Enrique V. Carrera

Time synchronization in Arduino-based wireless sensor networks

Wireless sensor networks have attracted a lot of attention due to their wide variety of applications, ranging from inventory management to battlefield surveillance. In most cases, an accurate time synchronization of the nodes is essential to provide a common time reference and to facilitate coordinated actions. However, important challenges arise when network nodes are clocked by low-cost, low-stability oscillators, besides suffering of strong power consumption and computational capacity constraints. In addition, wireless communication protocols used by those simple nodes often have functional and performance disadvantages. Based on that, this paper presents a synchronization technique for Bluetooth-based sensor networks populated by highly resource constrained sensing nodes. In this paper, the main factors influencing time synchronization are characterized and experimentally validated in a real system, including the Bluetooth communication channels. At the end, the proposed synchronization technique is implemented and evaluated in an Arduino-based wireless sensor network, whose motivational application is the localization of acoustic events.

Supervised evaluation of seed-based interactive image segmentation algorithms

Extensive research has been conducted in an effort to evaluate methods and techniques for image segmentation. However, while most literature has focused on evaluating automatic and semi-automatic algorithms, works evaluating interactive segmentation algorithms are less numerous. Note that interactive segmentation can improve results by adding prior knowledge from users into the process. Although this user guidance improves segmentation results, it also makes difficult to conduct objective evaluations. For this reason, some works only present non-canonical evaluations. In this paper, we present an objective and empirical evaluation of seed-based interactive segmentation algorithms. We first compare popular metrics that are employed in image-segmentation evaluations in order to define which one reflects most accurately the performance of segmentation algorithms. Then, in the aim of presenting reproducible results, we introduce a novel seed-based user input dataset that extends the well-known GrabCut dataset. In addition, we evaluate and contrast four state-of-the-art interactive segmentation algorithms. The analysis of the results demonstrates that Jaccard coefficient and Precision-Recall curves provide a good insight into the performance of the evaluated algorithms. Finally, the GrabCut algorithm presents the most robust and useful segmentation among all the evaluated algorithms.

Automatic Recognition of Long Period Events From Volcano Tectonic Earthquakes at Cotopaxi Volcano

Geophysics experts are interested in understanding the behavior of volcanoes and forecasting possible eruptions by monitoring and detecting the increment on volcano-seismic activity, with the aim of safeguarding human lives and material losses. This paper presents an automatic volcanic event detection and classification system, which considers feature extraction and feature selection stages, to reduce the processing time toward a reliable real-time volcano early warning system (RT-VEWS). We built the proposed approach in terms of the seismicity presented in 2009 and 2010 at the Cotopaxi Volcano located in Ecuador. In the detection stage, the recordings were time segmented by using a nonoverlapping 15-s window, and in the classification stage, the detected seismic signals were 1-min long. For each detected signal conveying seismic events, a comprehensive set of statistical, temporal, spectral, and scale-domain features were compiled and extracted, aiming to separate long-period (LP) events from volcano-tectonic (VT) earthquakes. We benchmarked two commonly used types of feature selection techniques, namely, wrapper (recursive feature extraction) and embedded (cross-validation and pruning). Each technique was used within a suitable and appropriate classification algorithm, either the support vector machine (SVM) or the decision trees. The best result was obtained by using the SVM classifier, yielding up to 99% accuracy in the detection stage and 97% accuracy and sensitivity in the event classification stage. Selected features and their interpretation were consistent among different input spaces in simple terms of the spectral content of the frequency bands at 3.1 and 6.8 Hz. A comparative analysis showed that the most relevant features for automatic discrimination between LP and VT events were one in the time domain, five in the frequency domain, and nine in the scale domain. Our study provides the framework for an event classification system with high accuracy and reduced computational requirements, according to the orientation toward a future RT-VEWS.

Automated detection of diabetic retinopathy using SVM

Diabetic retinopathy is a common eye disease in diabetic patients and is the main cause of blindness in the population. Early detection of diabetic retinopathy protects patients from losing their vision. Thus, this paper proposes a computer-assisted diagnosis based on the digital processing of retinal images in order to help people detecting diabetic retinopathy in advance. The main goal is to automatically classify the grade of non-proliferative diabetic retinopathy at any retinal image. For that, an initial image processing stage isolates blood vessels, microaneurysms and hard exudates in order to extract features that can be used by a support vector machine to figure out the retinopathy grade of each retinal image. This proposal has been tested on a database of 400 retinal images labeled according to a 4-grade scale of non-proliferative diabetic retinopathy. As a result, we obtained a maximum sensitivity of 95% and a predictive capacity of 94%. Robustness with respect to changes in the parameters of the algorithm has also been evaluated.

