Journal paper on blind parameter estimation accepted by the IEEE Transactions on Wireless Communications!
The paper focuses on blind parameter estimation and adaptive denoising of sparse signals.
Accurate knowledge of system parameters, such as the average noise power, average signal power, and signal-to-noise ratio (SNR), is critical in wireless communication systems, as many baseband processing tasks rely on these quantities. In this paper, we propose computationally-efficient algorithms to track such parameters in a blind manner, i.e., without the need of pilot signals. Our algorithms exploit the sparsity of channel vectors, considering that in various wireless settings, most of the energy is concentrated to a few entries of large-dimensional vectors. In addition, using Stein’s unbiased risk estimator (SURE), we propose a nonparametric estimate for the mean-square-error (MSE) loss of an arbitrary estimation function. SURE (unlike the MSE) is independent of the unknown signal, and therefore it can be used to tune parameters in a given estimation function, with the objective of optimizing the loss. In particular, we consider a denoising task, as we had done in our previous external page BEACHES algorithm. While BEACHES assumes perfect knowledge of the noise power, our new algorithm uses the proposed noise power estimate. We derive theoretical results for the accuracy of our estimators, and perform simulations in three different wireless scenarios.
The paper was co-authored by Alexandra Gallyas-Sanhueza and Prof. Christoph Studer and will appear in the IEEE Transactions on Wireless Communications. A preprint is available on external page arXiv.