Massive MIMO Pilot Assignment Optimization based on Total Capacity

Abstract

We investigate the effects of pilot assignment in multi-cell massive multiple-input multiple-output systems. When deploying a large number of antennas at base station (BS), and linear detection/precoding algorithms, the system performance in both uplink (UL) and downlink (DL) is mainly limited by pilot contamination. This interference is proper of each pilot, and thus system performance can be improved by suitably assigning the pilot sequences to the users within the cell, according to the desired metric. We show in this paper that UL and DL performances constitute conflicting metrics, in such a way that one cannot achieve the best performance in UL and DL with a single pilot assignment configuration. Thus, we propose an alternative metric, namely total capacity, aiming to simultaneously achieve a suitable performance in both links. Since the PA problem is combinatorial, and the search space grows with the number of pilots in a factorial fashion, we also propose a low complexity suboptimal algorithm that achieves promising capacity performance avoiding the exhaustive search. Besides, the combination of our proposed PA schemes with an efficient power control algorithm unveils the great potential of the proposed techniques in providing improved performance for a higher number of users. Our numerical results demonstrate that with 64 BS antennas serving 10 users, our proposed method can assure a 95%-likely rate of 4.2 Mbps for both DL and UL, and a symmetric 95%-likely rate of 1.4 Mbps when serving 32 users.