Semantic model-driven development for IoT interoperability of emergency services

Abstract

Disaster Risk Reduction (DRR) is a systematic approach to analyze potential disasters and reduce their occurrence rate and possible impact. The main DRR component is an Early Warning System (EWS), which is a distributed information system that is able to monitor the physical world and issue warnings if abnormal situations occur. EWSs that use Internet-of-Things (IoT) technologies, so called IoT EWS, are suitable to realize (near) real-time data acquisition, risk detection and message brokering between data sources and information receivers, comprising both humans (e.g., emergency managers) and machines (e.g., sirens). Over the last years, numerous IoT EWSs were developed to monitor different types of hazards. Multi-hazard EWSs are a special class of EWS that can detect different types of situations and, if necessary, react to them. Multi-hazard EWSs require integration of existing EWSs and seamless integration with new EWSs. Interoperability of EWS components is necessary for effective integration, e.g., so that sensors, devices and platforms work with each other and with other EWSs. Although IoT technologies offer possibilities to improve the EWS efficiency and effectiveness, this potential can only be exploited if interoperability challenges are addressed at all levels. In this thesis, we focus on how to improve the semantic interoperability of IoT EWSs. Semantic interoperability refers to the ability of two or more EWSs (or EWS components) to share data elements in a prescribed format (syntax) and precise unambiguous meaning (semantics). From a literature review on semantic IoT EWS approaches, we selected the three major challenges that need to be addressed together: 1) semantic integration of a variety of data sources that make use of different standards, ontologies and data models; 2) near-real-time processing in time- and safety-critical applications; and 3) data analysis for effective situation awareness and decision support. This thesis introduces the “SEmantic Model-driven development for IoT Interoperability of emergenCy serviceS” (SEMIoTICS) framework, which is a holistic approach for semantic IoT EWS. SEMIoTICS consists of a semantic model-driven architecture that facilitates the application of data representations, model transformations and distributed software components. SEMIoTICS is a framework that can be used to develop interoperable IoT EWSs for different domains, enabling an IoT EWS to act as a cloud-based semantic broker for situation-aware decision support. SEMIoTICS is leveraged by the adoption of ontology-driven conceptual modelling for situation-aware applications, covering both EWS design-time (specification and implementation) and runtime. Furthermore, the SEMIoTICS guides the application of the Findable, Accessible, Interoperable and Reusable (FAIR) data principles, in which the role of standardization is emphasized. SEMIoTICS was validated in the context of the H2020 INTER-IoT project, in which a semantic interoperable IoT EWS was developed to detect accident risks with trucks that deliver goods at the Valencia port area. The research in this case study addresses the semantic integration of a variety of data sources with processing in safety-critical applications for effective emergency response. The solution considers existing domain-specific ontologies and standards, along with their serialization formats. In this case study, accident risks are assessed by monitoring two types of data, namely (1) the drivers’ vital signs with electrocardiogram (ECG), and (2) the trucks’ position, speed and acceleration. The case study includes the detection of health issues with drivers and collisions with vehicles with dangerous goods. A special result of this research for this case study is SAREF4health, which extends the European semantic standard for IoT (Smart Appliances REFerence, SAREF) with the representation of ECG data. The framework has been validated in three ways with respect to non-functional aspects: (1) an analysis of the accuracy and efficiency of the semantic translations, (2) an analysis of the communication efficiency of JSON for Linked Data (JSON-LD) for IoT scenarios, and (3) an analysis of the scalability of the semantic brokering between data sources and information receivers. The most important contributions of this thesis are: • Improved IoT Semantic Interoperability: (1) semantic translations between IoT standards (W3C SSN/SOSA and ETSI SAREF) for semantic integration; and (2) SAREF4health, as the first extension of SAREF for the healthcare domain; • Improved Situation Identification for IoT EWS: higher semantic expressiveness with a new version of the Situation Modelling Language (SML) and Complex Event Processing (CEP) technology; • Interoperability reference for disaster services: improved reference architecture validated with an open source cloud-based IoT EWS for ECG monitoring.