PervasiveComputing is likely to be the downfall of desktop computer systems (and their massive operating systems), at least in the broad application base they are used now. Seeing people no longer slaved to a machine to do any work, but computers slaved to how people work seems like an admirable goal (cf. TheSingularity).
Generally, discussion falls into one of the following categories. See the rest of the page for some pointers.
Contributors: SunirShah, DaveJacoby, CliffordAdams, ErikDeBill, AlexSchroeder.
See also: PervasiveComputingArchived.
Pervasive vs. Wearable Computing
See CarryingGadgets for more on how to carry your computing infrastructure such as pocket calculators, cell phones, MP3 players. That includes belt clips and jackets with builtin computing infrastructure.
See WearableComputing for more on the effects of carrying your computing infrastructure around. Contrast this with accessing a computing infrastructure from everywhere without having to carry anything around.
Pervasive Computing vs. Home Automation
See HomeAutomation for more on simple computing in your home such as switching light on or off and controlling your stereo.
Problem: User Interface
PervasiveComputing will have to offer a UserInterface that is easy to use. Maybe agents can alleviate the usability problem. Whether new user interfaces should actually model human behavior (such as using VoiceRecognition) is open to debate. See IntelligentAgent, StupidAgent, HandwritingRecognition, HumanComputerInterface?.
Problem: Spill Over
Sometimes problems with crude interfaces spill over into social life: People use a particular handwriting they trained themselves to use for their HandHeld while writing to a fellow human. See the Hand Held Example below. It gets worse if VoiceRecognition follows HandwritingRecognition in forcing us to talk a certain way to the computer.
Problem: Carrying Capacity
This is discussed in ClothingAndTechnology.
A possible solution is the continuing convergence of cell phones and hand helds. One device is better than two, usually. -- SunirShah
On a recent hiking trip, I had a GPS receiver and my Palm. I remarked to a friend that if I owned a cell phone, I would have brought that too (turned off). Indeed what I actually want is a tricorder. --tb
Problem: Power Supply
If the computing infrastructure is mobile, then a power supply must be provided. See ClothingAndTechnology, WirelessEthernet.
Usually not a problem if the device is commonplace.
Example: Having your cell phone used to be an embarrassment. These days, however, that is no longer true. Just avoid disturbing people with excessively loud or long ringing or talking, and avoid having it ring in movie theaters, meetings and other such places. See below for more on cell phones.
Problem: Avoid Orwellian Future
BruceSterling did an interesting speech on 911.net, archived in one of his ViridianDesignMovement notes, where he mentions ubiquitous computing, and lists a number of areas where we might actually want it:
Example: Pocket Calculators
PervasiveComputing became real in one sense during the 1970s and 1980s when pocket calculators (and watch calculators) permitted anyone to add, subtract, multiply and divide 8 digit numbers anywhere that they were. Certainly that is computation, for small values of computation. It doesn't bring in connectivity, data storage or much of what people think about when they consider computing.
Example: Cell Phones
To many people, location is just about irrelevant for making and receiving phone calls. They also remind of the limitations: places where they don't work (cities with a lot of hills, areas far away from civilization), lack of etiquette (see above for more on etiquette), short battery life, tiny screens, etc.
Example: Hand Held
Part of the reason Newton failed while Palm succeeded was that Newton used real handwriting recognition while Palm uses fake HandwritingRecognition, which is less messy. Any HumanComputerInterface? person would be aghast at the thought of people adapting themselves to computers and not the other way around, but the typewriter keyboard interface of most computers pretty much adapting to the machine, to the point of adapting against yourself (RepetitiveStress? and CarpalTunnel? problems).
See HandHeld, PalmTop.
Time is startlingly disconnected within my house, and probably yours, too. Check your alarm clock, watch, PDA, desktop and car stereo, and I'm sure you'll find at least a 5-minute difference between two of them. If you're unlucky, 10. The first thing I'd want to do with PervasiveComputing is to align them.
Whether this will change our perception of time, talking about time, and being on time, remains open to debate.
In Xerox PARC, you have a badge that identifies you, and if someone calls for you, the call is routed to the nearest physical phone. At work, in a much less technologically-advanced workspace, I have a badge that identifies me to my computer, and locks the screen when I am no longer near. Even fancier bathrooms know when you (or at least someone) is there and when you leave.
The traditional hack is MotionSensors?: put a motion sensor on a light switch and the lights will be on when people are there. The problem is that things that block vision block motion detection (for example: the chalkboard at the front of the training room) and that people in deep thought often don't display much motion.
See DoorGames, now that HandHelds are swamping the market, the future of gaming is in the past.
If you're looking for more information about pervasive computing elsewhere you might find that UbiquitousComputing? (or abbreviated as UbiComp?) is another common name for it. The phrase was coined by MarkWeiser? at Xerox Parc.
Tech: Interfaces, Protocols
Personally, I tend to think that most devices should be relatively dumb, but have a standard interface controllable by a smarter element.
The "tiny web server" idea is an attractive one, especially if IPv6 takes off. HTTP is a dirt-simple protocol that is easy to write tools for.
See the iPic matchhead size web server that fits in less than 12kbits (1532 octets).
One big problem with the current Internet Protocol (IPv4) is the lack of addresses. IPv6 can give every person on Earth more addresses than the entire current Internet. (Your toaster could even run its own internal network. ;-)
See IPVersionSix? for a discussion of why your toaster might need an IP address (or not). Many IPv4 problems can be solved using NetworkAddressTranslation? and DHCP, plus firewalling.
Tech: Blue Tooth
See BlueTooth?, for a solution to networking in a pervasive computing environment.
While it seems nice and low-power some people recommend WirelessEthernet instead because of roaming and range issues. See below for more about that.
Tech: Wireless Ethernet
See WirelessEthernet for an alternative to Blue Tooth. Interferences from other appliances and speed continue to be issues. We're talking 11MBs, which was fine for 10baseT but doesn't quite cut it in the age of gigahertz. (With 802.11a, you can get up to 56MBs, but with a smaller range.)
Tech: Near-3G Cell Phone Protocols
Until the widespread distribution of WirelessEthernet, WirelessWideAreaNetwork? protocols like GSM and CDMA will be the most common means for network-enabling a mobile person.
A study in to the human side of wearable computing. http://www.media.mit.edu/wearables/lizzy/mit-ideo/index.html (Note the gender difference in look, feel and utilization) (See CarryingGadgets)
Goel, A. (2001) Urban pilot: a dynamic mapping tool for personalizing the city through collective memory. in Proceedings of the Fifth International Conference on Information Visualisation (IV'01), July 25-27, London, England, p.227-232. Abstract available from http://csdl.computer.org/comp/proceedings/iv/2001/1195/00/11950227abs.htm.