It was an interesting experience for me to finally visit the
PerCom conference itself for a first time, after having organized a workshop there quite some time ago [1]. It's hard to distinguish between ubicomp, pervasive, percom and many other second tier venues - but definitely the visitors at percom seem to be much more from an networking and sensor networks background.
The conference was opened with a fabulous keynote from Gregory Abowd. He was stating the fact that more and more scientific meetings center around ubicomp which inherently bears the danger of fracturization of our community. So the difference for papers at the major venues (ubicomp/pervasive/percom) really should be focus on a variety of perspectives instead narrow focus only. I also enjoyed his statement about stopping the quest for the killer-app and yet coming up with new scenarios but rather to listen to the users and domain experts in order to recognize needs and to try to tackle those. He reminded that also Marc Weiser focuses on human activities instead of machines. Gregory predicted that we would made three thirds of the way for making ubicomp as a reality. The interesting time of today is that dreams of 15 years ago can be realized now. Finally, he came up with following guidelines for successful ubicomp research:
- Include various disciplines - you can't do all yourself.
- Seek out critique: who are your friends motivating you, who are your enemies evaluating you and guiding you?
- Solving a real-world problem has the danger of NOT doing research - but it may rewarding anyway.
- Remember ubiquity: What can you assume in your home? Start from detectable events, e.g. power noise of switches.
- Make it matter, think about meaningful applications.
- Passion is not sufficient (but necessary): go one step beyond honorable causes
The second day was opened by another remarkable keynote given by Andrew Campbell. He set
out quite provocatively, especially for the percom audience: “Motes are dead: they not ubiquitous, suffer of power, people are out of the loop, they don't scale!”. Instead, he praised the emergence of mobile phones equipped with sensors that allow to study social relationships at a much bigger scale.
Andrew campbell
Motes are dead: not ubiquitous, power problem, people out of the loop, does not scale. He presented some examples from his own research, MetroSense [2], Bikenet [3], and Cenceme [4] - showing the capabilities of phones to detect context and to share on at larger scale. I found it striking that recognition of sitting, standing, walking - which has been shown years before [5] - now has a new quality and actually might become useful in the context of facebook. However, Andrew also had to admit that despite its ubiquity the mobile phone has not been quite build for sensing tasks, since after two hours of sensing the battery might be dead - so people could start de-installing such apps…
My question about his experience concerning adoption of his Cenceme was interesting: his students actually did not care about privacy, the off-button was an important design feature, control mechanisms of facebook allowing to grant access to friends he saw as the important features for getting it adopted. Most interestingly, Andrew himself said, he never would share sensor data about himself with others…
On the third day I was very glad of having been invited to the panel discussion about “Ubicomp is still not there, does nobody want it?” moderated by Christian Baecker. In my initial statement I said first of all we should be aware of the limitations of our research, it's not us driving adoption. If we want to do so we have to start listening more to users and stakeholders. Furthermore, it's not that bad, first devices such as Nabaztag, Tikitag, Choudy, picture frames with embedded WLAN actually show first instances of disappearing computing embedded in our homes. Cao build upon that and fostered the need for contuining research in systems that scale. Ty Zanati highlighted the need of understanding the nature of applications, focus on interoperability, privacy, and the seek for standards. Finally, Andrew Campbell was provoking again with questioning a common identity of the community: What defines us? Technology is still visible, what is our achievement, we need an identity. Then the discussion focused much about the prominent role of mobile phones: great platform and large scale adopted ubicomp device on the one hand but bearing the danger of occluding non-human related ubicomp apps on the other hand. After having a really appreciated discussion about that topic, Stephe Ward finally made a nice closing remark, taking out the almost religious belief in Mark Weiser: “The profound visions are those that disappear.”
Finally, I enjoyed two demos, one about robots controlled by RFID for providing a mobile simulation environment for testing WLAN
protocols [6], the other one about a mobile shopping assistant placing avatars on mobile phones giving product recommendations [7] which was close to what Felix von Reischach has presented at the conference [8]
[1]
Pervasive RFID/NFC Technology and Applications Workshop (
PerTec 2007),
www.autoidlabs.org/pertec[2] MetroSense,
http://metrosense.cs.dartmouth.edu/[3] Bikenet,
bikenet.cs.dartmouth.edu[4] CenceMe,
www.cenceme.org[5]
Junker, Holger and
Ward, Jamie A and
Lukowicz, Paul and
Tröster, Gerhard: User Activity Related Data Sets for Context Recognition, Proceedings of the Workshop on Benchmarks and a Database for Context Recognition, in conjunction with Pervasive Conference 2004.
[6]
RFID Based Localization for a Miniaturized Robotic Platform for Wireless Protocols Evaluation", Vikram P. Munishwar, Shailendra Singh, Christopher Mitchell, Xiaoshuang Wang, Kartik Gopalan, Nael B. Abu-Ghazaleh, Computer Science, Binghamton University (State University of New York) and S*ProCom2, The Cooper Union for the Advancement of Science and Art (USA)
[7] Social Mobile Augmented Reality for Retail, Sinem Güven, Ohan Oda, Mark Podlaseck, Harry Stavropoulos, Sai Kolluri, Gopal Pingali, IBM T. J. Watson Research Center (USA)
[8]
A Mobile Product Recommendation System Interacting with Tagged Products, F. von Reischach, D. Guinard, F. Michahelles, E. Fleisch,
Concise Contribution at the Seventh Annual IEEE International Conference on Pervasive Computing and Communications (PerCom 2009), United States, March 2009.
I'm just caming back from our half-yearly Auto-ID Labs meeting. It's always great meeting the colleagues from the other labs, exchanging ideas, getting new insights into on-going work, and to
consolidate the overall labs' strategy as such.
