Edinburgh Napier University

  The rise of Internet of Things has provided platforms mostly enhanced by real-time data-driven services for reactive services and Smart Cities innovations. However, IoT streaming data are known to be compromised by quality problems, thereby influencing the performance and accuracy of IoT-based reactive services or Smart applications. This research investigates the suitability of the semantic approach for the run-time validation of IoT streaming data for quality problems. To realise this aim, Semantic IoT Streaming Data Validation with its framework (SISDaV) is proposed. The novel approach involves technologies for semantic query and reasoning with semantic rules defined on an established relationship with external data sources with consideration for specific run-time events that can influence the quality of streams. The work specifically targets quality issues relating to inconsistency, plausibility, and incompleteness in IoT streaming data.
In particular, the investigation covers various RDF stream processing and rule-based reasoning techniques and effects of RDF Serialised formats on the reasoning process. The contributions of the work include the hierarchy of IoT data stream quality problem, lightweight evolving Smart Space and Sensor Measurement Ontology, generic time-aware validation rules and, SISDaV framework- a unified semantic rule-based validation system for RDF-based IoT streaming data that combines the popular RDF stream processing the system with generic enhanced time-aware rules.

The semantic validation process ensures the conformance of the raw streaming data value produced by the IoT node(s) with IoT streaming data quality requirements and the expected value. This is facilitated through a set of generic continuous validation rules, which has been realised by extending the popular Jena rule syntax with a time element. The comparative evaluation of SISDaV is based on its effectiveness and efficiency based on the expressivity of the different serialised RDF data formats.

The results are interpreted with relevant statistical estimations and performance metrics. The results from the evaluation approve of the feasibility of the framework in terms of containing the semantic validation process within the interval between reads of sensor nodes as well as provision of additional requirements that can enhance IoT streaming data processing systems which are currently missing in most related state-of-art RDF stream processing systems. Furthermore, the approach can satisfy the main research objectives as identified by the study.