Poster 2

Wednesday, 17 June 2020, 12:15-14:30 CEST, Recommended re-viewing,

Wednesday, 17 June 2020, 12:15-16:00 CEST, Non-Live interaction (Chat),  link sent only to Registered people


Edge Cloud Based Platform for Smart Parking

Naila Bouchemal (ECE Paris, France)
Nowadays, it is estimated that 24 hours a year are spent looking for a car parks place in urban city. This represents approximately 70 full days of life. Moreover, this results in fuel loss, increased stress but above all; increasing pollution in city centers. The concept of smart city aims to reduce theses issues. It consists of optimizing the cost, organization and residents well-being. Autonomous vehicles, sensors and infrastructures are an integral part of smart cities, their role is very important in the traffic flow, reducing greenhouse gas emissions and increasing road safety. Autonomous vehicles use wireless technologies to improve road safety and transport efficiency; this induce recent advances in mobile networks. The particularity of these advances (next generation 5G, Internet Of Things IoT and beyond) lies in the heterogeneous and intelligent networks design [1]. Intelligent Network design uses AI techniques, this allows better update, but may raise other constraints in terms of latency, network bandwidth and data security. This is where Edge Computing comes in. It is about being able to perform the calculations as closely as possible. For an autonomous car, it is about being able to make decisions in the fastest way.We propose in this paper a prototype of connected and intelligent parking for car parks places allocation in context of Smart City. This proposal is based on the development of AI algorithms at the Edge Cloud. This will allow the detection and prediction in real time of available car parks thanks to communication between the AI system and the Smart city environment: Sensors, infrastructure, and autonomous car.


Narrowband Internet of Things (NB-IoT) for Energy Efficient Body Area Network (BAN) Applications

Emil Novakov (IMEP, France)
The objective of this work is to study the possibilities to use the Narrow-band Internet of Things (NB-IoT) 4G cellular communication technology for Body Area Network (BAN) and energy-efficient wireless communications. To test the different ways of optimizing the energy consumption of an NB-IoT device, a specialized real-time power measurement system was developed. Measurements were carried out on a real 4G cellular network and real-time power and energy consumption were recorded. In real propagation conditions the energy used to transfer 128-byte blocks varies, depending on the software configuration, between 700E-6 Wh and 80E-6 Wh. According to the measurements for SARA-N211 NB-IoT device, two AA battery with 2.5 Ah capacity can provide the energy necessary to transfer 1300 bytes per day for 10 years.


5G Ecosystem: The 3 Combined Critical Requirements, Maximum Network Lifetime, Full Coverage and Connectivity Based Wireless Sensor Networks Communication

Amal Chaffai (Universidad Politecnica de Valencia, Spain)
To bring access to content wherever, whenever to face the whims of nature. A sustainable networks for new generation communication environment 5G trend to optimize reduction energy consumption. In order to extend the lifetime of wireless sensor networks, while ensuring less holes than possible and improving the coverage. Wherefore, our study focuses mainly on enhancing a reduction energy consumption of wireless sensor networks, based on a hopping technique. Accordingly, the numerical results affirm that energy efficiency can be achieved with 3 hops in a nonlinear architecture mixed routing scheme. Keeping in consideration that depend on the design parameter of the generique value radii
R = rmáx as minimum communication range required.


