Challenges in ubiquitous context recognition with personal mobile devices

Abstract

Ubiquitous computing applications are required to provide distraction-free task support by reacting on different context characteristics. With the wide-spread use of personal mobile devices, many users are in possession of a powerful platform for context recognition. This, in theory, should allow the recognition of a number of characteristics which a user experiences during the course of daily routine. However, in order to be suitable for personal mobile devices, existing systems are considering a small and static set of characteristics for a particular application. This enables the developers to manually optimize their systems. Yet, it limits the applicability of the systems to narrowly defined scenarios. We argue that context recognition systems must take heterogeneity into account in order to be practically applicable to ubiquitous computing on a large scale. Specifically, future systems must find ways to accommodate the heterogeneity of tasks and users which results in three novel research challenges, namely the dynamic integration, privacy-preserving cooperation and automatic personalization of context recognition systems. In this paper, we motivate these challenges and outline ways to address them.