Li-Fi: Edinburgh prof seeds LEDs for communication

Oct 06, 2012 by Nancy Owano weblog
Credit: pureVLC Ltd

(Phys.org)—"We believe wireless communications should be as reliable as lighting." That is a company motto of startup PureVLC (Visible Light Communication), which aims to connect the two—communications and light. They have an approach that could be an alternative to radiofrequency spectrum, as there are worries about "running out of gas" because demand grows for space. "We talk about the Internet of Things where every device is interconnected, but without more bandwidth it will be impossible to provide reliable communications to all of these 'things,'" say company sources.

PureVLC has in mind an alternative communications path that could hit 1Gbps wireless speeds. Harald Haas, the company CTO, has said his company is relying on the potential of the use of for transmitting wireless data, Rather than complex radio schemes at increasingly higher frequencies, "the performance of VLC looks very favorable," said the company. PureVLC intends to commercialize LED systems. The team hopes the technology can be integrated into mobile devices.

Haas is professor of at the University of Edinburgh. His team have developed their wireless networking system to handle data transfer using instead of . They have been able to transmit and download streamed video using the visible light method, according to a Professional Engineering report published earlier this week. The report looks at this work so far.

The approach is called "Li-Fi," where LEDs are used to transmit data to photo-sensor receivers by making changes in the so fast the eye cannot see. The pattern of this flashing light can project the data in the form of 0s and 1s at very high speeds. The data is encoded in the light and transmitted from the . He commented that it is not really a case of switching the light on and off but of changing the intensity in a subtle and specific way.

Haas also told Professional Engineering that the company was developing a smart lighting development kit (SLDK) that can enable lighting systems that are already in place to transmit and receive data, reaching network speeds of up to 50 megabits per second. (The SLDK will be capable of downloading and uploading data at speeds of between 15 and 50Mb/s.)

This video is not supported by your browser at this time.

According to the company, the SLDK will enable lighting manufacturers to add "visible light communications" to their products. The kit is designed to enable rapid implementation of VLC technology using standard lighting fixtures.

Haas last year performed the first public demonstration of visible light communications live at TED Global, where he showed an angle poise lamp fitted with an LED bulb transmitting high definition video displayed onto a massive screen. When he interrupted the light with his hand the video froze and it was then restored when he removed his hand.

Explore further: Apple gets smartwatch patent

More information: purevlc.com/
profeng.com/features/dawn-of-the-age-of-li-fi

Related Stories

Data are traveling by light

Aug 01, 2011

Regular LEDs can be turned into optical WLAN with only a few additional components thanks to visible light communication (in short, VLC). The lights are then not just lighting up, they also transfer data. ...

OMEGA at the top of the data transfer game

Aug 26, 2011

Researchers in Europe have successfully developed a new transfer technology for video data. An outcome of the OMEGA project, the team transferred data at an extremely fast rate of 100 megabits per second (Mbit/s) ...

Recommended for you

Medical advances turn science fiction into science fact

Jul 18, 2014

Exoskeletons helping the paralysed to walk, tiny maggot-inspired devices gnawing at brain tumours, machines working tirelessly as hospital helpers: in many respects, the future of medicine is already here.

User comments : 21

Adjust slider to filter visible comments by rank

Display comments: newest first

gwrede
not rated yet Oct 06, 2012
So, let's watch movies only when it's dark. Or have our car know about traffic jams ahead only in the night.

This is about as useful as the wall-outlet network, envisioned 15 years ago, when some startup thought that not installing network cabling would save a lot of money. Then even Elcos thought of using high tension wires for data transmission. Well, as we all know, it fizzled so that few of us even remember it.

The room lamp will fare just as well, i.e. nada, zero, zilch.

rfw
1 / 5 (2) Oct 06, 2012
Flashing lights? The supporters of this idea need to do their homework to show that their flashing frequencies have no negative side effects such as seizures...
Lurker2358
not rated yet Oct 06, 2012
So, let's watch movies only when it's dark. Or have our car know about traffic jams ahead only in the night.

This is about as useful as the wall-outlet network, envisioned 15 years ago, when some startup thought that not installing network cabling would save a lot of money. Then even Elcos thought of using high tension wires for data transmission. Well, as we all know, it fizzled so that few of us even remember it.

