I know that even cheap LEDs have rise and fall times of easily under 5 ns with the right circuitry[0]. This means that a transmitter of 100 MHz can be easily built with COTS parts, so the speeds they mention make sense to me.
However, when I built a cheap optical communications system with LEDs and $0.10 diodes, the tough part was always getting the photodiode capacitance low enough to receive high frequencies while still sensitive enough to detect the signal accurately[1]. This is why lasers are very popular for communications: a concentrated beam can deliver a very high signal-to-noise ratio. With LEDs, it's tougher because the beam is dispersed and you don't want to blind people with high power output, so the signal is a lot harder to receive. It looks like they went with nice avalanche diodes instead of cheap ones, and I'm sure that have a well-designed integrated circuit helps a lot too. I'm very excited to see how this plays out.
A note -- if you're linking to arXiv, it's better to link to the abstract (http://arxiv.org/abs/1011.1954) rather than directly to the PDF. From the abstract, one can easily click through to the PDF; not so the reverse. And the abstract allows one to do things like see different versions of the paper, search for other things by the same authors, etc.
> This is why lasers are very popular for communications: a concentrated beam can deliver a very high signal-to-noise ratio. With LEDs, it's tougher because the beam is dispersed and you don't want to blind people with high power output
I suppose I am missing something obvious... but laser beams are shaped into a beam by way of a lens at the end of a tube, right? The laser diodes I've seen emit light in all directions. You could just as well stick an LED at the bottom of that tube.
Classic lasers directly emit coherent beam without the need for additional optics. Output aperture of most lasers might seem to emit light into all directions, but most of the output power is concentrated into the direction of the resonant cavity. Due to manufacturing tolerances low power laser diodes tend to produce somewhat irregular output beam (both diverging and uneven) which is then compensated for using some external optics that also serve to set required width of the final output beam.
It is impractical to convert light from some arbitrary "point-like" source (ie. LED) into beam that has low divergence by lenses or similar optics.
Lasers generate coherent radiation, the frequency of the light is much more monochromatic (set by the optical cavity) than a bare LED (set by the thermally smeared band gap of the material).
Even a very cheap laser diode can have the output collimated with a single lens, resulting in a beam that diverges a fraction of a percent. Hence cheap laser pointers.
Try and do the best you can with an LED and you will end up with a torch.
About the issue of capacitance and detecting the signal (with visible light interfering with it, unless they work in IR only) you can modulate the signal in some way then filter for that
So the obvious question is... what happens when line-of-sight is lost between the endpoints? With near-visual wavelengths, someone walking past could interrupt the connection. I could see this being useful for very limited applications in settings where radio is completely unusable, but it's a stretch.
> In 25 years, every lightbulb in your house will have the processing power of your cellphone today
And in 50 years, every hammer and wrench in your toolbox will be able to beat Ken Jennings on Jeopardy!
To be fair, I replaced every lightbulb in my house with a Philips Hue -- which does have some limited processing power (enough to manage a Zigbee mesh connection).
So the "every lightbulb in your house" comment doesn't seem completely unreasonable to me.
Dim able and colour controlled without having to walk over to the switch on the wall. Integrate the controllers win IR sensors, you have lights that turn off when nobody is in the room. Built in sunrise alarm in every room! Turn the lights red when playing submarine simulators for extra authenticity :)
I have Hue bulbs everywhere too, and I really like them. Being able to sunrise and sunset them automatically is really nice, and they're a great ambient information display: One of mine near the door is blue if there is rain forecasted for today!
That seems naive. Current connections are encrypted. Those with the desire and means to break crypto aren't going to be deterred by line-of-sight issues.
Multiple lights is multiple paths, and maybe a slower connection off reflections on wall?
Even if the above is not true, there is usually a very useful niche for every communication, especially a faster one. Wifi doesn't work all that well throughout the house. Wifi + LiFi might be a much better experience.
I have another question, could be seriously stupid: Do you need to have a strong, glowing light on the top of your computer for this to work? Could it use the monitor backlight? Imagine if the Apple logo in a MacBook is actually a Li-Fi "antenna".
Whatever happened to UWB? Ten years ago, there was so much talk about UWB, and how it would provide so much bandwidth, etc. And now it's nowhere to be found. One huge advantage of radio is that it can work through walls; "Li-Fi" won't.
