Argyro Mavrogiorgou

Plug‘n’play IoT Devices: An Approach for Dynamic Data Acquisition from Unknown Heterogeneous Devices

The Internet of Things (IoT) is being hailed as the next industrial revolution and promises billion of IoT devices connected to the Internet in the near future. Emerging back-end support technologies not only have to anticipate this dramatic increase in connected devices, but also the heterogeneity of devices. This paper proposes a comprehensive approach for efficiently and rapidly integrating heterogeneous devices during runtime, in order to collect data. The proposed approach identifies both known and unknown devices’ specifications and Application Programming Interfaces (APIs), classifies the specifications and illustrates the APIs into ontologies, while the latter are being mapped in combination with the classification’s results to match the known with the unknown devices’ APIs. All of these steps are being performed as preliminary steps for the proposed Dynamic Data Acquisition API, which is the final outcome of the mechanism that is responsible for translating the devices’ APIs into a common format.

Towards a Secure Semantic Knowledge of Healthcare Data Through Structural Ontological Transformations

Current devices and sensors have revolutionized our daily lives, with the healthcare domain exploring and adapting new technologies, promising high quality of care. The diversity of healthcare data is leading to the independent operation of the latter, whilst the value emerging from their exploitation is limited. Most of the data is confined in data silos, without meeting the requirements of standards, and secure data exchange. Healthcare systems need to be able to communicate and exchange data anonymously, for better understanding about the results of prevention strategies, diseases, and efficiency of patient pathway management. Since healthcare interoperability is the only sustainable way to address these constraints, several techniques have been developed but they are partially applicable, rising the needs of a generic solution. The increased use of Electronic Health Records (EHRs) requires ontologies to capture domain knowledge, providing the basis for agreement within the healthcare domain. This manuscript focuses on the semantic interoperability of multiple EHRs – and their standards, proposing a way for primarily anonymizing the personal data stored into the EHRs, and then transforming the EHR datasets into XML Syntactic Models, for getting their semantic knowledge through their ontological representation.

A Generic Approach for Capturing Reliability in Medical Cyber-Physical Systems

Cyber-physical systems (CPSs) are slowly emerging to dominate our world through their tight integration between the computational and physical components. While the reliability evaluation of physical systems is well-studied, the one referring to CPSs is difficult due to the fact that software systems do not degrade, as they follow a well-defined failure model like in physical systems. Henceforth, a great attention has been given to tackle the challenge of reliability in CPSs, especially in the field of Medical CPSs (MCPSs) that are being considered as a powerful candidate for healthcare applications. This paper proposes a generic approach for effectively measuring reliability in MCPSs, taking into consideration the multiple MCPSs’ applications that exist. The proposed approach captures the MCPS’s reliability by initially modelling its components, accompanied with the selection of the evaluation environment, which is finally being followed by the failure analysis, and the reliability estimation, which are necessary for deciding whether a MCPS is considered as reliable or not.

A Comparative Study of Classification Techniques for Managing IoT Devices of Common Specifications

As information technology and telecommunication systems continue to grow in size and complexity, especially with the Internet of Things (IoT) domain that is being hailed as the next industrial revolution, emerging technologies have to anticipate this dramatic increase of heterogeneous connected devices. This paper proposes a solution that can be used to manage this huge number of devices, by classifying them and predicting their device’s type, based on their specifications. Four (4) classification algorithms are being applied on a dataset containing the specifications of known devices (in terms of known device type), which is being used for predicting the unknown devices’ types. These algorithms are analyzed using the WEKA data mining tool and a comparative study is undertaken to find the classifier that performs the best analysis on the dataset obtained, using a set of predefined performance metrics to compare the results of each classifier.

Acquiring the Ontological Representation of Healthcare Data Through Metamodeling Techniques

The recent trends in healthcare ICT technologies promise high quality of care, whereas the variety and diversity of healthcare data increase the challenges of information exchange. With healthcare systems facing poor communication and information exchange difficulties, interoperability has become the holy grail of Health IT. While several techniques and researches are conducted to face this challenge, a global solution is still missing, as only particular scenarios are being currently addressed. This paper emphasizes on achieving semantic interoperability across multiple electronic health records (EHRs), through ontologies and Model-driven Engineering techniques. A multi-stepped approach is proposed that transforms EHR datasets into syntactic models and metamodels, in order to be finally transformed into their semantic structure through a mechanism that translates them into a common ontological representation.

A Comparative Study in Data Mining: Clustering and Classification Capabilities

The ICT evolution has driven on the creation of a capable society, in providing new kinds and type of information. The gathered information is stored continuously, meaning that a great amount of databases has to be created. The problem that arises is whether there is a global manner of managing and gaining knowledge out of the rising variety and volumes of data. Many efforts have been developed for addressing the emerging challenges of data mining based on statistics and machine learning techniques that can significantly boost the ability to analyze data. In this paper, a detailed study on the data mining field takes place, followed by a comparative study between clustering and classification techniques, resulting that the integration of clustering and classification techniques can provide more accurate results than a simple classification technique that classifies datasets with priorly known attributes and classes.