In practice, data gathered by wireless sensor networks often belongs in a low-dimensional subspace, but it can present missing as well as corrupted values due to sensor malfunctioning and/or malicious attacks. We study the problem of Maximum Likelihood estimation of the low-rank factors of the underlying structure in such situation, and develop an Expectation-Maximization algorithm to this purpose, together with an effective initialization scheme. The proposed method outperforms previous schemes based on an initial faulty sensor identification stage, and is competitive in terms of complexity and performance with convex optimization-based matrix completion approaches.

JF - European Signal Processing Conference (EUSIPCO) CY - A Coruña, Spain ER - TY - CONF T1 - Distributed multivariate regression with unknown noise covariance in the presence of outliers: a minimum description length approach T2 - IEEE Workshop on Statistical Signal Processing (SSP) Y1 - 2016 A1 - R. López-Valcarce A1 - Daniel Romero A1 - Josep Sala A1 - Alba Pagès-Zamora KW - compass KW - wsn AB -We consider the problem of estimating the coefficients in a multivariable linear model by means of a wireless sensor network

which may be affected by anomalous measurements. The noise covariance matrices at the different sensors are assumed

unknown. Treating outlying samples, and their support, as additional nuisance parameters, the Maximum Likelihood

estimate is investigated, with the number of outliers being estimated according to the Minimum Description Length

principle. A distributed implementation based on iterative consensus techniques is then proposed, and it is shown effective

for managing outliers in the data.

JF - IEEE Workshop on Statistical Signal Processing (SSP)
CY - Palma de Mallorca, Spain
ER -
TY - JOUR
T1 - Multiantenna GLR detection of rank-one signals with known power spectral shape under spatially uncorrelated noise
JF - IEEE Transactions on Signal Processing
Y1 - 2016
A1 - Josep Sala
A1 - Gonzalo Vázquez-Vilar
A1 - R. López-Valcarce
A1 - Saeid Sedighi
A1 - Abbas Taherpour
KW - cognitive radio
KW - compass
AB - We establish the generalized likelihood ratio (GLR) test for a Gaussian signal of known power spectral shape and unknown rank-one spatial signature in additive white Gaussian noise with an unknown diagonal spatial correlation matrix. This is motivated by spectrum sensing problems in dynamic spectrum access (DSA), in which the temporal correlation of the primary signal can be assumed known up to a scaling, and where the noise is due to an uncalibrated receive array. For spatially independent identically distributed (i.i.d.) noise, the corresponding GLR test reduces to a scalar optimization problem, whereas the GLR detector in the general non-i.i.d. case yields a more involved expression, which can be computed via alternating optimization methods. Low signal-to-noise ratio (SNR) approximations to the detectors are given, together with an asymptotic analysis showing the influence on detection performance of the signal power spectrum and SNR distribution across antennas. Under spatial rank-P conditions, we show that the rank-one GLR detectors are consistent with a statistical criterion that maximizes the output energy of a beamformer operating on filtered data. Simulation results support our theoretical findings in that exploiting prior knowledge on the signal power spectrum can result in significant performance improvement.

VL - 64 IS - 23 ER - TY - JOUR T1 - Multiantenna GLR detection of rank-one signals with known power spectrum in white noise with unknown spatial correlation JF - IEEE Trans. on Signal Processing Y1 - 2012 A1 - Josep Sala A1 - Gonzalo Vázquez-Vilar A1 - R. López-Valcarce KW - cognitive radio KW - dynacs AB -Multiple-antenna detection of a Gaussian signal with spatial rank one in temporally white Gaussian noise with arbitrary and unknown spatial covariance is considered. This is motivated by spectrum sensing problems in the context of Dynamic Spectrum Access in which several secondary networks coexist but do not cooperate, creating a background of spatially correlated broadband interference. When the temporal correlation of the signal of interest is assumed known up to a scale factor, the corresponding Generalized Likelihood Ratio Test is shown to yield a scalar optimization problem. Closed-form expressions of the test are obtained for the general signal spectrum case in the low signal-to-noise ratio (SNR) regime, as well as for signals with binary-valued power spectrum in arbitrary SNR. The two resulting detectors turn out to be equivalent. An asymptotic approximation to the test distribution for the low-SNR regime is derived, closely matching empirical results from spectrum sensing simulation experiments.

VL - 60
IS - 6
ER -
TY - JOUR
T1 - Multiantenna Spectrum Sensing Exploiting Spectral a priori Information
JF - IEEE Trans. on Wireless Communications
Y1 - 2011
A1 - Gonzalo Vázquez-Vilar
A1 - R. López-Valcarce
A1 - Josep Sala
KW - cognitive radio
KW - dynacs
KW - spectrum sensing
VL - 10
IS - 12
ER -
TY - CONF
T1 - Multiantenna spectrum sensing for Cognitive Radio: overcoming noise uncertainty
T2 - International Workshop on Cognitive Information Processing (CIP)
Y1 - 2010
A1 - R. López-Valcarce
A1 - Gonzalo Vázquez-Vilar
A1 - Josep Sala
JF - International Workshop on Cognitive Information Processing (CIP)
CY - Elba Island, Italy
ER -