Eduardo Freire Nakamura

An Efficient Directed Localization Recursion Protocol for Wireless Sensor Networks

The establishment of a localization system is an important task in wireless sensor networks. Due to the geographical correlation between sensed data, location information is commonly used to name the gathered data and address nodes and regions in data dissemination protocols. In general, to estimate its location, a node needs the position information of at least three reference points (neighbors that know their positions). In this work, we propose a different scheme in which only two reference points are required in order to estimate a position. To choose between the two possible solutions of an estimate, we use the known direction of the recursion. This approach leads to a recursive localization system that works with low-density networks (increasing by 40 percent the number of nodes with estimates in some cases), reduces the position error by almost 30 percent, requires 37 percent less processor resources to estimate a position, uses fewer beacon nodes, and also indicates the node position error based on its distance to the recursion origin. No GPS-enabled node is required, since the recursion origin can be used as a relative coordinate system. The algorithms evaluation is performed by comparing it with a similar localization system; also, experiments are made to evaluate the impact of both systems in geographic algorithms.

An incremental technique for real-time bioacoustic signal segmentation

An incremental transformation of ZCR and energy without using temporal windows.With our method is possible to save memory and transmission costs.Solution to process large amounts of data by resource-constrained devices as WSN. A bioacoustical animal recognition system is composed of two parts: (1) the segmenter, responsible for detecting syllables (animal vocalization) in the audio; and (2) the classifier, which determines the species/animal whose the syllables belong to. In this work, we first present a novel technique for automatic segmentation of anuran calls in real time; then, we present a method to assess the performance of the whole system. The proposed segmentation method performs an unsupervised binary classification of time series (audio) that incrementally computes two exponentially-weighted features (Energy and Zero Crossing Rate). In our proposal, classical sliding temporal windows are replaced with counters that give higher weights to new data, allowing us to distinguish between a syllable and ambient noise (considered as silences). Compared to sliding-window approaches, the associated memory cost of our proposal is lower, and processing speed is higher. Our evaluation of the segmentation component considers three metrics: (1) the Matthews Correlation Coefficient for point-to-point comparison; (2) the WinPR to quantify the precision of boundaries; and (3) the AEER for event-to-event counting. The experiments were carried out in a dataset with 896 syllables of seven different species of anurans. To evaluate the whole system, we derived four equations that helps understand the impact that the precision and recall of the segmentation component has on the classification task. Finally, our experiments show a segmentation/recognition improvement of 37%, while reducing memory and data communication. Therefore, results suggest that our proposal is suitable for resource-constrained systems, such as Wireless Sensor Networks (WSNs).

Target Tracking for Sensor Networks: A Survey

Target-tracking algorithms typically organize the network into a logical structure (e.g., tree, cluster, or faces) to enable data fusion and reduce communication costs. These algorithms often predict the target’s future position. In addition to using position forecasts for decision making, we can also use such information to save energy by activating only the set of sensors nearby the target’s trajectory. In this work, we survey of the state of the art of target-tracking techniques in sensor networks. We identify three different formulations for the target-tracking problem and classify the target-tracking algorithms based on common characteristics. Furthermore, for the sake of a better understanding of the target-tracking process, we organize this process in six components: target detection, node cooperation, position computation, future-position estimation, energy management, and target recovery. Each component has different solutions that affect the target-tracking performance.

A prediction-based clustering algorithm for tracking targets in quantized areas for wireless sensor networks

AbstractnTarget tracking is an important application of sensor networks, particularly interesting for ecology applications related to wildlife monitoring. nIn this context, understanding the territorial occupation of animals is fundamental for understanding their habits. In this work, we propose the PRATIQUE—a prediction-based clustering algorithm for tracking targets considering a discrete sensor field divided into cells. This approach is based on two hierarchical levels: static clusters at the first level and dynamic clusters at the second level. This hybrid scheme reduces the cost of communication and ensures that all data generated by an event be delivered to a single node. We use Kalman, Alpha-Beta, or Particle Filters in order to predict the target’s position. Prediction is used to prepare the set of nodes that will detect the next event, thereby reducing the message overhead during the tracking task. Results show that prediction errors are close to one cell.

Cloud-assisted Computing for Event-driven Mobile Services

Today, software developers for desktop computing build request and respond applications to do what end users tell them to do and answer what they ask. In mobile computing, software developers will need to develop sense and response applications that will interact with the end user. These applications will notify or ask users what they want based on what they have sensed or on a personal profile. Mobile cloud computing has the potential to empower mobile users with capabilities not found in mobile devices, combining different and heterogeneous data sets. In this work, we discuss the importance and challenges in designing event-driven mobile services that will detect conditions of interest to users and notify them accordingly.

