In this work, we propose a convolutional neural network (CNN) based low-complexity approach for downlink (DL) channel estimation (CE) in frequency division duplex (FDD) systems. In contrast to existing work, we use training data which solely stems from the uplink (UL) domain. This allows to learn the CNN centralized at the base station (BS). After training, the network parameters are offloaded to mobile terminals (MTs) within the coverage area of the BS. The MTs can then obtain channel state information (CSI) of the MIMO channels with the low-complexity CNN estimator. This circumvents the necessity of an infeasible amount of feedback, i.e., acquisition of training data at the user, and the offline training phase at each MT. Numerical results show that the CNN which is trained solely based on UL data performs equally well as the network trained based on DL data. Furthermore, the approach is able to outperform state-of-the-art CE algorithms.
A low-complexity convolutional neural network estimator which learns the minimum mean squared error channel estimator for single-antenna users was recently proposed. We generalize the architecture to the estimation of MIMO channels with multiple-antenna users and incorporate complexity-reducing assumptions based on the channel model. Learning is used in this context to combat the mismatch between the assumptions and real scenarios where the assumptions may not hold. We derive a high-level description of the estimator for arbitrary choices of the pilot sequence. It turns out that the proposed estimator has the implicit structure of a two-layered convolutional neural network, where the derived quantities can be relaxed to learnable parameters. We show that by using discrete Fourier transform based pilots the number of learnable network parameters decreases significantly and the online run time of the estimator is reduced considerably, where we can achieve linearithmic order of complexity in the number of antennas. Numerical results demonstrate performance gains compared to state-of-the-art algorithms from the field of compressive sensing or covariance estimation of the same or even higher computational complexity. The simulation code is available online.