Tensor-Assisted CNN to Estimate Channel in Massive MIMO

In this paper, we propose a machine learning (ML) approach to channel estimation in Massive Multiple Input Multiple Output (MIMO) systems with orthogonal frequency division multiplexing (OFDM) signals. The proposed algorithm employs a sparse tensor signal representation in the angular and delay domains, which utilizes information about Base Station (BS) antenna elements placement. We show that such tensor representation helps to achieve a better performance than matrix signal representation by accounting for information between neighboring tensor samples via feed-forward denoising convolution neural networks (DnCNN).