Research: Communication & Signal Processing Group

Communications & Signal Processing Research Group

Welcome to the Communications, Networking, Image and Signal Processing Research Group, School of Engineering, Department of Computing, Electronics, and Mechatronics at the Universidad de las Americas Puebla, Mexico.  The group currently has about 5 members who have research interests in several communications, networking, signal, and image processing areas. We currently focus on the following research areas:

If you are interested in joining us as a postgraduate student or research fellow, please contact Prof. Alarcon-Aquino, Email: vicente dot alarcon at udlap dot mx

 (A beautiful view of campus, UDLAP) 

Check this out: UDLAP Gallery

Research Projects

Multiresolution Analysis for Transiting Exoplanet Identification Using Machine Learning

Collaborators: M. A. Jara-Maldonado (UDLAP, Mexico), V. Alarcon-Aquino (UDLAP, Mexico), R. Rosas-Romero (UDLAP, Mexico) 

An exoplanet (short for the extrasolar planet) is a planet that orbits a star different than our Sun. The task of identifying exoplanets requires looking for transits in light flux time series data. Transits are similar phenomena to eclipses, but instead of the moon being the one that dims the light received from the star, it is the exoplanet that crosses between the star and the observer. Manually searching for transits in light flux-related data (called light curves) consumes huge amounts of time and effort from astronomers, because of the size of the databases that have been generated by satellites and ground-based observatories. For this reason, the application of machine learning techniques to exoplanet identification has been developing in the last few years.

Source URL http://www.pitt.edu/~stepup/

Anomaly Detection Model based on Immune Danger Theory applied to Computer Networks

Collaborators: David Limon-Cantu (Ph.D. Student), V. Alarcon-Aquino (UDLAP, Mexico) 

Currently, David is centering his research in Evolutionary Computing Algorithms inspired by the Human Immune System and the Danger Theory (DT) model. The Dendritic Cell Algorithm (DCA) is a population-based binary classifier designed to detect anomalies and is inspired by the behavior of Dendritic Cells (DC), which are able to sense Danger Signals (DS) emitted by abnormal behavior of the tissues and cells of the body. The objective of Cantu's research is to enhance this algorithm and incorporate state-of-the-art techniques in order to increase its precision, adaptability, and scalability using computer network communications to influence security applications in the anomaly detection field, which aims to ultimately help detect unknown or dynamic threats in computer networks.

Source: Z. Chelly and Z. Elouedi, “A survey of the dendritic cell algorithm.,” Knowledge & Information Systems, vol. 48, no. 3, pp. 505–535, Sep. 2016.

Object Recognition using Multiresolution Analysis in Infrared Imaging for Defense Systems

Collaborators: Daniel Treviño-Sanchez (PhD Student), V. Alarcon-Aquino (UDLAP, Mexico) 

Daniel is currently working on recognizing vehicles, drones, and people (among other things) using infrared images. This is a sensitive topic for national defense systems, especially achieving this goal in total darkness. Since improving national defense systems against new threats has already become a major concern that requires state-of-the-art technology to fill some gaps and address vulnerabilities. Some key components for such tasks are Infrared technology (IR), which enables vision in many adverse circumstances.  This research is focused on Multiresolution techniques to extract the most relevant features with good accuracy and low computational cost from those images, and by combining the use of some advanced methods to enhance edges,  identify shapes and classify objects, and recognize those objects.

