DQ-MAN: A tool for multi-dimensional data quality analysis in IoT-based air quality monitoring systems

Air quality monitoring has traditionally been performed using robust specialized systems based on an air filter. These systems provide high quality data, but entail a high investment, thus limiting the scale of the deployment. An alternative way of measuring air pollution is the use of optical sensors, which are mounted on an embedded system, leading to a lower cost, as compared to the traditional solution. While these systems allow for a wider deployment at a lower cost, there is a concern on the quality of the data provided by them. In this context, the analysis of Data Quality (DQ) takes special relevance, in order to meet the requirements established by environmental agencies. In order to tackle this issue, this paper proposes a multi-dimensional model that estimates a unified DQ index, based on the integration of the relevant DQ dimensions and the subjective preferences of experts in the field. We present the development of DQ-MAN, a tool that allows the end-user to assess and visualize the DQ metrics over different time frames, and to compute the corresponding DQ index. Our tool allows the user to publish the summarized results in a web report. We validate DQ-MAN using a synthetic dataset to assess the correctness of our tool, as well as a real dataset of a low-cost monitoring system deployed in Medellín, Colombia. Based on the evaluation, we conclude that DQ-MAN is aware of changes in DQ, and how each dimension affects the overall DQ assessment. [Display omitted] •Key Data Quality dimensions for air monitoring systems have been identified.•We propose a DQ index, integrating multiple DQ dimensions and expert preferences.•We created a tool to evaluate DQ index, and applied to a low-cost sensor network case.