Reduced Complexity Precoding for One-Bit Signaling

A number of precoding methods have recently been proposed for massive MIMO scenarios that assume the use of one-bit digital-to-analog converters, or that constrain the signal to be of constant modulus. The methods that provide the best performance employ non-linear precoding techniques that require a search for a parameter vector of the same dimension as the number of antennas, which is prohibitively expensive for massive MIMO applications. In this paper, we propose a reduced complexity precoding method based on linear programming and constructive interference that exploits an initial approximate solution to significantly reduce the dimension of the search. In particular, the method uses the output of an initial quantized linear precoder and updates only those elements that do not satisfy a certain reliability condition. Simulations show that the resulting loss in performance due to the use of the simplified approach is insignificant provided that the reduced dimension is chosen to be large enough, but still much smaller than the number of antennas.