Section: New Results

FIESTA-IoT Ontology: Semantic Model for Federation & Interoperability among Platforms

Participants : Rachit Agarwal, Valérie Issarny, Nikolaos Georgantas.

Plethora of heterogeneous data is being generated and made available by diverse platforms. Such platforms can be those that are formed by the use of mobile application that act as interface between sensing devices and storage or between users and storage. The diversity and openness in the data generated isolate platforms and lead to interoperability issues between platforms, where much work has to be done in order to ensure compatibility. One has to understand the other's format, parse different data formats, and create the mapping between different data formats. One method to accomplish this interoperability is by attaching semantics to this data. Semantics provides meaning to the data and helps in (a) achieving common understanding and (b) performing analysis and reasoning. Many IoT-related semantic models (http://sensormeasurement.appspot.com/?p=ontologies) propose interoperability but have many issues like: observation graph is missing, are highly domain specific, and do not follow best practices. In order to address the above, we focused our research on: the identification of a unified semantic model that addresses the above, creation of a prototype application, and identification of guidelines for storing semantic data [13]. We report our following key results:

• State of art survey of semantic models that are available in literature in the domain of the Internet of Things: This survey gave us required knowledge needed for the semantic model from which concepts can be reused to create a unified ontology. This helps the semantic community by not overloading the domain with concepts similar to already existing concepts, and allows us to reuse concepts as much as possible. We identified that recent trends show more and more use of the SSN  [35] and oneM2M  [67] ontologies. However, these models are currently far from being able to address observation-related issues and lack domain taxonomy.

• Unified semantic model for enabling interoperability and federation of testbeds: Based on the analysis of the concepts from various onotologies identified, we unify specific concepts from these identified ontologies into one ontology. These ontologies being: SSN, oneM2M, IoT-lite  [27], WGS84 (https://www.w3.org/2003/01/geo/), DUL (http://www.ontologydesignpatterns.org/ont/dul/DUL.owl), TIME (https://www.w3.org/TR/owl-time/) and M3-lite taxonomy (created as a part of this research). Such unification gives our ontology the power to define meta data about the sensor that is producing the observation and the observation itself. The federation is achieved by the use of the taxonomy that each platform should follow.

• Best practices to publish data based on the unified model: In order to enable full interoperability, federation and usage of data, it is essential that best practices are followed while storing the data based on the unified model. We identify various best practices which form our recommendations to the platform owners towards annotating the data with respect to the ontology. This is supported by a reference annotator that also acts as a guide for developers to publish data.

These above-mentioned results are currently applied in the frame of the EU funded H2020 FIESTA-IoT project (see § 8.2.1.2).