Platys: from position to context

There is increasing recognition of the centrality of delivering high-quality user experience in networked applications. Conventional approaches to user experience tend to equate it with network performance measures such as throughput and latency.  However, user experience is more than just performance, and naturally incorporates considerations of user context. Although a user’s context is inherently a high-level concern centered on meaning, it turns out to depend crucially on low-level concerns centered on devices and infrastructure, and poses challenges to network architecture.  Thus context is a natural topic for cross-layer, indeed interdisciplinary, study—and consequently is well aligned with the goals of the present solicitation.

This project considers elements of context that are particularly related to mobile computing, and which can be exploited in many applications. Because mobile devices are always with a user, they offer a great opportunity for capturing some of the key aspects of context, namely, the user's location or other characteristics of the user’s environment through what is termed localization.  In current practice, the user's location is captured at the level of position, i.e., geospatial (latitude-longitude) coordinates. What matters is the user’s place: a location in conceptual terms such as “in a study group meeting,” “jogging,” or “grocery shopping”—descriptions that combine a set of positions with the user’s activities, properties of the user’s environment, and the activities of the users surrounding or interacting with the user.

To realize such a notion of place requires that information from devices and infrastructure flow in ways unanticipated in current network architectures. In connection with collaboration, it involves ways to enable opportunistic interactions while preserving the users’ privacy and designing incentive mechanisms to ensure cooperation by all without exploitation of any. These are inherently architectural concerns—they involve interconnections among components. Yet they lie far beyond traditional network topics such as routing.