802.15 WPAN - It's kinda stalled for some time.
And UWB really did get good press with some very promising features / performance back in the day.
If I remember they ended up using it in some system for firefighters that can 'see' through walls, and it might've been used for such a system in the army as well.
UWB is making steady progress in active tracking systems and passive tracking systems. I work for a company that does active tracking: http://pluslocation.com/.
PLUS is an offshoot of Time Domain, http://www.timedomain.com/. Time Domain does radar; I don't think they do any active tracking (could be wrong though). IIRC Time Domain is the company that did the see-through-walls system.
It's interesting how things go back and forth in science...
A Brazilian priest studied about using light to transmit data and achieved that around 1 century ago, even before Marconi
Unfortunately his government didn't supported him, because he "was a mad priest" that wanted to send messages to other worlds, he said:
"Give me a vibratory movement as large as the distance that separates us from these other worlds that roll our heads, or below our feet, and I will make the sound of my voice reach there."
Upon closer inspection however a small lens at the position of 12 o' clock on the watch face indicated the mode of the wireless data transmission through visible light.[4][16] Data was transmitted from the CRT of the computer through a series of pulsating horizontal bars,[17][18] that were then focused by the tiny lens and inputted into the watch EEPROM memory through an optoelectronic transducer operating in the visible light spectrum and employing optical scanning technology.[19][20]
The CRT synchronization was possible only for systems operating on Windows 95 and Windows 98. The watch was compatible with Schedule+ and for the Datalink 70 model the time needed to download seventy phone numbers was about twenty seconds.[14][17]
When my brother and I were younger we would sneak up to the door of each other's room and flip the light switch. It was mildly annoying. The kids of the future are going to go ballistic when a younger sibling flicks off their internet.
The article mentions industrial applications being the first to use this. What are some example industrial applications where you need a high-bandwidth, low-latency, link that has to be wireless, that WiFi is not good enough?
A lot of industrial processes generate immense interference - electroplating and arc welding spring to mind. I guess those would be a logical starting point.
How about medical instrumentation/telemetry? In the ER, ICU and other bedside locations ability to communicate high bandwidth data is very valuable. Wireless transmission allows greater flexibility choosing devices to measure various parameters and freedom to set them up optimally around the patient.
Another idea is having a completely sealed Li-Fi transmitter in an isolation chamber send experimental data through a transparent port for analysis. (My guess is Wi-Fi would be harder to package for the purpose.) I'm sure there are many other situations where high-speed short-distance transmission would be advantageous...
Have you heard of/seen Kiva robots? If the floor could just blink to the robot, that would be pretty cool! Especially if Li-Fi is cheaper / higher throughput.
Contrary to the article's title, LiFi doesn't _compete_ with the majority of WiFi use cases due to its limited propagation characteristics.
The article confirms as much: "“We don’t want to replace Wi-Fi,” he says. “That’s not our goal.” But Deicke says Li-Fi could complement existing communications technologies including WiFi"
Would this be less susceptible to local surveillance than wifi, e.g. within a few hundred feet? How would it deal with interference from overlapping beams?
Line-of-sight communications are definitely less susceptible to surveillance than WiFi that has omnidirectional broadcast. Even if the signal bleeds due to refraction, at such high frequencies, this loss of power is extremely detrimental to making sense of the waveforms.
Multipath interference can be managed with existing communications techniques, as light isn't the only EM wave that bounces off of things.
Hmm, true, but that could be remedied by transmitting the signal with something like Manchester encoding, so that the average intensity is constant whether or not data is being sent.
However, when I built a cheap optical communications system with LEDs and $0.10 diodes, the tough part was always getting the photodiode capacitance low enough to receive high frequencies while still sensitive enough to detect the signal accurately[1]. This is why lasers are very popular for communications: a concentrated beam can deliver a very high signal-to-noise ratio. With LEDs, it's tougher because the beam is dispersed and you don't want to blind people with high power output, so the signal is a lot harder to receive. It looks like they went with nice avalanche diodes instead of cheap ones, and I'm sure that have a well-designed integrated circuit helps a lot too. I'm very excited to see how this plays out.
[0]: http://arxiv.org/ftp/arxiv/papers/1011/1011.1954.pdf
[1]: http://www.thorlabs.us/tutorials.cfm?tabID=31760