GTTI English Version Versione Italiana

RIUNIONE ANNUALE 2018 - Bari, 25-27 Giugno 2018

Programma Dettagliato

Martedì 26 Giugno


GAUChO - A Green Adaptive Fog Computing and Networking Architecture


Fog Computing (FC) is an emerging paradigm that extends Cloud Computing towards the edge of the network. In particular, FC refers to a distributed computing infrastructure confined on a limited geographical area in which some applications/services run directly at the network edge in smart end-devices. The goal of FC is to improve efficiency and reduce the amount of data that needs to be transported to the Cloud for massive data processing, analysis and storage. However, in latency-sensitive and energy-efficient (i.e., green) applications, the FC paradigm per se might not be appropriate and (at least) part of the computation has to be transferred to the sensor/actuator end-devices level. Here, decisions must be usually taken in a very short time and energy and power constraints constitute a limiting factor. Furthermore, the design of efficient solutions within FC also requires investigate a novel communication/networking paradigm, called Fog Networking (FN), in order to meet specific configurability, adaptability, flexibility and energy/spectrum-efficiency constraints. Being currently FC and FN features designed, optimized and implemented independently each other, the GAUChO project aspires at designing a novel distributed and heterogeneous architecture able to functionally integrate and jointly optimize FC and FN capabilities in the same platform. The joint FC+FN architecture, representing the overall outcome of the project, aims at supporting low-latency and energy-efficiency as well as security, self-adaptation, and spectrum efficiency by means of a strict collaboration among end-devices and FC+FN units in a same integrated platform. In addition, the development of suitable analytic methods and definition of appropriate techniques will enable extra relevant characteristics of the FC+FN platform including ubiquity, decentralized management, cooperation, proximity to end users, dense geographical distribution, efficient support for mobility and real-time applications. To achieve this goal, the GAUChO project foresees to address several relevant and challenging research topics that require skills and knowledge in different scientific fields. The scientific contributions that the project aims at achieving, are: i) efficient schemes for the coordinated management of resources and interference in heterogeneous wireless communication systems; ii) joint optimization of communication and computing capabilities to support energy-efficient management and self-reconfigurations; iii) a learning modality permitting software agents to detect variations within the integrated FC+FN platform; iv) model-free fault diagnosis systems and comprehensive methodology integrating intelligence-based mechanisms for optimally managing energy consumption, detecting changes in environment/system under inspection, and evaluating and mitigating the possible occurrence of faults affecting the end-devices computing units. The GAUChO project is expected to have a significant technological impact on many up-to-date and relevant families of technologies, such as Smart Wireless Sensor Networks (SWSNs), Smart Objects of Internet-of-Things and Intelligent Embedded Systems. All in all, the project will allow the FC+FN paradigm to enter into a new phase where real-world problems emerging from complex applications are addressed and effectively solved. It is expected that project outcomes will move from basic research to mass production in years 2018-25 providing significant economic benefits to masses of potential end users, together with the added value of the offered application services, limited cost of the communication and processing infrastructure.




The Green Adaptive Fog Computing and Networking Architecture (GACho) Project General Overview

R. Fantacci, C. Alippi, A. Unicini, D. Tarchi

A Matching Theory Framework for Joint Fog Networking and Computing

F. Chiti, R. Fantacci,B. Picano

Anomaly behavior detection into physical data of Fog Computing and IoT systems

L. Pierucci, M. Roveri, T. Pecorella, F. Nizzi

The Gaucho Project Network Testbed supporting Fog Computing Service Models

D. Tarchi, W. Cerroni , C. Caini, A. Bozorgchenani.



Best GTTI PhD Awards 2018




Radio Access for the Internet of Things Traffic in Fifth-Generation Cellular Networks

Marco Centenaro

Abstract: In this talk, we deal with the fifth-generation (5G) cellular network as a fundamental enabling technology for massive machine-type communications (mMTC). 5G systems operate on licensed frequency bands, and exploit long-range wireless links and a star topology. Two contention-based radio access protocols designed in collaboration with Nokia Bell Labs Stuttgart are proposed, mathematically characterized, and compared against the current cellular standard (4G). The performance evaluation results show that the proposed protocols provide lower latency, higher throughput, and reduced downlink feedback thanks to their connectionless approach.

Bio: Marco Centenaro received the BSc degree in Information Engineering in 2012 and the MSc in Telecommunication Engineering in 2014, both from the University of Padova, Italy. From November 2014 until October 2017, he was a PhD student at the Department of Information Engineering of the University of Padova: during the doctoral course he carried out research on the support of massive Machine-to-Machine (M2M) traffic in heterogeneous networks and fifth-generation (5G) cellular systems. From September 2016 to December 2016 he was an intern at Nokia Bell Labs, Stuttgart, Germany; from January 2017 to July 2017 he was a visiting researcher at the Yokohama National University (YNU), Yokohama, Japan. Since November 2017, he is a postdoctoral research fellow at the Department of Information Engineering of the University of Padova, doing research on channel-state information signaling reduction in cellular networks based on frequency-division duplex (FDD).



