Precoder optimization with local and shared CSI on the K-user MIMO interference channel

Network MIMO provides significant gains compared to transmission schemes based on the interference channel in cellular networks. However, the interference channel is adopted very often in particular due to data sharing limitations of network MIMO. These limitations are due to backhaul links which have finite capacity and also introduce delay. In interference channels, even though data is not exchanged between the transmitters, the global channel state information (CSI) needs to be shared in order to attain the degrees of freedom (DoF) of the channel. This highlights the necessity of reducing the amount of CSI exchange and relying more on local CSI to design the precoders in particular when the local CSI is accurate. In this contribution, cellular downlink transmission using time division duplexing (TDD) is considered where the base stations (BS) can estimate their own channels through reciprocity during the uplink phase. The interfering BSs share their channel estimates via backhaul links of finite capacity. We propose a distributed precoding method based on interference alignment (IA) which is shown to require relatively less CSI exchange compared to other methods.