Channel measurements and analysis for very large array systems at 2.6 GHz

Very large MIMO is a technique that potentially can offer large network capacities in multi-user scenarios where the users are equipped only with single antennas. In this paper we are investigating channel properties for a realistic, though somewhat extreme, outdoor base station scenario using a large array. We present measurement results using a 128 element linear array base station and 26 different user position in line-of-sight (LOS) and 10 different user position in non line-of-sight (NLOS). We analyze the Ricean K-factor, received power levels over the array, antenna correlation and eigenvalue distributions. We show that the statistical properties of the received signal vary significantly over the large array. Near field effects and the non-stationarities over the array help decorrelating the channel for different users, thereby providing a favorable channel conditions with stable channels and low interference for the considered single antenna users.

I. Introduction

Very large MIMO, where a base station equipped with a large number of antennas (say 50–400) is serving several simultaneous single antenna users, is a new research field that shows many promising properties. The technology can be an enabler for the next generation power and spectral efficient cellular systems [1] [2]. It has shown very promising performance characteristics based on theoretical studies, but there are many open questions when it comes to practical implementations. One such question is whether the environment and setup provide enough decorrelated channels for different users in a multiuser scenario. The inner product between the propagation vector of different users is of special interest since it in some sense will determine the interference. If we increase the number of antenna elements and this inner product grows slower compared to the propagation vector itself, the interference will eventually approach zero [2]. In this paper we analyze propagation properties for very large MIMO based on measurements. Especially we are interested in answering the following question: What are the specific propagation phenomena that have to be taken into account for large array channel modeling when using a physically large linear array? We specifically target received power levels, singular value distribution, antenna correlation, angular power spectrum, and near field effects. To the authors' best knowledge there are no previous measurement based investigations of channel properties when using physically large arrays in cellular scenarios. Previous investigations using a large circular, but not physically large, antenna array showed a good potential for low correlation and good performance also for less complex linear precoders [5]. With this measurement campaign we are aiming to analyze further channel properties and to identify any differences when going for a very large array that also is physically large.