Edge-Cloud Integrated Data Warehousing Architecture for Efficient IoT Data Analytics

The rapid proliferation of Internet of Things (IoT) deployments has generated
urgent demand for data management architectures capable of reconciling the
competing requirements of low-latency local processing and comprehensive
cloud-scale analytics. This paper proposes the Edge-Cloud Integrated Data
Warehousing Architecture (ECIDWA), a four-layer hybrid framework that
employs metadata-driven query routing between distributed edge data marts and
a centralised cloud data warehouse. ECIDWA directly addresses the principal
limitations of both pure-cloud and pure-edge designs: the high bandwidth
overhead and round-trip latency of cloud-only approaches, and the constrained
computational capacity and limited analytical coverage inherent in edge-only
deployments. A lightweight orchestration engine dynamically classifies incoming
queries as either latency-sensitive or analytics-intensive and routes them to the
appropriate execution tier. The architecture is evaluated against cloud-only and edge-only reference
implementations using a controlled experimental protocol comprising 500 independent runs per architecture
across seven performance dimensions. Results indicate statistically significant improvements on all metrics (one
way ANOVA; all F > 1,800; all p < .001). Relative to the cloud-only baseline, ECIDWA achieves a 43.8%
reduction in query latency, a 37.7% reduction in bandwidth consumption, a 28.0% improvement in throughput,
and a 44.9% reduction in fault recovery time, while sustaining a query accuracy of 97.6% (SD = 0.8%). Scalability
analyses confirm that these advantages are maintained under concurrent device loads of up to 10,000 nodes.
Collectively, these findings position ECIDWA as a robust, production-ready framework for industrial IoT, smart
city, and real-time monitoring applications.