Pilot Contamination in Small-Cell Massive MIMO Systems

The increasing demand for wireless data services requires massive network densification which is neither economically nor ecologically feasible. The concept of cell-size reduction provides an effective and simple solution to this problem. Small-Cell Networks (SCN) are one of the promising architectures based on this idea. Generally, a SCN is a cellular network where the size of the cells are very small (compared to the conventional macrocell networks), with self-organizing, low-power, low-cost Base Stations (BS) employed in each cell. In addition to their benefits in the green system design point of view, SCNs are cosidered as the simplest and most effective way to increase the system capacity [1]. Furthermore, It has been shown in [2] that area spectral efficiency increases with reduction of the cell-size in cellular networks. Consequently, cell-size reduction designs has been attracted many interests in the literature [1], [2], and will play an important rule in the next generations of cellular networks. Massive MIMO technique is another approach to increase the system capacity. In multi-cell massive MIMO networks, each BS is equipped with a large number of antennas, which in turn, provide high multiplexing and diversity gains in both uplink and downlink directions. As a result, massive MIMO technique promises to increase the capacity of the network while reducing the transmission power [3]. It has been shown that in massive MIMO systems, the effect of AWGN and small scale fading are eliminated, and also, the simplest linear precoding and detectors perform optimally [4]. Therefore, the only limiting factor on the performance of massive MIMO systems is Pilot Contamination; i.e., the interference between the pilot sequences transmitted by the terminals during the channel estimation. This phenomenon introduces a finite Signal to Inteference Ratio (SIR) to the network, which in turn, imposes an upper-bound on the capacity of the system. Several techniques has been proposed in the literature to remove the effect of pilot contamination and therefore, increase the capacity of the network [5]–[7]. Since pilot contamination is essentially a consequence of linear precoding and detection schemes in massive MIMO systems (which are the most computationally efficient methods), all these proposed techniques introduce non-linear channel estimation methods which are very complicated compared to the linear techniques. On the other hand, as we will see in the next section, the SIR imposed on the system by pilot contamination effect depends only on the large-scale fading (path-loss and shadowing) coefficents between the BS and the terminals within a cell. Then, one question arises here: ” How does cell-size reduction affects the pilot contamination phenomenon?”. In this project we are going to answer this question. Specifically, we are trying to investigate if the cell-size reduction reduces or increases the pilot contamination effect on the performance of massive MIMO systems. Finally, our results will represent one of the following two cases: The cell-size reduction would decrease the effect of pilot contamination and therefore the simple linear precoding and detection schemes can still be employed effectively, and this might be considered as a new advantage of using cell-size reduction designs and SCNs. On the contrast, we may end up with the fact that cell-size reduction increases the effect of pilot contamination (reduces the capacity), and therfore, it is of utmost importance to come up with new channel estimation techniques for massive MIMO SCNs which are able to remove the pilot contamination effect and have low complexity. In the following section, we present our system model and formulate the problem we are going to address.