Models and Algorithms for Video Compression Using Wavelet Transforms and Fovea

Collaborators: J. C. Galan-Hernandez (UDLAP, Mexico), V. Alarcon-Aquino (UDLAP, Mexico), T. Stathaki (Imperial College London, UK) 

Digital image and video stores amounts of data that can be used for solving several computer vision problems. However, this is a challenging goal primarily because a large amount of data makes the task computational very intensive. The subject of video coding is of paramount importance for video processing since it can be of great impact on data transmission and manipulation. Video transmission can be greatly improved when a video compression algorithm is used reducing the communication overhead. With efficient coding, distortions can be assessed and controlled, bandwidth can be managed efficiently, and transmission time reduced. The structure of the human eye can be exploited for compression. The human eye experiments a form of aliasing from the fixation point to the edges of the image. This aliasing is exploited in fovea compression. Fovea compression can be applied over images with ROIs, the use of fovea around ROIs improves the image quality for the human eye. This research focuses on image and video compression using wavelet transforms and the fovea of the human eye for increasing the quality of the compressed image and video.

Epilepsy Seizures Classification in EEG Signals Using Wavelet-Based Neural Networks

Collaborators: E. Juarez-Guerra (PhD Student), V. Alarcon-Aquino (Universidad de las Americas Puebla, Mexico), P. Gomez-Gil (INAOE)  

The brain is one of the most important organs of the human body, controlling the coordination of the human muscles and nerves. The transient and unexpected electrical disturbances of the brain result in an acute disease called Epileptic seizures. These seizures are seen as a sudden abnormal function of the body, often with loss of consciousness, an increase in muscular activity, or an abnormal sensation. The human brain is obviously a complex system and exhibits rich spatiotemporal dynamics. Among the noninvasive techniques for probing human brain dynamics, electroencephalography provides a direct measure of cortical activity with millisecond temporal resolution. Electroencephalogram (EEG) signals involve a great deal of information about the function of the brain. But evaluation and classification of these signals are limited and since there is no definite criterion evaluated by the experts, visual analysis of EEG signals in the time domain may be insufficient. Routine clinical diagnosis needs the analysis of EEG signals. An important issue in epileptology is the question of whether epileptic seizures can be anticipated prior to their occurrence. The question of whether information extracted from the EEG of epilepsy patients can be used for the classification and prediction of seizures has recently attracted much attention. In this research, we focus on a study of biomedical signals to analyze, detect, classify and predict abnormalities such as Epileptic seizures, brain tumors, etc. using signal processing techniques such as wavelet neural networks and empirical mode decomposition.

Forecast, Localization, and Detection of Wildfires Using Multi-Resolution Optimization

Collaborators: M. Diaz-Romero (PhD Student), V. Alarcon-Aquino (Universidad de las Americas Puebla, Mexico) 

Regardless of its origin, a wildfire is an imminent danger or harm for people, property, weather, morphology, soils, economy, society, and the environment. It propagates without control in rural areas, through bushes and woody vegetation, alive and dead. The prevailing weather conditions, like high temperatures, low humidity, and wind, put an area in a moment of maximum probability of occurrence of wildfires. Most of the wildfires are caused by men or by elements like machines and facilities built by men. Wildfires management has been seen in the development of several analysis methodologies based on operation research techniques that serve as tools for the creation of policies. The goal of controlling wildfires is to minimize the negative impact they cause. This can be achieved through detection and attack of fires soon after it is informed or through effective control of fires. The overall objectives of this research are to create a model to maximize the weighted sum of demand points that are covered in a protected area given its different classes, considering the number of available resources, and the arrival time limit to extinguish the fire (which varies for each zone). The model used is a programming model that determines the resource deployment in such a way that any fire can be attacked within a certain time limit. The deployed resources must cover the largest geographical area that was previously classified for its level of importance. The problem of maximum coverage localization is known to be NP-Hard so there is no warranty that for bigger instances of the problem an optimal solution can always be found. Therefore, to find solutions in a reasonable time for big instances the design of heuristics is required. Here, we propose a new method using the multiresolution solution in the local search.