Acoustic event localization on an Arduino-based wireless sensor network

Valuable information from acoustic signals has not been fully considered into real applications associated with assisting people. In fact, the assessment of acoustic signals as an input parameter for localization techniques has useful applications in detecting the presence of living beings trapped in emergency situations by processing any type of sound emitted in face of danger. Indeed, acoustic signals have several technical considerations that make them very attractive, mainly because of their simple acquisition and processing. Thus, this paper presents the design, implementation and evaluation of a prototype that localizes a sound source through the processing of acoustic signal measurements. Concerning the acoustic event localization, two localization techniques are probed within a reverberated indoor environment. Despite the challenges of acoustic signal processing and real-time event localization, the average error obtained in our system is as low as 22 cm inside an area of 2×2 m.

Water Quality Sensing and Spatio-Temporal Monitoring Structure with Autocorrelation Kernel Methods

Pollution on water resources is usually analyzed with monitoring campaigns, which consist of programmed sampling, measurement, and recording of the most representative water quality parameters. These campaign measurements yields a non-uniform spatio-temporal sampled data structure to characterize complex dynamics phenomena. In this work, we propose an enhanced statistical interpolation method to provide water quality managers with statistically interpolated representations of spatial-temporal dynamics. Specifically, our proposal makes efficient use of the a priori available information of the quality parameter measurements through Support Vector Regression (SVR) based on Mercer’s kernels. The methods are benchmarked against previously proposed methods in three segments of the Machángara River and one segment of the San Pedro River in Ecuador, and their different dynamics are shown by statistically interpolated spatial-temporal maps. The best interpolation performance in terms of mean absolute error was the SVR with Mercer’s kernel given by either the Mahalanobis spatial-temporal covariance matrix or by the bivariate estimated autocorrelation function. In particular, the autocorrelation kernel provides with significant improvement of the estimation quality, consistently for all the six water quality variables, which points out the relevance of including a priori knowledge of the problem.

Development of Radial Waveguide Dividers with Large Number of Ports

Power division/combination is a very important function for achieving high-performance power amplification in microwave and millimeter-wave systems, allowing the use of simpler individual amplifiers to provide microwave/millimeter wave high power signals. In this paper we address the development of divider/combiners in waveguide technology for large number of ports. Radial waveguide dividers will be studied, starting from the S-parameters representing the N-port waveguide junction, and ending with the full-wave simulations. Two different designs in this technology are studied. The experimental evaluation will come from a prototype at Ku-band, fabricated and tested in order to validate the design.

A Finger-vein Biometric System Based on Textural Features

Biometric systems are being widely used today for automated authentication purposes. In particular, vascular biometrics or vein recognition is receiving a large amount of attention because of its several advantages related to security and convenience. However, images containing vein patterns normally include more information than just those structural arrangements. Thus, we propose a finger-vein biometric system based exclusively on textural features to evaluate the usefulness of the remaining information around vein patters. Textural features are obtained through gray-level co-occurrence matrices from the wavelet detail coefficients belonging to finger-vein images. The evaluation of the proposed biometric system is based on a standardized finger-vein database and its results show favorable improvements on the finger-vein authentication accuracy when textural features are incorporated in the biometric process.

ECG signal denoising through kernel principal components

Heart electrical activity is measured on the body surface; this measure is known as electrocardiogram (ECG). The ECG signals are commonly accompanied by different types of noise, that can lead to a difficult and imprecise computational process to diagnose heart diseases. In this paper, we propose the Kernel Principal Component Analysis (KPCA) method, usually used in image denoising, for minimizing the noise presented in ECG signals. The aim is to present a high-performance ECG signal denoising process. We also include a denoising benchmark among Discrete Wavelet Transform, Principal Component Analysis, and KPCA methods. Physionet database was used and the accuracy was realized with the Mean Squared Error (MSE) metric. The lower MSE values were obtained in the majority of cases with the KPCA method, considering signals with SNR of 5dB.

Event localization in wireless sensor networks

Wireless sensor networks have important military and civilian applications where determining the geographical location of a specific event is crucial. High-accuracy event localization is a challenging task that depends on the type of event to track. Hence, this work proposes a general approach for achieving event localization in wireless sensor networks, keeping independence from the kind of signals emitted by such events. The proposed approach estimates the distance between an event and a sensor node according to the intensity of the signal perceived by that sensor. In addition, lateration and fingerprinting techniques are implemented to determine the location of particular events. In order to evaluate our proposal, acoustic event localization is implemented in a real wireless sensor network based on the Arduino hardware. Results from this specific implementation show an event localization error of approximately 0.2 m inside a 2×2 m area.