Evolving complex networks and their applications in web services

Organized by: Olayinka Adeleye, Auckland University of Technology, New Zealand

 

Despite the continual increase in the number of Web-APIs available on the internet, it is still challenging for API consumers to discover appropriate Web-APIs that could satisfy requirements. One of the main reasons for this is that Web-APIs registered on online directories such as ProgrammableWeb are in general isolated, as they are registered by diverse providers independently and progressively, ignoring continuous interactions among these APIs, which could enhance their discoverability. In this tutorial, we present an evolving complex network based approach for constructing evolving networks for Web-APIs that are capable of enhancing their discoverability. We first introduce mashups and Web-APIs interactions in the service ecosystem, and analyze their popularity distributions, and quantitatively measure two key node attachment dimensions within the ecosystem: Preferential Attachment and Similarity. Based on the analysis, we then propose two methods for constructing evolving Web-API networks using the theoretical procedures of the Barabasi-Albert and the Popularity-Similarity Optimization evolving complex network models.

Ontologies for Smart Mobility

Organized by: Bilal Farooq, Ph.D., Laboratory of Innovations in Transportation (LiTrans), Ryerson Universityn, Canada

Ontology is the explicit and formal representation of the concepts in a domain and relations among them. Transportation science is a wide domain dealing with mobility over various complex and interconnected transportation systems, such as land, aviation, and maritime transport, and can take considerable advantage from ontology development. In this context, there exists a strong potential to develop and study comprehensive smart mobility ontologies. The objective of this tutorial is to present different aspects of ontology development in general, such as ontology development methods, languages, tools, and software. Subsequently, we will present the currently available mobility-related ontologies developed across different domains, such as transportation, smart cities, goods mobility, sensors. Current gaps in the available ontologies are identified, and future directions regarding ontology development are exposed that can incorporate the forthcoming autonomous and connected vehicles, mobility as a service (MaaS), and other disruptive transportation technologies and services.

Reinforcement Learning for Robotics

Organized by: Prof. Dr. Ayşegül Uçar, Engineering Faculty Mechatronics Engineering Department, Firat University, Turkey

In this tutorial, we plan to summarize and explain the reinforcement learning methods that are alternative to classical control methods. There are a lot of reinforcement methods such as tabular Q-learning, Sarsa, actor-critic methods, and policy gradient. In recent years, new reinforcement learning methods including the Deep Neural Networks (DNNs) have appeared. They are Deep Q Networks (DQN),
DQN with Prioritized Experience Replay (DQN+PER), Double DQN (DDQN), Double Dueling DQN (D3QN), Reinforce, Asynchronous
Advanced Actor Critic Asynchronous (A3C) and synchronous Advanced Actor Critic Asynchronous. This tutorial is going to introduce all of them and apply the cart-pole balancing problem / inverted pendulum on the OpenAI GYM environment and webots.

An Overview of the Security Challenges of IoT Solutions

Organized by: Arthur Desuert and Amir Ali Pour, PhD students at Laboratoire de Conception et d’Intégration des Systèmes (LCIS) Alpes, France

Connected devices are getting really present in the daily life of human users. These devices are the foundation of smart spaces which aim at offering useful services like remote control of users’ homes or efficient energy usage across large buildings. However they also raise significant challenges, concerning for instance their security. In this tutorial, we aim at clearly defining some of those security challenges. To do so, we present a classical architecture to connect devices and applications together for commercial or industrial deployment of smart spaces. We then present for several
key points of this architecture the associated security needs. To meet those needs, current technologies are presented as well as proof of concept of recent research work in this area. The presentation also features a set of potential points of combinatory utilization of Physically Unclonable Function (PUF); a pervasively expanding hardware primitive, and machine learning methods, that can open in conjunction with each-other some interesting protocols for IoT security. To illustrate the presentation, a demonstration of a real-life IoT scenario illustrating the presented challenges conclude the tutorial. At the end of the tutorial, the attendees will have a clear understanding of the security challenges of IoT solutions and some ways to address them

