Antonio Tedeschi

A real-time automatic pavement crack and pothole recognition system for mobile Android-based devices

Due to the rapid growth of vehicles and traffic accidents caused by road pavement defects, road safety has become a pressing concern worldwide. For this reason, Countries and Federal States have started focusing their resources on the analysis of civil infrastructures to assess their safety and serviceability. The analyses are performed by specialized teams of inspectors and structural engineers who manually inspect road infrastructures and provide detailed reports about the detected pavement distresses and their magnitudes. This work aims at providing a new system able to detect the framed distress using solely the computational resources provided by a mobile device To reach this goal, an automatic pavement distress recognition system based on the OpenCV library is developed and embedded in a mobile application, enabling the recognition of three common pavement distresses: Pothole, Longitudinal-Transversal Cracks, and Fatigue Cracks. Our method, tested on several Android mobile platforms, is able recognize the pavement distresses of interest reaching more than 0.7 of Precision, Recall, Accuracy, and F-Measure. This application promises to improve the on-site work of inspectors by decreasing the time required to perform inspections while ensuring, at the same time, a higher level of accuracy.

Performance improvements of cooperative spectrum sensing in cognitive radio networks with correlated cognitive users

In cognitive radio networks, secondary (unlicensed) users opportunistically access vacant frequency bands licensed to primary users. Spectrum sensing is the technique exploited to discover these unused bands. Cooperative spectrum sensing is the solution to problems that arise in the presence of fading and shadowing effects. Observations from the cooperative sensors are usually assumed independent, in order to simplify the analysis. Conversely, we consider the case of correlated observations and propose an innovative test for cooperative sensing, in order to improve the detection performance. Our test combines two different fusion rules at once (i.e. the OR and the Majority voting rules), recovering all those unlucky detection cases that are not sensed due to strong fading and shadowing. Our simulation results show the efficacy and effectiveness of our method for application to cooperative sensing in the presence of correlated cognitive users.

A mobile android application for road and pavement inspection by GPR data processing

Ground penetrating radar (GPR) is widely used in the field of civil engineering, as a non-intrusive technique for road and pavement monitoring. This work presents an innovative application for mobile platforms (smartphones and tablets) for real-time GPR data processing. Our work aims at providing a useful support for engineers and technicians as well, for road and pavement inspection. According to the presented results, this application can play an important role for all the agencies involved in roads and highway management. In particular, our application will provide strategic and innovative potentialities, by improving the on-site efficiency and effectiveness of the works.

Big Data Sentiment Analysis for Brand Monitoring in Social Media Streams by Cloud Computing

The rapid growth of the World Wide Web and social media allows users playing an active role in the contents’ creation process. Users can evaluate the brands’ reputation and quality exploiting the information provided by new marketing channels, such as social media, social networks , and electronic commerce (or e-commerce). Consequently, enterprises need to spot and analyze these digital data in order to improve their reputation among the consumers. The aim of this chapter is to highlight the common approaches of sentiment analysis in social media streams and the related issues with the cloud computing , providing the readers with a deep understanding of the state of the art solutions.

Moisture Content Evaluation for Road Surfaces Monitoring by GPR Image and Data Processing on Mobile Platforms

This work introduces a mobile application for moisture content evaluation aimed at road-surfaces monitoring. Our application exploits Ground Penetrating Radar (GPR) image and data processing in real-time on Android platforms. GPR is one of the most advanced technology in civil engineering applications for road pavement monitoring, while water intrusion in structural layers is one of the most relevant causes of damage of road-surfaces. The novelty of this work (that is the power of our application) is to show the consistency of the GPR diagnosis regarding these hidden conditions, implementing GPR image and data processing on smartphones and tablet computers.

PV Panel Modeling: A Mobile Application for Modeling Photovoltaic Panels Using Datasheets Information

Over the past years, different approaches have been proposed for circuital modeling photovoltaic modules. Most of these approaches introduce simplifications and/or approximations techniques or exploit inaccurate data interpolated from the plots representing panel characteristic curves: this in order to overcome the treatment of transcendental equations or to add points for numerical fitting. Unfortunately, this leads to ineffective algorithms and inaccurate results. In this work, we propose a fast and powerful mobile application, named PV Panel Modeling, for the identification of PV panels from datasheet. PV Panel Modeling is based on a complete theoretical and practical analysis on the extraction of one-diode model parameters. In particular, the proposed procedure separates the independent parameters from the dependent ones by using suitable algebraic manipulations of the commonly used equations. Hence, it is possible to verify the open circuit, short circuit and maximum power point conditions, and to reduce the dimensions of the search space of the parameters. The application and its core are designed in order to be cross-platform and ensure high performances. Here, we propose an implementation for handheld devices based on Android operating system. Finally, the obtained results have been validated on hundreds of PV panels belonging to the California Energy Commission database.

