I've mentioned in past columns that I'm a private pilot and that I use a Palm VIIx for receiving weather radar data in the air. Bob Anderson, a developer working on an advanced version of the same service, recently told me a fascinating story that might have implications for ground-based users. In brief, he found that General Packet Radio Service (GPRS) doesn't work well when a device is moving.
Anderson was trying to use GPRS instead of the low-bandwidth Mobitex network that Palm VIIx and Palm i705 devices use. (With the additional bandwidth, he expected to deliver color imagery and support Pocket PCs.) He discovered that although GPRS gave him high-bandwidth connections on the ground (provided a GPRS tower is in range), it didn't deliver any connection in the air. Worse, he encountered the same problem while traveling by car. "I drove Interstate 5 from Washington state to Southern California," Anderson said, "but during the entire trip I was able to complete only one data transaction while the car was moving."
Anderson isn't alone in his findings: A quick Web search turned up a report that states, "Poor TCP behavior is reported for connections that are used in motion." For more information, see the following URL. (The document at this URL is a raw postscript file that's best viewed by using Google's text-translation capability.)
http://nestor.ccs.neu.edu/wwic2002/papers/105.ps
These findings aren't entirely surprising. Radio transmission from moving objects such as airplanes and cars is subject to "shadowing and multipath fading, particularly when one of the terminals is in motion," according to T.S. Rappaport's "Wireless Communications: Principles & Practice" (Prentice Hall, 1995).
However, obtaining a workable network connection from a moving vehicle is certainly possible. I regularly use the low-bandwidth Mobitex network on my Palm VIIx and have also used a Cellular Digital Packet Data (CDPD) modem to achieve much higher-bandwidth connections than Mobitex—both on the ground and in the air. The document at the following URL discusses a San Diego State University experiment in which an 802.11 connection (at Ethernet data rates) was established between an airplane and a ground station.
http://hpwren.ucsd.edu/news/020430.html
If technologies as disparate as Mobitex, CDPD, and 802.11 work, why doesn't GPRS? One possible answer is that GPRS is designed to share bandwidth with digital voice signals, which can reduce the signal strength available for high-bandwidth digital connections. GPRS might also be vulnerable to radio-frequency noise sources—for example, the magneto-based ignition system common in small airplanes or a radar detector in your car.
Also worth noting is that in the United States, GPRS is a new technology with limited coverage. A friend who recently spent a week in Europe told me that Web access for GPRS worked for him "nine times out of ten," when he was both on the move and stationary. (However, he also said that the data rate was highly variable.) GPRS in the United States might simply be undergoing "teething troubles."
If you've had experience with GPRS—positive or negative—write to me at jruley@ainet.com. Much rides on this technology: Microsoft and Palm are both betting on GPRS in their latest wireless products!