Statistical Models for the Characterization, Identification, and Mitigation of Distributed Attacks in Data Networks

Mario Di Mauro

Abstract: The thesis focuses on statistical approaches to model, mitigate, and prevent distributed cyber-attacks. Three fundamental issues emerge distinctly. The first issue concerns the threat propagation across the network, which entails an "avalanche" effect, with the number of infected nodes increasing exponentially as time elapses. The second issue regards the design of proper mitigation strategies (e.g., threat detection, attacker's identification) aimed at containing the propagation phenomenon. Finally, in order to guarantee protection even when the attacker cannot be defeated, it is desirable to act on the system infrastructure to grant a conservative design through a controlled degree of redundancy. The presentation will focus on the first two aspects: i) with regard to the first issue, I will introduce Kendall's birth-and-death process as a powerful analytical model to deal with the cyber-threat propagation phenomenon; and ii) with regard to the second issue, I will present a rigorous statistical characterization for a novel kind of randomized Distributed Denial of Service (DDoS) attack, along with the analysis of a designedfrom-the-scratch algorithm for botnet identification.



Detection and motion parameters estimation techniques in Forward Scatter Radar

Nertjana Ustalli

Abstract: Forward Scatter Radar (FSR) systems exploit the extreme bistatic geometry, reached when thebistatic angle approaches 180°. FSR systems are receiving a renewed interest in the last years dueto their potentialities for the detection of low-observable targets (i.e. small targets or stealthtargets). This thesis reports on research into this field of radar systems with additionalcontributions to two broad areas: target detection and motion parameters estimation. The mainobjectives of this work were (i) the development of innovative techniques for target detectionjointly with the analytical characterization of the performance in terms of probability of false alarmand probability of detection, and (ii) the derivation of kinematic parameters estimation techniquesjointly with an accuracy analysis of the estimated parameters. During this talk, the main resultsreached with respect to these objectives will be discussed.



Sessione CNIT - 5G: ricercare, sperimentare, applicare


In August 2017 Italy's Ministry of Economic Development (Ministero dello sviluppo economico, MISE) has selected companies to carry out pre-commercial trials of 5G technology using spectrum in the 3.6GHz-3.8GHz range in five cities (Milano, Bari-Matera, Prato-L'Aquila). Vodafone Italia, which has won the tender announced by MISE for the area of Milan, has officially launched its project at the end of 2017. Vodafone involved in the experimentation three research centers (the Politecnico di Milano, CNIT and IIT), 25 companies and 10 endorsers among local institution and companies. The activity consists of 41 projects out of 9 trasversal sectors (health and wellness, security and surveillance, smart grids and smart city, transportation, manufacturing, education and entertainment). CNIT in particular is involved in 2 use cases and 2 horizontal research activities supporting the whole project. The 5G Special Session includes an introduction by Andrea Abrardo, scientific responsible for CNIT activity within the 5G project in Milan. In the Introduction e general view of the project and the specific role of CNIT will be presented.


Then, the Session will be divided into the following four parts, each describing in more details the CNIT activity within the project:


Development of a 5G System level simulator

Giuseppe Piro (Politecnico di Bari)


Inertial Navigation System on 5G network for tourism 4.0 applications

Alessansdro Nicoli (Università di Parma)


5G network slicing and the role of network softwarization technologies

Claudia Campolo (Università di Reggio Calabria)


Network security challenges in 5G

Giuseppe Bianchi (Università di Roma Tor Vergata)


Mercoledì 27 Giugno


Prospettive nell'uso del Machine Learning




Deep learning techniques for multimedia applications (Tecniche di deep learning per applicazioni in campo multimediale)

Enrico Magli

Abstract: The talk will first describe the use of deep convolutional networks for visual analysis, particularly the problems of license plate recognition and vehicle classification. Then, I will briefly introduce generative adversarial networks and show some recent developments of an adversarial network for data defined on a graph. Such network has been trained for generation of point clouds, and is able to learn graphs with meaningful features.

Bio: Enrico Magli is a Full Professor with Politecnico di Torino, Italy, a Fellow of the IEEE and has been an IEEE Distinguished Lecturer from 2015 to 2016. He is an Associate Editor of the IEEE Transactions on Circuits and Systems for Video Technology and the EURASIP Journal on Image and Video Processing. His research interests include machine learning, compressive sensing, image and video coding.



Approaching UAV Image Analysis with Support Vector Machines - (Analisi di Immagini UAV mediante Support Vector Machines)

Farid Melgani

Abstract: Unmanned aerial vehicles (UAV) are recognized as a valid alternative or a complementary solution to satellite sensors particularly for small coverage or inaccessible areas. The extremely high spatial resolution of UAV images however raises methodological challenges for their effective exploitation. In this talk, we will describe two very recent solutions based on support vector machines (SVM) for object detection and classification issues in urban monitoring.

Bio: Farid Melgani received the Ph.D. degree in electronic and computer engineering from the University of Genoa in 2003. He is an Associate Professor of Telecommunications at the University of Trento. He is the Responsible of the Master degree in Telecommunications Engineering and the head of the Signal Processing and Recognition (SPR) Laboratory at the same university. His research interests are in the areas of remote sensing, signal/image processing, machine learning and computer vision. He is coauthor of about 200 scientific publications. He is an IEEE Fellow and an Associate Editor of the IEEE Geoscience and Remote Sensing Letters, International Journal of Remote Sensing, Remote Sensing, and Sensors.