EXPLORING DATA EXHAUST IN IOT DEVICES WITH A FOCUS ON VOICE ASSISTANTS
Abstract
In the realm of the IoT, the notion of data exhaust becomes pivotal, representing unintentional data generated during device interactions in the digital domain. This inadvertent data holds valuable insights for third parties interested in understanding user behavior. Despite this interest, our survey of 48 research papers revealed a surprising dearth of focus on data exhaust, particularly across different types of IoT devices. Our primary objective is to bridge this knowledge gap by offering a nuanced analysis of data exhaust issues.
Simultaneously, our thesis addresses the critical challenge of data privacy by developing a predictive modelling scheme. Focusing on Voice Assistants, we delve into the ecosystem surrounding these devices, emphasizing the lack of user awareness regarding data collection. Leveraging real-world data from Amazon Alexa Traffic Traces, our proposed approach employs machine learning methods to predict data exhaust. This method enhances user awareness, contributing to a more privacy-centric IoT landscape.