The room lamp will fare just as well, i.e. nada, zero, zilch.


You miss the point. Eventually, you could develop light traps or other tools to redirect signals in the same way antenna works for radio or television.

There are problems with this technology, but it is not useless, because you could still eventually use it for specialized cases where more bandwidth is needed than existing systems.

there's also potential use for "shoutcast" style super computing, where different systems run algorithms in parallel and communicate almost like the old two-way radio...
ValeriaT
5 / 5 (1) Oct 06, 2012
to show that their flashing frequencies have no negative side effects such as seizures
We are already living in the world of stroboscopes already. The LED and fluorescent lights are flashing way more intensively with 50/60 Hz frequency already, than the 50 MHz Li-Fi technology proposed will ever do. The question rather is, where such a technology could be used effectively.
Lurker2358
not rated yet Oct 06, 2012
Now I mentioned two concepts above:

Light traps:

Very simply put, allows you to circumvent the "obstructive object" problem, by using larger light collectors to increase the area where reception of the signal can occur.

shout-cast supercomputing:

(my idea from many months ago when this technology was being announced).

This could allow cross-communication between processors or servers in ways that were not physically possible in the past.

My ideas were both the "broadcast" computing and a directed "laser network" computing whereby a server farm could have computers access one another by orientable laser, to travel directly to one another, instead of navigating around corners or cables.

These were also ideas I thought could be useful in Neural nets, because the individual "circuits" in such a neural net can reconfigure not only themselves, but their interconnectivity dynamically without the need for a "wire," hub or router at all, giving a "Universal Optical Bus" for all devices
Lurker2358
not rated yet Oct 06, 2012
The question rather is, where such a technology could be used effectively.


See above.

Optical neural net.

Shout-cast querying of many different servers, where no one server knows where the data is actually stored, nor needs to know, except the one that actually has it, etc.

Server A gets a query from a user for File X. Server A does not have file X, but somewhere in the room there is a server that does have file X. Instead of searching all servers individually, Server A sends a single optical signal which propogates through the room, and the first server to find File X transmits it back.

This is a one step search algorithm that instantly finds it's target every time. (Ok, not exactly because it's still a hybrid with the conventional algorithms on the existing machines, but you see the point).

You can see how this could be miniaturized, in theory, to make a self re-configuring optical neural net...if only we had the functional understanding of how to design such a thing.
Lurker2358
1 / 5 (1) Oct 06, 2012
But think about collective problem solving.

If your teacher at school asked the class a question, they are "shouting" the question. The first student to raise their hand gives the answer.

Similarly, if you are in an emergency board meeting, and the president of the company announces that he's out of ideas and asks "anybody got any suggestions," then the first person to have a real idea speaks up.

The president doesn't need to know the answer, he may not even need to know the correct question, because someone else in the room knows.

That's how collectives work, and that's probably, almost certainly, how neural nets really work.
Lurker2358
1 / 5 (2) Oct 06, 2012
Ok, and here's how a two-way radio conversation might go:

Speaker A: "This is Bald Eagle, on 7. Does anybody out there know the best place to eat in (obscure local town name)?

Speaker B (random person who hears): "Hey, this is fuzzy wuzzy, sure get off at exit 3 and take a left, there's a mom and pop diner, best place in the county!"

See how that works?

Speaker A doesn't know who will have his answer, but speaker B knows the answer anyway.

In existing computing, Speaker A needs to know who has the answer ahead of time, in most cases, or you need complicated software and hardware (routers) to figure that out. That isn't a requirement of networking in general, it's just a protocol due to our existing technology, which ironically is becoming so advanced that it is paradoxically self-limiting.
Lurker2358
1 / 5 (1) Oct 06, 2012
I have also talked about the idea of wireless optical buses between boards and cards in the same computer, as it saves space on the board, and you don't need to wait for the signal to travel the full length of the card, wire, or board, in some cases.

