Aggregating Life Tags for Opportunistic Crowdsensing with Mobile and Smartglasses Users

Uncovering Practical Security and Privacy Threats for Connected Glasses with Embedded Video Cameras

“…By adopting this definition, we follow the classification of Kress et al [52] from their discussion of various types of eyewear and the segmentation of the Head-Mounted Displays (HMD) market. 1 The key word that specifies the scope of our investigation is "connected," where the application running on the glasses needs to stream image or video data as part of its functionality, e.g., for the purpose of lifelogging [38,39], linking to social networks [53], or requiring access to specific services, such as image classification in the cloud [1,54]. We specifically focus in this work on security threats involving the wearers of connected camera glasses, such as device discovery, user tracking, unauthorized access, denial of service, video sniffing, and video hijacking; but some of these threats, such as video sniffing, may equally affect bystanders, who may be rightfully concerned about being video-recorded in public settings without their consent [23,38,39,43,48,50] with consequences about compromising their privacy and security as illustrated with Andrew's story.…”

“…Aiordăchioae and Vatavu [3] designed "Life-Tags, " a wearable camera-based system for lifelogging that, instead of storing photos and videos, abstracts them in the form of clouds of tags and concepts automatically extracted from live video. A follow-up work [1] extended this approach to mobile crowdsensing involving both glasses and smartphones. However, because of the dependency of these lifelogging systems [1,3] on external services from the cloud to classify images and extract concepts, they are exposed to possible breaches of security and privacy requirements that we present in detail in the next section, such as video sniffing, video hijacking, and denial of service.…”

“…A follow-up work [1] extended this approach to mobile crowdsensing involving both glasses and smartphones. However, because of the dependency of these lifelogging systems [1,3] on external services from the cloud to classify images and extract concepts, they are exposed to possible breaches of security and privacy requirements that we present in detail in the next section, such as video sniffing, video hijacking, and denial of service.…”

“…Lastly, our investigation has specifically addressed glasses devices that stream video for various purposes, such as lifelogging records [38,39], linking to social networks [53], or that require access to services from the cloud to operate [1,54]. High-end HMDs do not need to offload video processing to a connected device since they embed the required resources for on-board video analysis.…”