Sounded good to me

[Jenni]

I was talking to my brother about my computer case's cable mess. I want to get it all nice and neat. He jokingly suggested wireless drives. That sounds like a GREAT idea, but how would you pull it off, or is it impossible?

[Toaster]

Jenni,
Its not at all impossible as its actually already been done.
Its more of practicality. The few remaining devices are the monitor and modem connections and the fact that infa-red has many limitations.

[Statica]

Imagine a cluster of computers, sitting side by side? How would one go about isolating the transmissions meant for one system. & with the transient noise from the power supply and other electrical components, going to be one icky situation?

[Toaster]

Ideally, using infa-red, each device works in a differing "band" of the infa-red spectrum. Much like a LAN with defined IPs. However, this is also where the problems come in. Infa-red communication can be fooled by radiant heat from devices to some extent sothese devices need to be "trained" to ignore anything outside the band and to regard it as "noise".
The best approach so far is "fire wire".

[Statica]

Also IR is very sensitive to transient light. An IR-device's performance can be affected by something as the lighting in your room.

Firewire is good too.

My thought is eliminate the bundle of wires, just by eliminating wires totally isnt going to solve it. But if you can cut down from the 40-wire IDE cable to one cable it would make things a whole lot clearer.

Given those choices USB is impractical because the bandwidth is too small, even comparing to IDE vs Firewire-
USB: 12 Mbits/s = ~1.5MB/s
IDE: ~max at 16MB/s
Ultra-ATA: ~max at 33 MB/s
Firewire: ~400Mbits = ~50MB/s

Bluetooth has great potential, but would it ever be fast enough?


How about something like this, just shooting the breeze, hopefully you wont fall off ur chairs laughing.
Something along the lines of a USB controller/hub, have periferal connector hubs, cheap to manufacture.. strewn along different places of the mobo. Say a couple near the 5.25" bays, a few near 3.5" bays ..
All you do is be able to plug the "wire" from the device to the nearest hub. Instead of the conventional system of stringing them on to the appropriate port... simply connect & daisychain to any port.

The last issue I have is with security, unless there is a physical connect, for something as integral as data between the different devices, it is scary when portals for exploits are created.

[Edited by Statica on 04-08-2001 at 12:23 PM]

[Toaster]

When dealing with "wireless" anything, security becomes a major factor. This includes cell phones which we all know to be grossly insecure. I found, in my dealings, that infa-red interfaces leave much to be desired, this is why the interface fails to grow by any extent. Infa-red has been around for quite a long time but because of its issues it fails to hit the mark. USB on the other hand I feel to be just as unstable a medium as DVD which will fail before long when a decent alternative is found. For the interim, they provide a solution but it will take little to unseat them as a "favorite". The bandwidth of USB is not static at all but rather dynamic IMHO and hence the reason for its eventual demise. For now, any other real alternatives are hypothetical but I tend to think that a "voice controlled" subsystem will be the key. Simply, it might work as so:
1. Get E-Mail
2. Go to "X"
3. Launch/run "X"

Voice recognition works but few companies are actively researching it. That and also standardization. Once these hurdles are overcome, then we may see some real progress.

[Statica]

Originally posted by Toaster
For now, any other real alternatives are hypothetical but I tend to think that a "voice controlled" subsystem will be the key. Simply, it might work as so:
1. Get E-Mail
2. Go to "X"
3. Launch/run "X"

Voice recognition works but few companies are actively researching it. That and also standardization. Once these hurdles are overcome, then we may see some real progress.

This is off topic but..

Try integrating voice recognition on Linux with ViaVoice modules. I even had a test installation that ViaVoice recognition engine for logging in. Was really cool at the time to have my computer log me in with voice command in full view of all my colleagues. Not practical cos the engine itself was far from perfect. Subsequent engines have been phenomenal, compared to the miniscule amount of processing power Linux actually uses. Currently there are many such programs like <A HREF="http://www.compapp.dcu.ie/~tdoris/Xvoice/" TARGET="_blank">Xvoice</A>, CVoiceControl

In addition, stuff that we havent

IBM, has been kind enough to lend its ViaVoice engines from its troubled OS/2 release for OpenSource development.

http://oss.software.ibm.com/developerworks/opensource/tclsmapi/

IMO the OSS is the onethat is doing so much away from the limelight of the expensive and shrinkwrapped boxes you see on shelves.
There are simply too many URLs to show here, but
http://freshmeat.net/search/?q=speech
http://freshmeat.net/search/?q=voice
http://freshmeat.net/search/?q=voice+recognition
etc are good places to start :)

[Toaster]

Former Mr. Cling,
Thanks for that web site, its an interesting concept and looks to be both usable and workable.