Towards Operating Systems for the Real World

Organized by: Tatsuo Nakajima and Risa Kimura, Waseda University, Japan

Information technologies dramatically changed our daily life. Now, every day, we use the Google search engine to find everything we like to know. Also, when we travel to somewhere, we use Google Map to navigate us toward the destination place. Software is becoming a key element for adding values to everything. However, we currently do not know what are fundamental high level platforms for adding values in our daily life services. In this tutorial, after showing some backgrounds, we like to present three research topics, which we are recently working on, and aim to seek future fundamental high level platforms for adding values for software-based daily like services. The first research topic is called “alternate reality experience” that aims to guide people’s attitude and behavior, which will become future user interface component in the real world operating system. The second research topic is called “sharing everything” that aims sharing and protecting any physical resources in the real world, where sharing and protecting are essential functions in operating systems. The third research topic is called “physical embodiments” that aims to offer novel I/O device managements to interface to the real world, where physical actuation is an essential function that operating systems communicate with the real world. The above approaches will allow us to consider what kind of operating systems we can investigate for developing future advanced services. Finally, we like to show some future research agenda in this research.

Vehicular network systems in smart cities

Organized by: Edna Iliana Tamariz Flores, Autonomous University of Puebla, Faculty of Computer Sciences, Puebla, Mexico

The need for connectivity in motion has gained interest in vehicular network systems due to the
services they can provide in the transformation of transport through the Internet of Things (IoT)
application for a Smart City. The evolution of wireless network technology, protocols, and standards
in a vehicular network aims to transform transport towards a more intelligent transport that meets
users’ needs. Thus, the implementation of a vehicular network, through smart vehicles and road
infrastructure, allows vehicles or mobile nodes to communicate with each other to send information
that provides society with numerous services ranging from decreased traffic up to the users´
security. Therefore the wireless network in motion applied in the transport is unprecedented.
Innovative applications focused on improving lifestyles through smart transportation are
highlighted in this chapter. Besides, diverse literature in wireless networks is presented to enhance
connectivity between moving vehicles and the development of future intelligent transport systems
(ITS).
The motivation of this tutorial is the pursuit of research within the area of transport for smart cities
employing vehicular networks. Smart transport has become highly important and relevant to meet
the demand for services that society needs for a better lifestyle. The main objective of this tutorial
is to highlight the need for the development and innovation of more intelligent transport systems
that are capable of defining today’s future across vehicular networks. The contributions of this
tutorial are numerous—excellent literature aimed at the latest advances, challenges, and solutions
in mobility in vehicular networks. In addition, the literature shows protocols, standards, and
implementations to improve smart transport through different wireless technologies. agenda in this research.

GDPR: one small step for people, one giant leapt for our digital privacy – implications in software development and IoT scenarios

Organized by: Ricardo J. Rodríguez, Dpto. de Informática e Ingeniería de Sistemas, Universidad de Zaragoza, Spain

Do you know the GDPR? I’m pretty sure so, if I tell you that the annoying messages that fill your screen when you visit a website are motivated by the European General Data Protection Regulation (GDPR), which came into effect on May 25, 2018. The GDPR gives us new rights as citizens over our personal data, with the aim of improving transparency and trust in data collection and exchange procedures. In this tutorial, I will summarize the main benefits that the GDPR brings to our digital lives and the implications for organizations and entities that handle personal data of European citizens. In addition, I will present the principles of privacy by design, applied by the GDPR. Finally, I will describe an example of the impact of the GDPR on IoT scenarios and briefly present the Jean Monnet project “EU-GDPR: New data privacy regulation in the European Union – Impact on EU citizens and organizations (EUpriv8),” GA 611826-EPP-1-2019-1-ES-EPPJMO-PROJECT, in which I participate with colleagues from the University of León (Spain).

From Centralized to Distributed Machine Learning

Organized by:

Mikel Larrea, University of the Basque Country UPV/EHU, Spain
Iñigo Perona, University of the Basque Country UPV/EHU, Spain

Machine Learning (ML) is a branch of Artificial Intelligence that aims to develop algorithms that “learn” behaviors by using data or through experience. In many applications, data is collected massively, and in order to run ML algorithms in a reasonable time it is necessary to run them in a distributed computational environment. Moreover, in many current application scenarios (Internet of Things, Smart City, social networks, blockchain…) the information to be treated is inherently distributed. All this makes that many centralized ML solutions (based on a single computer) are being adapted to distributed environments. In turn, within distributed solutions, some solutions are designed to be executed in a cluster of computers, i.e., in a High Performance Computation environment, where one node is selected as a master and the others are workers. Other solutions are designed to be executed on top of distributed computers over the Internet. In the latter scenario, the most developed topology schema, according to scientific literature, is the one composed of one central aggregator node and multiple participant nodes. Nevertheless, there are solutions from a centralized schema, all the way through a schema where all the nodes have equivalent capabilities. 

