SensorsConnect: World Wide Web for Internet of Things

Abstract

The widespread adoption of the Internet of Things (IoT) has led to a surge in smart sensing devices connected to the Internet. While IoT enables machines, embedded systems, and appliances to access the Internet, they do not interact with it as humans do through the World Wide Web (WWW). Unlike humans, IoT devices lack a unified framework like the WWW for collaboration and data sharing. This is primarily due to (1) the separate infrastructure often required for IoT security and privacy and (2) the challenges of limited connectivity, device heterogeneity, and evolving technology. This thesis presents SensorsConnect, a platform that connects IoT devices in a WWW-like framework, enabling real-time sensing data searches across a broad IoT context. It defines the architecture, core processes, significant challenges of SensorsConnect, and strategies to address them. In addition, the thesis presents a motivating scenario illustrating its potential impact in real-life situations. Using real-time road status and service occupancy estimated by Google Maps, SensorsConnect enhances Google Maps service recommendations, not only minimizing customer wait times but also distributing workload more evenly across service points. Additionally, spotting the light on the search component of SensorsConnect, the thesis introduces a real-time IoT Agentic Search Engine (IoT-ASE) that leverages the Large Language Models (LLM) models and Retrieval Augmented Generation (RAG) techniques. It also introduces the IoT-Retrieval Augmented Generation Search Engine (IoT-RAG-SE) and deploys it as an agent within IoT-ASE. Furthermore, it discusses a use-case scenario where IoT-ASE is hypothetically deployed in Toronto to improve the quality of service recommendations by leveraging real-time IoT data. IoT-RAG-SE retrieves the intent service as the first-ranked result by 92%. Additionally, the evaluation highlights a sample of queries to assess the performance of IoT-ASE compared to Gemini. IoT-ASE responses are more concise, relevant, engaged, and persuasive than Gemini responses, which tend to be generalized without fully addressing preferences embedded in queries. Furthermore, having access to real-time data provides an additional privilege to IoT-ASE to generate responses based on real-time context. Our discussion reveals that this promising framework could improve public access to live data, support real-time decisions, and improve quality of life.