5G-SOLUTIONS Analysis of Living Labs and KPIs Definition Methodology

Ioannis Markopoulos (FORTHNET, Greece); Andrea Di Giglio (Telecom Italia, Italy); Baruch Altman (LiveU, Israel); Håkon Lønsethagen (Telenor Research, Norway); Christos Verikoukis (CTTC & UB, Spain); Angelos Antonopoulos (Telecommunications Technological Centre of Catalonia (CTTC), Spain); Silvia Canale (Applied Research to Technologies, Italy); Matteo Grandi (IRIS Technology Solutions, Spain); Sofiane Zemouri (IIX, IBM, Ireland)
The scope of this paper is to provide an analysis of the 5G-SOLUTIONS Living Labs (LLs), the associated use cases, as well as the requirements and the target Key Performance Indicators (KPIs) definition methodology that will set the benchmarking for the actual measurements. 5G-SOLUTIONS is a 5G-PPP project supporting the EC’s 5G policy by implementing the last phase of the 5G-PPP roadmap. It aims to prove and validate that 5G provides prominent industry verticals with ubiquitous access to a wide range of forward-looking services with orders of magnitude of improvement over 4G, thus bringing the 5G vision closer to deployment. This will be achieved through conducting advanced field-trials of innovative use cases, directly involving end-users across five significant industry vertical domains (Living Labs in term of Use Case categorization): Factories of the Future, Smart Energy, Smart Cities, Smart Ports, Media & Entertainment. 5G-SOLUTIONS Living Lab (LL) analysis is dependent on usability needs. The project aims at capturing the requirements from the end-user stakeholders, as well as the relevant target technological and business Key Performance Indicators (KPIs), which will be validated in the LLs. This outcome will feed other critical tasks and point out the technological enablers for facilitating the execution of the field trials. To this end, the use cases will be validated towards their conformance to target 5G KPIs, service types i.e. eMBB, URLLC and mMTC as well as their business potential, ethical and social acceptance. This paper defines in a clear and solid way the initial version of the LL analysis, in order to prove and validate that the 5G technology can provide prominent industry verticals with ubiquitous access to a wide range of forward-looking services with orders of magnitude of improvement over 4G.


Non-linear Regression of Delay Percentiles in PONs Using Machine Learning Techniques

José Alberto Hernández (Universidad Carlos III de Madrid, Spain); Amin Ebrahimzadeh (INRS, Canada); Martin Maier (Institut National de la Recherche Scientifique (INRS), Canada); David Larrabeiti (Universidad Carlos III de Madrid, Spain)
This article shows how to learn delay models in Passive Optical Networks (PONs) using Supervised Machine Learning techniques. Essentially, a non-linear regression ML algorithm is trained with PON simulation data, showing that it can provide accurate equations for several delay-based metrics, in particular, delay percentiles. In particular, we obtain R-score above 99% for high-delay percentiles and provide a general equation for any delay percentile in the upstream channel of a PON employing IPACT. Applications of this technique include the ability to dimension PONs for scenarios where worst-delay bounds need to be guaranteed, like in PONs for the transport of delay-sensitive fronthaul flows.


On an Access Control Model Enhancement for the 5G System

Luis Suarez (IRT bcom & Université de Bretagne Occidentale (UBO), France); David Espes (University of Brest & LabSTICC, France); Frédéric Cuppens (IMT Atlantique, France); Philippe Bertin (Orange Labs & Bcom, France); Cao-Thanh Phan (BCOM, France); Philippe Le Parc (University of Brest, France)
The realization of communication services over 5G needs resource sharing as a way to achieve network coverage. To do so, it is necessary to consider security access mechanisms to regulate how interconnections are made. The existing models do not address all the needs inherent to the 5G architecture, such as access control mechanisms, multi-tenancy, multi-domain and multiple security levels. This position paper presents the state of the art of access control models and their application in 5G networks. Then, points out problems that are not addressed and establishes the conditions that such access control scheme must obey in order to be suitable for its utilization in the 5G system.


The SPIDER concept:A Cyber Range as a Service Platform

Christos Xenakis (University of Piraeus, Greece); Anna Angelogianni and Eleni Veroni (University of Piraeus, Greece); Eirini Karapistoli (CyberLens, United Kingdom (Great Britain)); Matthias Ghering (CYBERLENS, United Kingdom (Great Britain)); Neofytos Gerosavva and Vasileios Machamint (EIGHT BELLS, Cyprus); Pierluigi Polvanesi and Angela Brignone (ERICSSON TEI, Italy); Jeronimo Nuñez Mendoza (Telefonica, Spain); Antonio Pastor (Telefonica I+D, Spain)
The evolving cyber security threat landscape along with the extensive application of the 5G technology in various sectors brings together several cyber security concerns. To that extent, there is an apparent need for appropriate testing and training before the massive commercial deployment of 5G. The SPIDER platform proposes an innovative Cyber Range as a Service (CRaaS) platform that extends and combines the capabilities of existing telecommunication testbeds and cyber ranges into a unified facility for (i) testing new security technologies, (ii) training modern cyber defenders in near real-world conditions, and (iii) supporting organizations and relevant stakeholders in making optimal cybersecurity investment decisions.