In this tutorial a survey of this decentralizing path of ML algorithms will be presented, pointing out the algorithms that have been milestone and their most maintained implementations, including some demonstrations of these solutions. These demonstrations will show most promising frameworks to use in a distributed and heterogeneous system, which is the scenario where the Intelligent Environments are created. In this sense, some cases of use will be presented using a Federated Learning framework and also using a peer-to-peer solution.

 

Embedding lightweight software agents into single board computers for controlling IoT scenarios

Organized by:

José G.Caicedo-Ortiz, PhD student
Universidad de la Costa (Colombia)
Departamento de Lenguajes y Sistemas Informáticos (LSI), Universidad de
Granada (Spain)
Juan A. Holgado-Terriza, PhD
Departamento de Lenguajes y Sistemas Informáticos (LSI)
Universidad de Granada (Spain)
Pablo Pico-Valencia, PhD
Escuela de Sistemas
Pontificia Universidad Católica del Ecuador, Esmeraldas (Ecuador)

 

The Internet of Things (IoT) is an emerging paradigm that has gradually been
introduced in several areas to support the automation of tasks that solve real-world problems and help people have a better quality of life. In IoT environments, mechanisms that enable things, objects or devices to have greater autonomy, degree of collaboration and adaptation at runtime are increasingly necessary. In this sense, agents play an important role because they are entities with high communication capabilities that can modify their behavior according to environment changes and user and developer needs. In consequence, the agents should require reconfiguration or reprogramming for adapting to the novel conditions. The objective of this tutorial is to present two techniques to identify IoT ecosystems, that is, through embedded agents running on the devices themselves, and through Multiagent systems running in the cloud. More specifically, the proposed tutorial emphasizes the agentification technique using embedded software agents on the devices themselves. For this, it is proposed to use Raspberry Pi as the support to embed the agents, which in turn will be programmed in Python using the OsBrain tool. Through embedded agents, it is proposed to create proactive object networks to overcome the limitations of IoT devices, which are currently passive and can only be reconfigured under arguments defined by the manufacturer.

Telepresence in the Context of Intelligent Environments

Organized by:

Adrian Stoica (NASA Jet Propulsion Laboratory, USA)

Saeid Nahavandi (Deakin U., Australia)

Decades after first used to control operations in hazardous environments and in space, telerobotics, combined with autonomy, is getting ready to significantly impact our lives in major ways. From tele-health to tele-education and teleoperation in several industrial sectors, we are seeing the appearance of a large number of applications that will irreversibly change how we work and live.

The talk will overview technical aspects of telepresence. It will refer to perception of the remote environment via sensors some placed in the infrastructure or intelligent environment and some on robotic avatars. Information from these sensors creates the visual, auditory, and haptic perception which allows a human operator – or several of them, to observe and also to intervene at the remote location. Acting on the remote environment is usually done by robots, endowed with mobility and manipulation capabilities; their actions affect the natural or artificial environment (as in agricultural equipment or in warehouses) as well as on people (for example in telehealth applications). In addition, telepresence often involves transferring the physical appearance of the operator to the remote location. The simplest way, already common in deployed applications this is done by a visual/audio appearance through a screen mounted on the telepresence robot; more sophisticated solutions provide robotic avatars with arms which allow physical interaction with humans; holograms also have been used in applications such as teleducation.  Bandwidth limitations and delays impact teleoperations and require in many situations to have at least a minimal level of autonomy in the robotic avatar. Processing can be done locally or after being transmitted – all depending on tradeoffs between available power, computing, communication capabilities etc. The talk will also discuss services that are appearing as technology in research labs matures and products move to consumers. It will highlight the symbiosis between teleoperations and autonomy, and the role of AI. It will analyze the symbiotic relationship between telepresence and intelligent environments.