Digital Processing of EEG Signals oriented to the Development of BCIs

Collaborators: G. Rosas-Cholula, J. M. Ramirez-Cortes (INAOE), V. Alarcon-Aquino (Universidad de las Americas Puebla) 

Brain-Computer Interfaces (BCIs) are systems that allow people to control computer applications using their brain signals. In this research, we focus on the detection of a P-300 rhythm for potential applications on BCI using an Adaptive Neuro-Fuzzy algorithm (ANFIS). P300 evoked potential is an electroencephalographic (EEG) signal obtained at the central-parietal region of the brain in response to rare or unexpected events. The P300 evoked potential is obtained from visual stimuli followed by a motor response from the subject. The EEG signals are obtained with a 14 electrodes Emotiv EPOC headset. Preprocessing of the signals includes denoising and blind source separation using an Independent Component Analysis algorithm. The P300 rhythm is detected using the discrete wavelet transform (DWT) applied on the preprocessed signal as a feature extractor, and further entered into the ANFIS system. 

Security in Communications Networks Using Biometric Systems

Collaborators: H. A. Garcia-Baleon (PhD Student at Imperial College London, UK), V. Alarcon-Aquino (Universidad de las Americas Puebla) 

Combining biometrics and cryptography is becoming a matter of interest for some researchers due to the fact that this combination can bring together the better of the two worlds. The former guarantees the identification of individuals based on measuring their personal unique features with a high degree of assurance, while the latter mainly assures a high degree of trust in the transactions of information over non-secure communications networks. In this research, we focus on a bimodal biometric system for a cryptographic key generation that works with speech and electrocardiogram (ECG) signals using wavelet transforms. This work is based on the uniqueness and quasi-stationary behavior of ECG and speech signals with respect to an individual. The architecture of the proposed system considers three security factors, namely, user password, biometric samples, and a token. The stages that comprise the architecture are one-time enrollment and key derivation. The system architecture is able to verify the identity of individuals offline avoiding the use of a centralized database for storing the biometric information. The system also implements an error-correction layer using the Hadamard code. The performance of the system is assessed using ECG signals from the MIT-BIH database and speech signals from a speech database created for testing purposes. The random cryptographic key released by the system may be used in several encryption algorithms.

Anomaly Detection and Prediction in Communication Networks Using Wavelet Transforms

Collaborators: V. Alarcon-Aquino (UDLAP, Mexico), J. A. Barria (Imperial College London, UK)

It is important for service providers to monitor their systems in order to detect network anomalies and performance degradations in advance of network/service disruptions. In this regard, several anomaly detection schemes have already been proposed in the literature. These schemes are in most cases based on parametric models and thresholding techniques. The underlying aim of this research is to develop models and algorithms based on wavelet transforms for analysing the statistical behaviour of network metrics in order to detect and predict network anomalies and performance degradations in communication networks. A novel wavelet-based algorithm is proposed for detecting network anomalies in communication networks. The wavelet-based algorithm is then used to detect events in different network metrics of a Dial Internet Protocol service and corporate Proxy servers. A sensor fusion scheme, which combines local decisions made from dispersed wavelet-based sensors, is also investigated. This sensor fusion scheme incorporates the spatial dependencies among the monitored network metrics and hence reduces the number of false alarms generated by each network metric. Furthermore, a novel learning algorithm is investigated for time series prediction based on finite impulse response (FIR) neural networks and the multiresolution analysis framework of wavelet theory. A gradient descent method is used to adapt the gain of the non-linear functions in FIR networks at each level of resolution. The multi-resolution learning algorithm is compared with previously reported algorithms using a benchmark time series. The algorithm is also applied to network traffic prediction in an Ethernet environment. The results show that the generalisation ability of the FIR network is improved by the multiresolution learning algorithm. A method is also investigated for predicting and monitoring communication network metrics. The proposed method learns to predict the normal behaviour of the monitored network metric and together with an online decision-making algorithm detects and classify deviations from the normal operating region. The proposed method is used to predict and monitor events incorporate Proxy servers and Local/Wide area network traces. Experimental results show that the proposed method is able to identify moderate and severe abnormal network behaviours in advance of reported network faults and thereby providing a useful method for proactively managing communication networks.