Spectral Efficiency of Dense Multicell Massive MIMO Networks in Spatially Correlated Channels

This paper is on the spectral efficiency (SE) of a dense multi-cell massive multiple-input multiple-output (MIMO). The channels are spatially correlated and the multi-slope path loss model is considered. In our framework, the channel state information is obtained by using pilot sequences and the BSs are deployed randomly. First, we study the channel estimation accuracy and its impact on the SE as the BS density increases and the network becomes densified. Second, we consider the special case of uncorrelated channels for which the stochastic geometry framework helps us to simplify the SE expressions, and obtain the minimum value of antenna-UE ratio over which the pilot contamination is dominant rather than the inter- and intra-cell interference. Finally, we provide some insights into the obtained SE for the spatially correlated channels, from a multi-cell processing scheme as well as the single-cell ones in terms of the BS density. Our results show that while all the detectors result in non-increasing SE in terms of the BS density, their area SE increases exponentially as the network becomes densified. Moreover, we conclude that in order to achieve a given SE, the required value of antenna-UE ratio decreases as the level of channel correlation increases.