2026-10-14T10:06:23Z https://api.figshare.com/v2/oai

oai:figshare.com:article/34029933 2026-09-30T05:12:53Z category_26557 category_26560 item_type_12 month_year_09_2026

<b>Empirical Performance Evaluation of Heterogeneous IoT Sensor Clusters for Real-Time Industrial Gas Dispersion Tracking</b> Davina Liberata Dallo (25141658) Air pollution modelling and control Environmentally sustainable engineering gas dispersion tracking; heterogeneous sensor clusters; instrumental odour monitoring; field performance evaluation; metal-oxide sensors; industrial IoT. <h2 dir="ltr"><i>Heterogeneous sensor clusters are now the standard instrument for tracking industrial gas dispersion, yet the field lacks an evaluation framework that measures what these clusters actually deliver. Existing evidence is fragmented in a specific and consequential way: performance is reported per device, per algorithm, or per isolated experiment, while the quantity that determines whether a monitoring system is useful is the performance of the cluster as a tracking instrument under real emission and meteorological conditions. This article assembles the empirical record and shows that it supports stronger conclusions than are usually drawn from it. Published field evaluations establish that wireless gas detection networks can achieve detection rates of 91% with average detection delays of 108 s and localization accuracy of 5 m; that instrumental odour monitoring systems can exceed 95% classification accuracy once humidity and temperature interference is compensated; and that metal-oxide sensor output can be regressed against field olfactometry with RMSE between 4.02 and 6.70, depending on the regression technique. Read together, these results identify three conditions that determine cluster-level tracking performance reference anchoring, interference compensation, and deployment geometry informed by dispersion behaviour and one that is almost never reported, namely the temporal coherence of the cluster. We therefore propose an evaluation framework organized around four dimensions, each with an explicit protocol and reference metric, and derive testable propositions from the reported evidence. The framework's central claim is that a cluster should be evaluated as a distributed measuring instrument, not as a collection of devices.</i></h2><p></p> 2026-08-31T00:00:00Z Text Preprint 10.6084/m9.figshare.34029933.v1 https://figshare.com/articles/preprint/_b_Empirical_Performance_Evaluation_of_Heterogeneous_IoT_Sensor_Clusters_for_Real-Time_Industrial_Gas_Dispersion_Tracking_b_/34029933 CC BY 4.0