Exploiting Antenna Correlation in Measured Massive MIMO Channels.

Jose Flordelis,Sha Hu,Fredrik Rusek,Ove Edfors,Ghassan Dahman,Xiang Gao,Fredrik Tufvesson

DOI: https://doi.org/10.1109/pimrc.2016.7794664

2016-01-01

Abstract:We investigate antenna correlation of an M-antenna massive multiple-input multiple-output (MIMO) setup with the purpose of obtaining a low-rank representation of the instantaneous massive MIMO channel. Low-rank representation bases using short-term and long-term antenna correlation statistics are defined, and their performance is evaluated with data sets obtained from channel measurements in both indoor and outdoor environments at 2.6 GHz. Our results indicate that the short-term bases can capture a larger amount of the channel energy compared to the long-term ones, but they have a limited timespan, one coherence time or less. On the other hand, the long-term bases are stable over time-spans of a few seconds. Hence, they can be obtained relatively easily. We also investigate a rank-p vector-scalar LMMSE channel estimator that exploits antenna correlation. Our results show that the investigated estimator can achieve a performance similar to that of full-rank LMMSE at a (2p + 1)/M times lower cost. The investigated estimator may be used in conjunction with estimators that exploit correlation in the frequency and time domains or, alternatively, in situations in which these estimators cannot be used, e.g., when pilot separation is larger than the channel coherence bandwidth or time.