1Gbps Optical Wireless Network Might Replace Wi-Fi

Mark.JUK writes "Pennsylvania State University has developed a new method of indoor Optical Wireless network that does not require a line-of-sight and runs at speeds of 1Gbps+. The system uses a high-powered laser diode — a device that converts electricity into light — as the optical transmitter and an avalanche photo diode — a device that converts light to electricity — as the receiver. The light bounces off the walls and is picked up by the receiver. Traditional radio frequency systems (Wi-Fi , WiMAX etc.) do not require line of sight transmission, but can pass through some substances and so present a security problem. Light, in a room without windows, will not escape the room, improving security."

HP used to sell a product like this

[Animats]

IR office networks were popular around 15 years ago. HP used to have a "NetBeame" IR access point product line. (There's one on eBay for $49. [ebay.com]) There's Linux support for IRNet. The Infrared Data Association is already promoting gigabit IrDA.

The concept of diffuse IR networking works fine, but it never really caught on. You can usually get a signal with one bounce, typically off the ceiling, but more than one bounce and it tends not to work. You don't get any useful diffraction around obstacles at IR frequencies, so shadows are a problem. If you populate the ceiling with little IR domes, it works fine, and I've seen that done, but it's obsolete technology now.

Re:Wrong on one count

[TheRaven64]

640KB is enough for some people. 6.4MB is enough for a few more, 64MB for a lot more, 640MB for a lot of people. Each increment increases the things you can do. 64KB is enough for editing text. 640KB for rich text and small images. 6.4MB for larger images. 64MB for large raw images. 640MB for SD video. 6.4GB for HD video. 64GB for volumetric 3D images (the visible human dataset is around 40GB). 640GB for volumetric 3D movies. Are there things that 640GB isn't enough for? Almost certainly, but the number of people wanting to do them is relatively small. Far more people want to edit text than want to edit volumetric data.

In terms of network speed, 802.11n is fast enough to stream HD video. BluRay movies are at most 50Mb/s, while 802.11n has a theoretical speed of 300Mb/s and can get 50Mb/s in the real world quite happily. Are there uses where it's not fast enough? Sure, but not many yet. Eventually it probably will become common to do things that make 802.11n seem slow, but it isn't yet. The only reason why GigE was deployed in a lot of places was that it was as cheap as 100Mbit Ethernet.

I moved from 100Mb/s wired Ethernet to 802.11g because, for most uses, the convenience of not having to use a wire was more useful than the extra speed. From there, I'd rather move to 802.11n and have one access point for the house than 1Gb/s optical networking and need one in every room.

Point to point

[symbolset]

It's a laser so the beam can be tight. That lends itself to geeking it to reflective targets mutually visible in outdoor applications. If people can get wifi at 237 miles [engadget.com], this tech may be capable of extending both the range and bandwidth of point to point communications. That would extend the reach of the Internet to a lot of people isolated by distance and infrastructure. That would be cool.

And then there's the neighborhood network thing. I can gather maybe 250 single family homes into a network with a fenceline network without crossing a right-of-way with a cable. Leveraging this tech I could probably extend that reach to 30,000 families. If you can build a 1 Gbps network that large in the US, the Internet will beat a path to you because you've got something they want: earners with eyeballs. Real bandwidth becomes free, which changes a whole bunch of things in a totally positive way

Re:Headaches...

[sillybilly]

Persistence of vision kicks in above 25 Hz, and some people can still sense an 85-100 Hz refresh on CRT monitors with their periferal vision, but anything vibrating at a higher frequency is oblivious to the eyes. I'm guessing stuff transmitting in the GBit/s range is operating near 1GHz or close to it, which is well above the 100 Hz flicker you can see. If anything, the laser hitting you in the eyes, especially if it's infrared and invisible, might be the problem.
One of the issues becomes the data interference from multipath reflection, that used to be a problem with nondirectional rabbit ear antennas in television. The speed of light is 3x10^8 m/s, the frequency we talk about is 1x10^9 /s, so that gives a wavelength for the modulated signal of 0.3 m, or about 1 foot. So path differences on this order or greater cause interference - light reflected off the far corner carrying the old databit comes in 10 nanoseconds later than the light reflecting from the nearby wall, carrying the current databit. This of course can be avoided if the path is restricted, such as aiming to a white spot on a wall, and then focusing a telescope with full magnification zoom onto that spot. This should be cheaper than running cables, however people walking through the room blocking line of sight have to be considered, and the ease of knocking the computer table/devices out of alignment. But a homing-zooming type wide viewfield webcam setup that finds you 10 different dots on the wall, and tells you to pick the one talking to you in software, then zooming in onto that dot, and keeping homed in onto it, that could be done. Or something transparent and automated, such as in a netcafe, you get assigned a new ID/key combo, and you type that in, and the computer goes from dot to dot, and shoots a reply dot 2 cm to the left of the dot, to try to establish handshake, and tries each dot iteratively til it finds the one belonging to it.
I think even now there is a way to listen to people having conversations by the modulation the window puts onto a reflected infrared laser light. There is already technology that extracts info from laser dots on a wall.
Still, software can do something about weeding out some of the multipath noise. Or even just focus on a small area of the wall, but broadcast the reply. In a spatially broadcast and not limited to a dot reply world everyone shares the same channel, and from the encoded info the master router station has to decipher which packet came from who. Less bandwidth, but also less hardware, as the router has to have a laser aiming for each of the channels onto a separate spot on the wall. Also, I haven't talked about repeaters, that go around doors and rooms and staircases, possibly being blocked by people walking there. With staircases pulling a wire through the floor might be more straightforward. Or a combo - dot on the wall within a room, and short cabling intra-floor or throught he walls.

Re:Headaches...

[Shatrat]

I'm going on what I know about fiber, but I assume the same holds true for air but with higher values of attenuation and maybe a shift up or down in frequency.
Basically at frequencies of 1310 nanometers and lower your attenuation is mainly due to scattering.
Your photon will scatter off any imperfection in the medium larger than it's wavelength so short wavelengths scatter a lot. This continues in fiber up to around 1600 NM where the main cause of attenuation becomes absorbtion of the photon by the atoms it is passing through.
Another problem with this tech that I didn't think of is dispersion. Dispersion is what makes multimode fiber only good up to around 500 meters. Multimode fiber is going to produce a lot less modal dispersion than something with as many modes as a room. I recommend reading up on single-mode and multimode fiber and modal and chromatic dispersion. It's pretty cool stuff from a geeky perspective.