We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Electrical Engineering and Systems Science > Signal Processing

Title: Learning to Estimate RIS-Aided mmWave Channels

Abstract: Inspired by the remarkable learning and prediction performance of deep neural networks (DNNs), we apply one special type of DNN framework, known as model-driven deep unfolding neural network, to reconfigurable intelligent surface (RIS)-aided millimeter wave (mmWave) single-input multiple-output (SIMO) systems. We focus on uplink cascaded channel estimation, where known and fixed base station combining and RIS phase control matrices are considered for collecting observations. To boost the estimation performance and reduce the training overhead, the inherent channel sparsity of mmWave channels is leveraged in the deep unfolding method. It is verified that the proposed deep unfolding network architecture can outperform the least squares (LS) method with a relatively smaller training overhead and online computational complexity.
Comments: 5 pages, 7 figures, submitted to IEEE WCL (under second round review)
Subjects: Signal Processing (eess.SP); Machine Learning (cs.LG)
Cite as: arXiv:2107.12631 [eess.SP]
  (or arXiv:2107.12631v2 [eess.SP] for this version)

Submission history

From: Jiguang He [view email]
[v1] Tue, 27 Jul 2021 06:57:56 GMT (564kb)
[v2] Fri, 3 Dec 2021 03:22:13 GMT (570kb)

Link back to: arXiv, form interface, contact.