This paper addresses the problem of estimating the Potts parameter  jointly with the unknown parameters of a Bayesian model within a Markov chain Monte Carlo (MCMC) algorithm. Standard MCMC methods cannot be applied to this problem because performing inference on  requires computing the intractable normalizing constant of the Potts model. In the proposed MCMC method, the estimation of  is conducted using a likelihood-free Metropolis–Hastings algorithm. Experimental results obtained for synthetic data show that estimating  jointly with the other unknown parameters leads to estimation results that are as good as those obtained with the actual value of . On the other hand, choosing an incorrect value of  can degrade estimation performance significantly. To illustrate the interest of this method, the proposed algorithm is successfully applied to real bidimensional SAR and tridimensional ultrasound images.

Spectral analysis along with the detection of harmonics and modulation sidebands are key elements in condition monitoring systems. Several spectral analysis tools are already able to detect spectral components present in a signal. The challenge is therefore to complete this spectral analysis with a method able to identify harmonic series and modulation sidebands. Compared to the state of the art, the method proposed takes the uncertainty of the frequency estimation into account. The identification is automatically done without any a priori, the search of harmonics is exhaustive and moreover the identification of all the modulation sidebands of each harmonic is done regardless of their energy level. The identified series are characterized by criteria which reflect their relevance and which allow the association of series in families, characteristic of a same physical process. This method is applied on real-world current and vibration data, more or less rich in their spectral content. The identification of sidebands is a strong indicator of failures in mechanical systems. The detection and tracking of these modulations from a very low energy level is an asset for earlier detection of the failure. The proposed method is validated by comparison with expert diagnosis in the concerned fields.

Vibration data acquired during system monitoring periods are rich in harmonics characterizing the presence of several mechanical parts in the system. Periodic variations of the torque or of the load create modulation sidebands around those harmonics. Even if the energy impact of the sidebands is small compared to the total energy of the signal, they are strong indicators of failures in mechanical systems. Unfortunately, these effects are of little concern in most condition monitoring systems. When considering the problem from a signal processing point of view, the demodulation of those sidebands allows for a time visualization of the modulating functions which are a precise image of the torque or the load variations. This demodulation can be done on the analytical signal directly derived from the original data. But to do that, data and specifically its spectrum should respect some constraints. The purpose of this paper is to underline those often neglected constraints. In particular, the respect of the non-overlapping condition in the Bedrosian theorem is discussed for signals and modulation rates that can be encountered on rotating machines. The respect of the constraints depends on the monitored phenomenon (e.g., gear mesh, rotating shaft), the modulation phenomenon (e.g., belt frequency, rotor current) and the type of medium (e.g., vibrations, electrical current). In the case where the constraints are not satisfied, we explain the consequences in terms of signal processing. These results are illustrated by an industrial case study.

Discontinuities at origin have been used to better approximate measured curves in recent papers but generally not explicitly and their physical validity has not always been demonstrated. In this communication, we show that these discontinuities can be explained by physically acceptable discontinuities in the real physical device. We propose simple criteria to accept or reject these discontinuities, in either passive or active devices, depending on the order of the discontinuity. In addition, we show that models having such discontinuities behave differently from classical models. In particular, these discontinuities explain non-integer dB/dB slopes of harmonic power and intermodulation power as a function of input power. Recent and older measurements of intermodulation products in passive devices, telephony base-station and RF transistors show such a behavior so that supposed lack of measurement cannot be used as a reason to reject discontinuities as non-physical.

Broadcasting systems have to deal with channel variability in order to offer the best spectrum efficiency to the receivers. However, the transmission parameters that optimize the spectrum efficiency generally leads to a large link unavailability. In this paper, we study the performance of hierarchical and non-hierarchical modulations in terms of spectrum efficiency and link unavailability for DVB-S2 systems. Our first contribution is the design of the hierarchical 16-APSK for the DVB-S2 standard. Then we introduce the link unavailability to compare the per- formance of hierarchical and non-hierarchical modulations in terms of spectrum efficiency and link unavailability. The results show that hierarchical modulation is a good alternative to non-hierarchical modulation for the DVB-S2 standard.

