THESIS OBJECTIVES. Static Canopy Model (SCM); Dynamic Canopy Model (DCM); Enhanced Static Trunk and Ground(ESTG) model;

Transcription

1 THESIS OBJECTIVES Static Canopy Model (SCM); Dynamic Canopy Model (DCM); Enhanced Static Trunk and Ground(ESTG) model; Combination of individual models into a generic doubly-selective 3D vegetation model; Final model validation. 1 Vigo 22 nd June 2017

2 SDN-ORIENTED GLOBAL NETWORK OPTIMIZATION ALGORITHM MHIRI Saber, Francisco Javier González Castaño AtlantTIC, GTI Group University of vigo Real AP network (Wifi + 5G) Control Plane Location profiler 75% Flow profiler 85% Real user behavior APs assignment algorithm 52%

3 SENTIMENT ANALYSIS IN SOCIAL MEDIA CONTENTS USING NATURAL LANGUAGE PROCESSING TECHNIQUES Author: Tamara Álvarez López Advisors: Enrique Costa Montenegro, Milagros Fernández Gavilanes Affiliation: AtlantTIC Research Center, University of Vigo MOTIVATION OBJECTIVES Growth of social networks and their users. New way of expression. Very valuable information for companies, who need to analyse it. Automatic sentence-level SA system. Fine-grained SA system, by means of aspect detection. Application and adaptation to different datasets and contexts. Implementation for English and Spanish. e.g. Service was nice but price was not worth it and food quality was very miserable. Great amount of information available online. Difficult to analyse. Natural Language Processing (NLP) techniques. Applied in fields like: Sentiment Analysis (SA), Aspect Extraction, Text Generation, Entity Recognition, etc. Sentence-level SA Aspect-level SA Service Price Food quality

4 PRESENTATION ATTACK DETECTION ON FACE RECOGNITION SYSTEMS IN MOBILE DEVICES Artur Costa-Pazo, Esteban Vazquez-Fernandez and José Luis Alba-Castro Thesis Objectives Analysis of Presentation Attack Detection PAD in Real Scenarios. Evaluation Framework. Develop robust face-pad methods in mobile devices Preliminary Results Proposal and commitment for writing a chapter on a book. Handbook of Biometric Antispoofing Participation on IJCB 2017 competition on generalized face presentation attack detection in mobile authentication scenarios obtaining the 1st rank across all the four protocols. Figure 1: Samples of the different presentations attack instruments, PAI. Replay-Mobile database was captured within the framework of collaboration between Idiap and Gradiant. Workshop on Monitoring PhD Student Progress acosta@gradiant.org

5 Homomorphic Lattice Cryptosystems for Secure Signal Processing Workshop on Monitoring PhD Student Progress. June 22-23, Main objective Alberto Pedrouzo-Ulloa, Advisors: Juan Ramón Troncoso-Pastoriza, Fernando Pérez-González Privacy protection when dealing with sensitive signals in untrustworthy environments through the use of SSP (Secure Signal Processing) Specific objectives Privacy protection when dealing with multidimensional signals Design of new primitives and protocols for encrypted signal processing Security analysis and development of encrypted compression schemes Secret key owner Service A Encrypted data Encrypted results Encrypted data Encrypted results Service B Service C Sensitive signals Service D 1

6 PROPOSAL OF A SOCIAL INTERACTION TOOL, INTEGRATED IN FACEBOOK, FOR E-LEARNING author HELDER RODRIGO SOARES PINTO thesis advisor MARTÍN LLAMAS NISTAL PhD in ICT ARCHITECTURE AND TELEMATIC SERVICES and FERNANDO MIKIC FONTE E-LEARNING ENVIRONMENT OF A SOCIAL INTERACTION DEVICE FACEBOOK APP STORAGE WEB SITE WEB BROWSER MOBILE SQL DATABASE WEB SITE ACHIVEMENTS

7 Thesis Objectives Results and Discussions Motivation Radio Frequency Propagation, Characterization And Measurements For Antenna Sensor Networks With Applications In Smart Farming Hicham Klaina, supervised by Ana Vazquez Alejos and Otman Aghzout Hicham Klaina, Ana Vazquez Alejos and Otman Aghzout Radio frequency propagation study for nearground antenna sensor networks dedicated to agriculture under review. Hicham Klaina, Ana Vazquez Alejos and Otman Aghzout RFID Tags Detection Effects for Smart Agriculture Applications under review. Fig 1. Path loss in agriculture fields at 2.4GHz for H1=20cm & H2=40cm. Measurements show that a significant change in path loss occurs when lowering the height of the antennas near to ground after a break distance, changing the propagation channel model. WSN for farm monitoring using Raspberry: Fig 2. RSSI values using ALN G tag attached to a plastic bottle at 7cm from the bottom. Measurements demonstrate that RFID tags type, positioning and food dielectric constant have a big influence on tags attached to food detection performance. Novel design needed. Traceability of food transportation using Arduino: Design and development of an ideal smart farming system for crops monitoring from cultivation to transport and storage in warehouses. Using this WSN topology, farmer is allowed to monitor his farm in real time using his smartphone or laptop whenever and wherever he is. Data model offered to farmers to trace their food transportation which is a part of the new V2X technology, from the farm to the table. Food traceability in real-time on Google Maps will accessible to the farmer. Workshop on Monitoring PhD students Progress - PhD program in Information and Communications Technology - June 2017

8 DISTRIBUTED COLLABORATION IN MOBILE CONTEXT-AWARE SYSTEMS Bluetooth WiFi Accelerometer GPS Local proximity Urban areas Social proximity Proximity Magnetometer Gyroscope Microphone Fátima Castro Jul Ana Fernández Vilas Rebeca P. Díaz Redondo I & C Lab AtlantTIC Research Center 2 Universidade de Vigo

9 Evaluate Share Use Remix Create Explore By: Carmelo B. España V.

10 DATA-DRIVEN ALLOCATION OF NETWORK RESOURCES IN UDNs SILVESTRE LOMBA MALTA, SUPERVISED BY MANUEL FERNANDEZ VEIGA Ultra-dense networks (UDN) have been considered one of the most advanced technologies in 5G and can be the key to meet users expectations. Due to UDN characteristics, energy management and energy consumptions are serious challenges to network operators. Studies have concluded that Base Stations consume about 60%-80% of energy use in cellular networks. In my thesis, I would like to contribute with: New techniques for sampling, compressing and summarizing network mobility information in realtime. Evaluate the performance improvements of knowing mobility and traffic patterns in UDNs, as far as energy-efficiency is concerned. Develop deterministic mean-field approximations for predicting the performance in large UDNs.

11 Recommendation of Tourism Resources Supported by Crowdsourcing Recommendations based on Crowd-sourced information Tourism Crowdsourcing State-of-the-art review Tourist behaviour questionnaire Research Plan Profiling Data collection Crowd-sourced data analysis Multi-criteria profiling Trust & Reputation modeling Done Recommendations User-based and item-based recommendations Off-line && On-line processing Final Thesis Fátima Leal, University of Vigo, June 2017 Workshop DOC-TIC

12 Motivation Genomic data are sensitive. Genomic data are unique identifier of each person. Genomic data are not revocable. Thesis Objectives E-Health applications privacy analysis. Develop and device countermeasure. Results Novel protocol of paternity testing. Achieving anonymous credential for access control in genomic area. Sketch of the protocol featuring an inherent access control through attributes enforcing the patient s access policy Implementing "entropy-based" privacy measurement Dynamic Privacy-Preserving Genomic Susceptibility Testing.