Cooperative spectrum sensing for positioning in cognitive radios

The cognitive radio technology promises to become the solution to the spectrum scarcity problem, allowing secondary (unlicensed) users to opportunistically access vacant frequency bands licensed to primary users. Cooperative spectrum sensing is the technique used to discover the unoccupied frequency bands, in the presence of strong fading and shadowing effects. In addition, cooperative spectrum sensing allows to exploit positioning information as well as location awareness. The cooperative sensors are usually assumed independent each other, in order to simplify the analysis. Conversely, here we consider the case of correlated observations and propose a new double threshold test, in order to improve the detection performance, while reducing the time requested to detect a primary user. Our test combines two different fusion rules at once (i.e. the OR and the Majority voting rules), recovering all those unlucky detection cases that are not sensed due to strong shadowing. Finally, our simulation results show the effectiveness of the proposed method for application to cooperative sensing for positioning in the presence of correlated observations.

Performance improvements of reputation-based cooperative spectrum sensing

Cognitive radio is establishing itself as the most promising technology in tackling the problem of spectrum scarcity. It allows secondary users to opportunistically access and re-use the spectrum resources under-utilized by licensed or primary users. Spectrum sensing (either cooperative or non-cooperative modes) is exploited to identify those unused (free) frequency bands. This new technology has been proved to be vulnerable to an increasing number of attacks, such as Byzantine attacks. Here, we propose a novel reputation-based cooperative spectrum sensing method that first detects, and then rejects any malicious secondary user, thus improving the system performances. We evaluate the error probabilities (detection vs. false alarms) of the system, as well as the number of both correctly identified attackers and discarded honest users. Finally, we compare our method to a recently published reputation-based method (either blind or non-blind). As shown by the simulation results, the proposed method outperforms the conventional approach, confirming its robustness in detecting different types of Byzantine attacks.

Statistically-Enhanced Fine-Grained Diagnosis of Packet Losses

Packet losses are an important class of adverse events in Wireless Sensor Networks (WSNs), they can be caused by either a compromised or misbehaving node, or an attack focused on the wireless links of the network. Understanding the underlying cause is critical for the deployment of effective response measures aimed at restoring the network functionality. Shebaro et al. [9] proposed an initial approach for fine-grained analysis (FGA) of packet losses, and implemented and evaluated a tool based on it. Such approach profiles the wireless links between the nodes using resident metrics, such as the received signal strength indicator (RSSI) and the link quality indicator (LQI) for every packet, in order to achieve an accurate diagnosis of their root causes. The accuracy of their approach relies on the correct choice of some system parameters and thresholds, and empirically-determined values such as those proposed in their work might not always be optimal. Moreover, to reduce the burden on the network administrator, their approach uses a single threshold value set for the entire WSN, which can be suitable for some neighborhoods but not appropriate for others. In this work, we design an approach that builds a statistical model for determining optimal system thresholds by exploiting the variances of RSSI and LQI. Our model also has the advantage of allowing the setting of an individual threshold for each link. We have validated our approach through extensive MATLAB simulations based on real sensor data, showing that our model is accurate and its system parameters lead to an optimally-accurate fine-grained analysis of the underlying causes of packet losses.

Performance Analysis of Tracing Watermarking in the YST Domain for 3G Video-on-Demand Applications

In the last years, tracing watermarking has been proposed as a technique to provide a blind measure of the quality of service of the communication link, focusing on multimedia communication scenarios. This paper is focused on 3G video-on-demand (VOD) communication scenarios. We have exploited the new color space YST, recently intro- duced in the literature, where the luminance component (Y) is the same as in conventional YUV space, while the vectors S and T lie within the chrominance (UV) plane. In particu- lar, the third component T, defined as orthogonal to the YS plane, is chosen as the channel for the embedding process (instead of the Y channel). The results show the benefits obtained in digital watermarking by the new representation versus the conventional approach, i.e. the YUV color space. The main contribution of our work is twofold: the sensitivity of the YST representation outperforms the conventional one in terms of both quality (i.e. minimization of the alterations endured by the video during the embedding process) and security issues (i.e. detection of the watermark). Hence, the proposed procedure can be suitably applied for watermark- ing of 3G video-on-demand applications.