A Comparative Analysis of Data Models for Heterogeneous Sensor Data Management

In today’s technological landscape, data has become a cornerstone for numerous applications, witnessing exponential growth driven by the advent of sophisticated devices capable of generating and analyzing user data daily. This surge in data volume has spurred a competitive push among industries and firms to develop scalable applications designed to efficiently take advantage of this wealth of information to offer services that capitalize on the vast data repositories. Simultaneously, the Internet of Things (IoT) has seen a remarkable rise in popularity, resulting in the expanding range and significance of sensor applications. Sensors now play a pivotal role across various domains, providing real-time insights into the monitored subjects’ status. This paper introduces a comparative analysis between different data modeling techniques using a unified schema for heterogeneous IoT sensor data. This analysis aims to identify the most suitable approach for handling both structured and unstructured data within a sensor integration framework. Performance evaluation shows that the document data model, implemented by MongoDB, demonstrates superior efficiency in the READ, UPDATE, and DELETE operations under high data loads compared to traditional relational models. Despite a slower performance in CREATE operations due to overheads associated with index creation and document structuring, the document data model still maintained superiority over other approaches. Additionally, when compared with various CRUD operations, the document data model showed increased throughput with higher workloads, outperforming both the column and the EAV data models.