Content distribution over ad-hoc networks has been widely studied and numerous solutions can be adapted to VANETs (vehicular networks). VANETs, however, can also benefit from an infrastructure in order to improve the efficiency of any content dissemination technique, while allowing more transmissions resources to be available for safety applications. Unfortunately, so far, no such solution has been proposed. In this paper, we introduce the use of an 802.11p infrastructure based on Road Side Units (RSU) for downloading data (eg a map) to vehicles on a highway. At the application level, the main challenge is then how to deliver data to a large number of moving receivers with limited connectivity. While broadcasting data through each RSU should efficiently provide the cars with most of the data, one could believe that some specific transmissions based on vehicles needs could help to reach full downloads. Using simulations, we observed however that the best performances are achieved by a pure broadcasting system.

The kernel least-mean-square (KLMS) algorithm is a popular algorithm in nonlinear adaptive filtering due to its simplicity and robustness. In kernel adaptive filters, the statistics of the input to the linear filter depends on the parameters of the kernel employed. Moreover, practical implementations require a finite nonlinearity model order. A Gaussian KLMS has two design parameters, the step size and the Gaussian kernel bandwidth. Thus, its design requires analytical models for the algorithm behavior as a function of these two parameters. This paper studies the steady-state behavior and the transient behavior of the Gaussian KLMS algorithm for Gaussian inputs and a finite order nonlinearity model. In particular, we derive recursive expressions for the mean-weight-error vector and the mean-square-error. The model predictions show excellent agreement with Monte Carlo simulations in transient and steady state. This allows the explicit analytical determination of stability limits, and gives opportunity to choose the algorithm parameters a priori in order to achieve prescribed convergence speed and quality of the estimate. Design examples are presented which validate the theoretical analysis and illustrates its application.

This paper presents an unsupervised algorithm for nonlinear unmixing of hyperspectral images. The proposed model assumes that the pixel reßectances result froma nonlinear function of the abundance vectors associated with the pure spectral components. We assume that the spectral signatures of the pure components and the nonlinear function are unknown. The Þrst step of the proposedmethod estimates the abundance vectors for all the image pixels using a Bayesian approach an aGaussian process latent variable model for the nonlinear function (relating the abundance vectors to the observations). The endmembers are subsequently estimated using Gaussian process regression. The performance of the unmixing strategy is Þrst evaluated on synthetic data. The proposed method provides accurate abundance and endmember estimations when compared to other linear and nonlinear unmixing strategies. An interesting property is its robustness to the absence of pure pixels in the image. The analysis of a real hyperspectral image shows results that are in good agreement with state of the art unmixing strategies and with a recent classiÞcation method.

Periodic nonuniform sampling of order three (PNS3) is a sampling scheme composed of three periodic sequences with the same period. It is well-known that this sampling scheme can be useful to remove aliasing. Previous studies have provided conditions on the spectrum support for exact reconstruction in the case of functions. This paper deals more generally with the best mean-square interpolation for stationary processes with any known power spectrum, from PNS3 and possibly aliasing. We show that the best estimation is based upon particular linear filters, which depend on the gap between the sampling sequences. The mean-time error also depends on this gap. The errorless interpolation is a particular case. It requires the knowledge of the spectral support rather than the spectral true values.

Multifractal analysis has matured into a widely used signal and image processing tool. Due to the statistical nature of multifractal processes (strongly non-Gaussian and intricate dependence) the accurate estimation of multifractal parameters is very challenging in situations where the sample size is small (notably including a range of biomedical applications) and currently available estimators need to be improved. To overcome such limitations, the present contribution proposes a Bayesian estimation procedure for the multifractality (or intermittence) parameter. Its originality is threefold : First, the use of wavelet leaders, a recently introduced multiresolution quantity that has been shown to yield significant benefits for multifractal analysis ; Second, the construction of a simple yet generic semi-parametric model for the marginals and covariance structure of wavelet leaders for the large class of multiplicative cascade based multifractal processes ; Third, the construction of original Bayesian estimators associated with the model and the constraints imposed by multifractal theory. Performance are numerically assessed and illustrated for synthetic multifractal processes for a range of multifractal parameter values. The proposed procedure yields significantly improved estimation performance for small sample sizes.