Flexible Data Access Mechanisms for IoT Ecosystems: From Content-Based Addressing to LLM-Driven Data Acquisition
Year: 2026
Type: preprint
Abstract: The continuous development and integration of public and private Internet of Things (IoT) infrastructures require flexible and scalable data access mechanisms that ensure seamless access to IoT data regardless of its source of physical address. Traditionally, retrieving data from different physical sources requires rigid, topology-dependent protocols and high technical skills; this may limit accessibility for a wider range of end-users.Content-Based Addressing (CBA) accesses data based on its attributes or content rather than its network location. Users need an intuitive mechanism to extract IoT values from multiple sources (e.g. public and private sources), removing the need for technical database query knowledge. In a similar way to web search (e.g. Google), users ask questions in natural language based on content, location and time without direct reference to the physical location of the IoT devices that provide the necessary information. This paper proposes a novel, hybrid framework for flexible IoT data access that combines CBA with a Large Language Model (LLM)-based conversational interface. The architecture is built on two core pillars: a Content-Addressable Storage (CAS) layer, coupled with a device registry, that unifies heterogeneous data sources, public APIs and private edge devices, into a single content-addressed store secured by a granular Role-Based Access Control (RBAC) mechanism. The second pillar is a Natural Language Query Engine that translates chat-based interrogations into executable system queries.An adaptive imputation service maintains data completeness by detecting missing sensor readings, predicting replacement values, and reconciling predictions when delayed measurements arrive. Together, the framework enables an intuitive experience for IoT data exploration with no technical training needed.
Source: Research Square
Institution: Technical University of Cluj-Napoca
Language: English
Topic: IoT and Edge/Fog Computing
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7236563162