A coverage-based drop-policy in wireless sensor network with disruptive connections

Many applications in Wireless Sensor Networks (WSNs) consider remote and large scale monitoring. For those scenarios, the whole network is difficultly fully connected all the time. A possible vision is the union of WSNs and Disruptive Tolerant Network(DTNs) concepts, where mobile nodes make collect data in sparse networks and deliver them to a base station. This work presents a buffer management strategy, which is a basic principle in DTN networks. The proposed solution considers the global coverage to choose which messages are dropped, therefore, minimizing the impact on monitoring application. Such solutions are important for WSNs applications with limited resources. We show through simulation that the proposed Coverage-Based Drop-Policy in Wireless Sensor Network with Disruptive Connections (CBDP) improves the network coverage compared to traditional DTN drop policies like Drop Last Packet (DL) and Drop First Packet (DF).

For or Against?: Polarity Analysis in Tweets about Impeachment Process of Brazil President

Social networks define a major media for users to express their opinion regarding different matters. As a consequence, these networks naturally provide information that allow us to detect user behaviors, opinions, and sentiments about diverse events around the world. One event that called attention in Brazil is the impeachment process of the Brazillian President. The goal of this paper is to infer and characterize the opinion (polarity) of the Brazilians about the impeachment process in Brazil. We used a supervised learning approach and compared three classifiers: Max Entropy, Support Vector Machine (SVM), and Multinomial Naive Bayes. The SVM presented the best performance for detecting the comments polarity about the impeachment process. In some cases, the SVM presented an F-score at least 1.03% higher than the others.

How to Correctly Evaluate an Automatic Bioacoustics Classification Method

In this work, we introduce a more appropriate (or alternative) approach to evaluate the performance and the generalization capabilities of a framework for automatic anuran call recognition. We show that, by using the common k-folds Cross-Validation (k-CV) procedure to evaluate the expected error in a syllable-based recognition system the recognition accuracy is overestimated. To overcome this problem, and to provide a fair evaluation, we propose a new CV procedure in which the specimen information is considered during the split step of the k-CV. Therefore, we performed a k-CV by specimens (or individuals) showing that the accuracy of the system decrease considerably. By introducing the specimen information, we are able to answer a more fundamental question: Given a set of syllables that belongs to a specific group of individuals, can we recognize new specimens of the same species? In this article, we go deeper into the reviews and the experimental evaluations to answer this question.

Using Complex Networks to Assess Collaboration in Rap Music: A Study Case of DJ Khaled

DJ Khaled is a popular musician that is known for having many collaborators in his songs. Hence, in this paper, we model the evolution of DJ Khaleds collaboration network as nine different networks that incrementally consider the albums of his discography. The network of each album includes the collaborations from previous ones and adds the collaborations from the new album. The artists are represented as nodes and the edges are the number of songs they appear together. Our focus is to answer whether or not: (i) we can identify meaningful communities in this network; and (2) there is an artist who has greater influence as networks emerges. By using the network average clustering coefficient, we found that the artists in the the network tend to naturally cluster in a logical manner. As a result, we identified nine communities, six of them are meaningful, and we identified the rapper Rick Ross as the most influential artist of the network.

Recognizing Family, Genus, and Species of Anuran Using a Hierarchical Classification Approach

In bioacoustic recognition approaches, a “flat” classifier is usually trained to recognize several species of anuran, where the number of classes is equal to the number of species. Consequently, the complexity of the classification function increases proportionally to the amount of species. To avoid this issue we propose a “hierarchical” approach that decomposes the problem into three taxonomic levels: the family, the genus, and the species level. To accomplish this, we transform the original single-label problem into a multi-dimensional problem (multi-label and multi-class) considering the Linnaeus taxonomy. Then, we develop a top-down method using a set of classifiers organized as a hierarchical tree. Thus, it is possible to predict the same set of species as a flat classifier, and additionally obtain new information about the samples and their taxonomic relationship. This helps us to understand the problem better and achieve additional conclusions by the inspection of the confusion matrices at the three levels of classification. In addition, we carry out our experiments using a Cross-Validation performed by individuals. This form of CV avoids mixing syllables that belong to the same specimens in the testing and training sets, preventing an overestimate of the accuracy and generalizing the predictive capabilities of the system. We tested our system in a dataset with sixty individual frogs, from ten different species, eight genus, and four families, achieving a final Micro- and Average-accuracy equal to 86 % and 62 % respectively.