This repository provides an estimator for the problem of recovering the channel vector
Using the estimator is a two-step process.
Training data
Step 1 only needs to be done once and it can be done in a preparatory offline training phase.
Assuming that
The GMM estimator
For more details, kindly take a look at the first two references below.
The code is written in Python.
Required packages are numpy
, scikit-learn
, and scipy
.
To generate plots, matplotlib
is required as well.
The file examples.py demonstrates the use of the GMM estimator in various settings. Invoking
python example.py
runs all examples. Alternatively,
python example.py --nr n
runs example
- The observation matrix
$\mathbf{A} = \mathbf{I}$ is the identity matrix and full GMM covariance matrices are used. - The observation matrix is a selection matrix and full GMM covariance matrices are used.
This repository is joint work of Michael Koller and Benedikt Fesl.
The implementation makes use of Benedikt Fesl's repository GMM_cplx which allows fitting a GMM with complex-valued quantities.
The following reference provides more details and properties of the GMM estimator.
- Koller, Fesl, Turan, Utschick, "An Asymptotically MSE-Optimal Estimator Based on Gaussian Mixture Models," IEEE Trans. Signal. Process., 2022. [IEEEXplore] [arXiv]
The estimator has been used in the following references.
- N. Turan, B. Fesl, M. Grundei, M. Koller, and W. Utschick, “Evaluation of a Gaussian Mixture Model-based Channel Estimator using Measurement Data,” in Int. Symp. Wireless Commun. Syst. (ISWCS), 2022. [IEEEXplore]
- B. Fesl, M. Joham, S. Hu, M. Koller, N. Turan, and W. Utschick, “Channel Estimation based on Gaussian Mixture Models with Structured Covariances,” in 56th Asilomar Conf. Signals, Syst., Comput., 2022, pp. 533–537. [IEEEXplore]
- B. Fesl, N. Turan, M. Joham, and W. Utschick, “Learning a Gaussian Mixture Model from Imperfect Training Data for Robust Channel Estimation,” IEEE Wireless Commun. Lett., 2023. [IEEEXplore]
- M. Koller, B. Fesl, N. Turan and W. Utschick, "An Asymptotically Optimal Approximation of the Conditional Mean Channel Estimator Based on Gaussian Mixture Models," IEEE Int. Conf. Acoust., Speech, Signal Process. (ICASSP), 2022, pp. 5268-5272. [IEEEXplore] [arXiv]
- B. Fesl, A. Faika, N. Turan, M. Joham, and W. Utschick, “Channel Estimation with Reduced Phase Allocations in RIS-Aided Systems,” in IEEE 24th Int. Workshop Signal Process. Adv. Wireless Commun. (SPAWC), 2023. [arXiv]
- N. Turan, B. Fesl, M. Koller, M. Joham, and W. Utschick, “A Versatile Low-Complexity Feedback Scheme for FDD Systems via Generative Modeling,” 2023, arXiv preprint: 2304.14373. [arXiv]
- N. Turan, B. Fesl, and W. Utschick, "Enhanced Low-Complexity FDD System Feedback with Variable Bit Lengths via Generative Modeling," in 57th Asilomar Conf. Signals, Syst., Comput., 2023. [arXiv]
- N. Turan, M. Koller, B. Fesl, S. Bazzi, W. Xu and W. Utschick, "GMM-based Codebook Construction and Feedback Encoding in FDD Systems,"in 56th Asilomar Conf. Signals, Syst., Comput., 2022, pp. 37-42. [IEEEXplore]