Streamlining Aviation Operations: A Medallion Architecture Approach to Real-Time Flight, Baggage, and IoT Analytics

DOI

No DOI

Published In

Volume 7, Issue 2 (2026)

Published Online

March-April 2026

Pages

192-202

Article Type

Original Article

Check for updates

No DOI available for cross-reference

Today, international airports are one of the most complicated operational environments within the global transportation system. As an example, Skyhaven International Airport (SHI), which processes over 40 million passengers a year via 180 daily flights to and from three passenger terminals, illustrates the complex data management issues that face other high-volume aviation facilities worldwide. The use of older, siloed reporting methods at SHI have resulted in a three-day delay in obtaining On-Time Performance (OTP) data, only reactive identification of baggage misconnects, and the inability to predict terminal congestion events, effectively keeping SHI in an "Operational Blindness" condition. This paper describes the design, implementation, and anticipated outcomes of the Skyhaven International Airport Operations Intelligence Platform (AOIP). The AOIP platform is built on a multi-layer (Bronze, Silver, and Gold) architecture in a unified data Lakehouse model that has been deployed utilizing Databricks. The platform combines heterogeneous real-time and batch data streams from the Airport Operations Database (AODB), Kafka-driven gate management systems, the Baggage Information System (BIS), and approximately 1,000 IoT passenger counting sensor networks. Core technical contributions include the following: ingestion of Bronze layer via MERGE-based loading into Databricks Auto Loader with schema evolution support; Silver layer UPSERT pipeline for consolidated operational timelines; delay propagation engine tracking aircraft registration at both inbound and outside legs to predict cascading delays 2 hours prior to departure; SLA-validated turnaround risk assessment by aircraft type - narrow vs. wide-bodied, with corresponding minimums of 45 and 90 minutes respectively; and PySpark 15-minute Tumbling Window aggregations over approximately 1.4 million daily IoT sensor readings for terminal congestion detection. Projected results from the resulting platform will include a 30% reduction in uncontrolled baggage connections, improved gate utilisation from 40% idle baseline to sub 25% target, and automated DGCA compliant OTP reports generated by 06:00 daily - effectively transforming SHI from reactive operations hub into Predictive Airport.