This year's UbiSys workshop focused on the 4 interrelated themes of common abstractions, new and emerging middleware and software engineering approaches, wide scale inter-domain deployment and techniques and benchmarks for effective evaluation of UbiComp systems. Due to the large size (47 participants) the day started with a 1 minute madness where all participants were able to state their position on these issues and then continued with a mix of 2 paper sessions and 3 panels. During the course of lively discussion a number of key issues emerged. The issue of a common set of abstraction for Ubicomp and the possibility of the community adopting a toolkit or platform to facilitate research was the subject of lively debate. Many felt this would help address a second issue that was frequently raised, namely the problem of evaluation and benchmarking of Ubicomp systems research. A key aspect of this, described as ‘magic in ubicomp systems’ was the problem that many researchers failed to adequately explain their assumptions and relied on some ‘magic’ in the future to help justify their approach. An equally important issue discussed was the poor record that the community has in addressing resilience and reliability. This was perceived by many to be a stumbling block to real world deployment of Ubicomp technologies. As an antidote to this problem, a well supported suggestion was that increasingly UbiSys researchers should strive to reuse their own work and to try and reduce the amount of reinvention and replication within the community. The meta-theme that evolved during the day was the need for a better co-ordination amongst the community which was evident in the last panel that focused on common scenarios to aid evaluation as well as the need to develop a community supported set of ‘grand challenges’.

Organizers: Rodger Lea and Mike Blackstock, University of British Columbia, Canada

12:15 - 1:30 Lunch and Posters
Posters will be on display in the workshop area and presenters available for discussion.