Improving Apache Spot Using Autoencoders for Network Anomaly Detection

Athanasios Priovolos, Georgios Gardikis and Dimitris Lioprasitis (Space Hellas S.A., Greece); Socrates Costicoglou (Space Hellas SA, Greece)
Apache Spot is an increasingly popular opensource platform for advanced network insights, focusing on the detection and analysis of anomalies, which can potentially correspond to security incidents. In this paper, we propose an improvement over Apache Spot’s built-in Machine Learning algorithm (Latent Dirichlet Allocation – LDA), replacing it with an Autoencoder based on deep learning techniques. We implement the Autoencoder functional block and deploy it into the Apache Spot’s pipeline, integrating it with Hadoop and Spark. Finally, we evaluate and benchmark the Autoencoder against the built-in LDA, using a publicly available network traffic dataset with cyber-attacks. The result is a considerable increase of accuracy, precision and recall.


In-Network Knowledge Reasoning with New IP

Lijun Dong (Futurewei Technologies, USA); Lin Han (Futurewei Technologies Inc., USA); Richard Li (Futurewei Technologies, USA)
Knowledge is a collection of statements, which can be stored in a knowledge base. Take the vehicular network as an exemplary use case, one of the most important knowledge that a driver wants to know is the traffic condition on the path from one location to another. If the path contains many roads, the traffic condition on the path needs to be reasoned from the traffic conditions of those roads. On the other hand, even if the path only contains one highway, different locations of the highway may have different traffic conditions (i.e. congested or not). For example, a car wants to know whether the highway is congested between the exit 1 and exit 4. Such knowledge can be reasoned by the following rule: If Location1 does not have congestion AND Location2 does not have congestion AND Location3 does not have congestion, then the highway between the exit 1 and exit 4 does not have congestion. The Location1, Location2 and Location3 are three locations that are picked as sampling locations between the exit 1 and exit 4 on the highway. In the use case, we assume that the traffic server stores all the traffic reports from any car on the highway, which is willing to notify the real-time traffic condition periodically and proactively when it detects congestion on the road. The existing knowledge request process requires the data acquisition from the traffic server. Although the current information centric networking has the in-network caching, and store-and-forward routing capabilities, which have been proven to greatly improve the data acquisition performances, the responses to the knowledge reasoning requests from clients can still be slow if all the data needed for reasoning come from the traffic server directly. Recently, New IP has been proposed as an evolutionary extension to the current IP packet to enable user-oriented, context-aware, intelligent services in the Internet. A New IP packet carries a Contract block between the header and payload. The Contract could include contract clause(s) and associated metadata. A contract clause describes how the routers treat the packet as it traverses the network based on the predefined triggering event and condition. The “”Metadata”” contains contextual information about the packet itself, user’s requirement or preference, etc. The paper proposes knowledge reasoning functionalities to be supported in a distributed manner by network nodes. The recently thriving Internet framework, i.e. New IP could facilitate the enablement of the proposed functionalities with in-network intelligence. The performance improvement over the legacy approach is analyzed and illustrated.


Design and Implementation of an IoT-based Ambient Intelligence Framework for Smart Built Environments

Reza Tasooji, Archi Dasgupta and Denis Gracanin (Virginia Tech, USA); Kresimir Matkovic (VRVis Research Center, Austria); Matthew LaGro and Mike Mihuc (OSIsoft, USA)
We describe the design and implementation of an Internet of Things based framework for monitoring environmental conditions and energy consumption in smart built environments (such as smart homes) using low-cost sensory devices. The framework incorporates machine learning algorithms to analyze the collected data, detect trends and predict behavior in support of ambient intelligence. We describe a reference implementation of the developed framework. The system provides real-time data collection and analysis of the collected longitudinal data. A web-based user interface provides data visualization and real-time control of connected devices. The results from three case studies support the utility of the implementation and ability of the framework to incorporate different analytic methods.