Massive MIMO and Small Cells: How to Densify Heterogeneous Networks

We study two time division duplex (TDD) based schemes to enable the coexistence and integrate the benefits of a non-cooperative cellular network overlaid with small cells (SCs). In TDD networks, the uplink and downlink communications are synchronized across both tiers. On the contrary, in reverse TDD (R-TDD) networks, the BS communicates in the downlink while SCs operate in the uplink and vice versa. This leads to different sets of interfering links compared to those of the TDD networks. In particular, the significance of the channel reciprocity offered by TDD based networks is two fold. First, as channel reciprocity can be exploited, the channel estimation overhead is not a function of the number of antennas at the BS. Hence, a large number of transmit antennas can be deployed at the BS. Second, it enables the BS to estimate the covariance of the interference it imposes on SCs during its downlink transmission from its received uplink signal. This paper establishes a framework to exploit this knowledge which leads to implicit coordination across the tiers without introducing any signaling overhead. In this regard, precoding vectors are jointly designed to serve independent data streams for cellular users while being orthogonal to the subspace spanned by the strongest interference directions, thereby minimizing the sum interference enforced on SCs. Our simulation results suggest that both TDD and R-TDD in conjunction with covariance based precoding improve, respectively, the downlink and uplink rates of SCs and provide significant gains in terms of sum achievable rates for individual tiers.