A Strategy For Data Quality Assessment In IoT-based Air Quality Monitoring Systems

Abstract

ABSTRACT: With the upcoming growth of IoT, which is translated into millions of interconnected devices reporting a high volume of data coming from heterogeneous sources (sensors), it is necessary to assess the confidence of the data in order to provide the system with trustable information that can be used to get real insights from the physical world and thus take proper decisions or actions over it. Having in mind that ensuring data quality is key to ease user engagement, acceptance of IoT services and large scale deployments [1], a new critical issue arises which is related to the quality of the data in IoT applications. In order to get a correct insight or interpretation of the physical world, IoT users and upper layers need to be provided with reliable data and also need to be able to judge whether the data is reliable or not. Moreover, different users and applications will have different requirements and conceptions of data quality, making of this a subjective matter that should be included in the analysis. In this research, we investigate the data quality term and how it has been treated over several studies and applications, aiming at the definition of a set of attributes and metrics to assess quality of data in the Internet of Things. We also investigate on how to integrate objective measurements and subjective perceptions of data quality, to provide a single index that informs about the data quality status of the system. Our approach is implemented in a software tool, which is evaluated on a synthetic dataset to test its awareness to induced data quality changes. The tool is also tested with a real dataset retrieved from the SIATA’s citizen science system, an air quality monitoring application that can be encompassed within the IoT paradigm, and that is composed by more that 230 nodes deployed all over the Aburrá Valley in Antioquia, Colombia. The results show that feasibility assessing data quality and the importance of data quality awareness for an IoT application, as a way for it to take proper actions on the real world.