMC-OQAM) signaling scheme, for example, many theoretical discussions and achievements surrounding its waveform structure, signal processing methods, filter design techniques, and so on. This paper looks at the impact of massive multiple-input multiple-output (MIMO) systems on FBMC-OQAM. The mutual influence is definitely an interesting matter.

This paper mostly covers theoretical characteristics, with performance analysis discussions via the output mean squared error (MSE) of the estimated transmitted symbols in an uplink massive MIMO system. It presents a good review of massive MIMO FBMC-OQAM systems and introduces three linear receivers: zero forcer (ZF), linear minimum mean squared error (LMMSE), and matched filter (MF). Using random matrix theory, the authors also constitute a framework with the following parameters: “the number of [base station, BS] antennas N, the number of [single antenna] users K, the spatial correlation between the BS antennas, small-scale and large-scale fading, and the different [power delay profile, PDP] of each user.” With a more practical and realistic view, where N is finite and not much larger than K, the framework details the bounds of signal-to-interference-plus-noise ratio (SINR) for asymptotic MSE expressions.

For the system model, the authors describe classical approximation, where “the channel is assumed to remain constant over the frame duration and frequency flat at the subcarrier level.” However, in practical scenarios, at channels with high frequency selectivity, inter-user interference (IUI), inter-carrier interference (ICI), inter-symbol interference (ISI), and nonperfect inversions of prototype functions are probable. So, based on MSE approximation and new channel modeling, a new comprehensive system model is devised and the deteriorating factors (IUI, ICI, and ISI) are extracted. Accordingly, performance analyses of the ZF, LMMSE, and MF receivers, via the asymptotic behavior of the MSE, are explained.

The paper explains: “if the users are well synchronized, the different terms that compose the MSE ... become negligible for large values of the ratio N/K”; this “self-equalization” has remarkable influence on system performance. The paper includes insightful examples interwoven with both practical and theoretical discussions. It sheds some light on the hidden corners of massive MIMO applications with tangible results.