Wednesday, November 12, 2008

University mobile phone research

Second of four parts: Part 1 (teaching mobile phones), Part 3 (sensor research), Part 4 (Georgia Tech).

I was extremely lucky to spend the last two weeks of October (after classes were over) visiting some of the top U.S. research universities to hear about how they do research and teaching using mobile phones. It is a real eye-opener to see what’s going on in universities.

Once upon a time, mobile phones made phone calls, but now researchers are treating them as mass-produced mobile computers with large LCD screens and the ability to return data. Did I also mention that these are devices that thousands of readily available test subjects (18-22 year-olds) will gladly carry in their purse or pocket?

On Oct. 22 I visited UCLA, then on Oct. 29-31 I went with representatives of Symbian to MIT, Dartmouth and Georgia Tech. We did not have time to visit all schools that have active research programs on mobile phones, which would also include Stanford, Berkeley and UCSD.

Most (but not all) of the research is going on with the highest-end phones, like the Nokia N95 or iPhone 3G. We came across three platforms and vendors over and over:

  • Nokia. For many researchers, the Nokia N95 was their favorite phone because of its capabilities. But there are also the people: the (PhD-ridden) Nokia Research Center labs in Palo Alto and Cambridge are actively seeking out research collaborations with a small set of universities like Stanford and MIT. The Symbian OS with the S60 UI is also the world’s most popular and tested smartphone platform with the widest range of handsets.
  • Microsoft. As with Nokia, Microsoft Research is also filled with PhD researcher seeking collaborations with top universities. This is the natural choice for those owning Windows laptops, as the Windows APIs are familiar to many programmers and there are a variety of devices.
  • Apple (i.e. the iPhone) This is mainly driven by the pull from the students (or faculty) that went and bought iPhones on their own. It also has fans (particularly among Mac owners) because it runs Unix and has a good Mac/iPhone cross-development environment. Apple provides some limited support for university software development — a customized version of its app store.
We saw some interest in BlackBerry, Linux phones, and some projects that ran on the generic JME-enabled handset. A year from now, I’d expect Android to have a major presence — given that it’s open source and has Linux APIs — but until recently there was no source, documentation or devices available.

Other key issues:
  • Handset Costs. Most of this research is funded by competitive research grants (typically NSF), while others are industry or DoD funded. Some researchers get the phones donated by the manufacturers while in other cases they have a budget to buy them. Free phones are much more important when they are being given to a 100 test subjects than if there’s just one or two demo units.

    The other way to do a large-scale deployment is to use the students’ own phones, either through the Java least common denominator or by targeting a single popular platform (which here in the US is the iPhone).
  • Airtime costs. Researchers also need a way to pay for the air time for their phones — in most cases, involving expensive data plans. Strikingly, I didn’t see a single CDMA phone (probably something I would have seen at UCSD), so this becomes an issue of who’s paying for the SIM card to activate the phone. Occasionally, they get could get SIM cards donated by the phone makers, but this is pretty expensive. In other cases, experiments can be done on the Wi-Fi networks owned and operated by most university campuses.
When looking for mobile phone research, there is the question of the theoretical discipline. At least in the US, there are few professors of mobile phones: communications groups (within E.E. departments) would study new radio technologies, but this wouldn’t be done using existing mobile phones.

Instead, the action we saw was in software applications running on the phones. I identified four major fields within computer science:
  • sensor networks: in which a mobile phone is a portable computer with attached sensors. This is represented by last week’s 6th ACM Conference on Embedded Networked Sensor Systems (SenSys 2008) — hosted by NC State that attracted many of the people we met.
  • embedded computing: We expected to find people in this area, but not in the universities we visited. (From a Google search, it would appear the Americans would attend the International Conference on Distributed Computing Systems while the Asians attend Embedded and Real-Time Computing Systems and Applications (RTCSA).
  • ubiquitous or pervasive computing: the idea that comptuers will be everywhere in our life and things not previously computerized (like a shirt) gets a computer. They have the ACM-sponsored UbiComp conference and the IEEE-sponsored PerCom conference.
  • mobile computing: which has the ACM-sponsored MobiSys and the MobiCom conferences, both sponsored by the ACM’s Sigmobile.
The latter two fields appear to also attract researchers from information systems (MIS) departments in a business school.

In addition to CS and IS, mobile devices also seem to be attracting attention from others that use computer-enabled media — media studies, gaming, virtual reality — things that were on PCs that are nwo moving to mobile phones.

Later this week, I hope to post my thoughts on two specific areas of mobile phone research: sensor networks and the large Georgia Tech operations. I’ve already posted my notes on the use of mobile phones in the classroom. Stay tuned.

No comments: