On one-bit quantized ZF precoding for the multiuser massive MIMO downlink

We study low complexity precoding for a downlink massive MIMO multiuser system assuming a base station that employs one-bit digital-to-analog converters (DACs) in order to mitigate power usage. The use of one-bit DACs is equivalent to constraining the transmit signal to be drawn from a QPSK alphabet. While the precoding problem can be formulated using a standard maximum likelihood (ML) encoder, the implementation cost is prohibitive for massive numbers of antennas, even if a sphere encoding approach is used. Instead, we study the performance of a one-bit quantized zero-forcing precoder, and we show that it asymptotically provides the desired downlink vector with low complexity. Simulations show that the quantized ZF precoder can actually outperform the ML encoder for low to moderate signal-to-noise ratios.