This paper considers a massive multiple-input-multiple-output (MIMO) systemwith low-resolution analog-to-digital converters (ADCs). In this system,inspired by supervised learning, we propose a novel communication frameworkthat consists of channel training and data detection. The underlying idea ofthe proposed framework is to use the input-output relations of a nonlinearsystem, formed by a channel and a quantization at the ADCs, for data detection.Specifically, for the channel training, we develop implicit and explicittraining methods that empirically learn the conditional probability massfunctions (PMFs) of the nonlinear system. For the data detection, we proposethree detection methods that map a received signal vector to one of the indexesof possible symbol vectors, according to the empirical conditional PMFs learnedfrom the channel training. We also present a low-complexity version of theproposed framework that reduces a detection complexity by using asuccessive-interference-cancellation (SIC) approach. In this low-complexityversion, a symbol vector is divided into two subvectors and then these twosubvectors are successively detected using SIC. When employing the proposedframework with one-bit ADCs, we derive an analytical expression for thesymbol-vector-error probability. One major observation is that thesymbol-vector-error probability decreases exponentially with the inverse of thenumber of transmit antennas, the operating signal-to-noise ratio, and theminimum distance that can increase with the number of receive antennas.Simulations demonstrate the detection error reduction of the proposed frameworkcompared to existing detection techniques.
展开▼
