Transparent use-case management for AIoT devices
Jeff Will, Hanno Homann · 2024
This paper presents an understandable approach to managing use-cases for Artificial Intelligence of Things (AIoT) devices in smart home environments, with a focus on balancing user privacy, functionality, and convenience. Recognizing the growing integration of smart home technologies and AI, particularly in aiding elderly individuals, we propose a transparent, user-centric model that allows individual users to control the extent of data transmission and processing. We categorize users into three distinct groups: privacy-oriented users, who prefer local data processing; convenience-oriented users, who favor cloud-based AI services for natural language processing; and technically-informed users, who desire a transparent choice between local and cloud-based processing.(p)(/p)We demonstrate the feasibility of this approach with a smart home assistant prototype, showcasing the flexibility in use-case management and its potential to cater to diverse user preferences. Our model employs a RaspberryPi-based smart home assistant with sensor integration, capable of both local processing as well as interfacing with cloud-based Large Language Models (LLMs) like ChatGPT for complex query handling. The system architecture supports MQTT protocol for IoT device communication and ensures data security through robust encryption protocols. (p)(/p)This work contributes to the discourse on smart home technologies by offering a framework that respects user autonomy in data privacy while harnessing the benefits of AI, thereby enhancing the acceptance and effectiveness of smart home systems in modern living environments.