To my knowledge, unless it's something radically advanced, there's nothing like that being used yet in widespread applications in computing. Maybe the military uses something like that, I wouldn't know.
Neurons_At_Work
not rated yet Oct 06, 2012
Some obvious things getting overlooked here, I'm afraid. First current LED lighting does not flash at 60 Hz. Neither do CFL's, nor most standard fluorescents installed in the last 10 years. Next, the high speeds and low modulation levels as a percentage of full brightness are such that any change in the overall average brightness would not be sensed by a human in the room. Finally, an IR module could replace or augment a bulb of any type allowing use in the dark (albeit that would be a little silly).
Green_Dragon
not rated yet Oct 07, 2012

In existing computing, Speaker A needs to know who has the answer ahead of time, in most cases, or you need complicated software and hardware (routers) to figure that out. That isn't a requirement of networking in general, it's just a protocol due to our existing technology, which ironically is becoming so advanced that it is paradoxically self-limiting.


No,in wireless routers the signal is broadcast to everything within range and only the device intended to receive the packet replies. This isn't an unknown concept at all.
antialias_physorg
not rated yet Oct 07, 2012
So, let's watch movies only when it's dark. Or have our car know about traffic jams ahead only in the night.

Light can also mean infrared.

Basically any EM frequency that remains local to your home (i.e. is blocked by walls) is usable for this.
alfie_null
not rated yet Oct 07, 2012
Optical neural net.

It scales poorly. Consider the order of complexity.

Given one questioner and n information repositories you proposed, you could divide your body of data amongst the repositories to gain a corresponding increase in response speed, but at a cost of n times the fixed resources needed to support a repository (e.g. power, maintenance). And you have to funnel all your questions through one node.

Also note that it's not the way neural nets work. Nor the way organic brains work. Or ever have worked, even for infants.

You could argue that nodes could prune the links that aren't useful, but you are still left with the problem that all links share the same medium, meaning the bandwidth per link is affected by n^2 number of nodes. Or worse yet, if you don't have an out-of-band way to enforce collision avoidance.
Lurker2358
not rated yet Oct 07, 2012

No,in wireless routers the signal is broadcast to everything within range and only the device intended to receive the packet replies. This isn't an unknown concept at all.


Well, that's true, but the difference in wireless routers is they are using radio, which would not be secure for the type of computing I'm talking about. You don't want a hacker to have access to inside communications within a cloud network or an individual supercomputer, for example, it's bad enough they have access to outside communications.

The advantage of visible (or infrared) is that it is coherent long enough to be useful for a short-distance signal, but can't easily be intercepted by a guy on the other side of the block or across the street.

Additionally, wireless routers are not exactly what I was talking about anyway, because there you are dealing with outside clients, so you only want exactly one client to "understand" and respond to the signal.
Lurker2358
not rated yet Oct 07, 2012
alfie null:

Well of course organic neural nets work by storing information and skills in the physical and chemical configuration of countless dynamic interconnects via axons and dendrites.

An electronic neural net would have a problem because it's impossible to spontaneously reconfigure billions of wires dynamically on the fly with our existing technology. So to solve that problem you'd either need billions of microscopic routers allowing temporary on/off combinations as they are needed, or else each electronic neuron would need thousands of wires connecting to thousands of others, but only a few are used simultaneously. This would be incredibly wasteful in space and materials.

With an optical system, you could produce dynamic neural interconnects by simply re-orienting the light source and/or receivers.

I am not talking about bits in the sense of normal computing, but I am talking about multi-direction and parallel signal processing that mimics organic networks, but at light speed
Eikka
not rated yet Oct 07, 2012
Lurker2358: Brevity is the soul of wit.

Anyways, the problem with this idea is that using light is a line-of-sight technology only.

The whole point of using radio waves for communication is that they travel through objects, which means you can answer your cellphone indoors, and you don't have to have your wireless router in the same room with you. Reflections provide a partial remedy as long as the reciever can see the reflection, but 2nd and 3rd order reflections have too much noise to the signal to be useful.

In practical terms it means you have to hold your futuristic pad computer just the right way or the signal will diminish.
LED Guy
not rated yet Oct 07, 2012
@Eikka
This is not line of sight only You can easily handle a few reflections. The real key is developing fast (nanosecond response) detectors with a good signal to noise ratio that don't cost a fortune.

Simple math says that at 50 MB/sec you have a 20ft time of flight for photons. That limits how many reflections you can tolerate, but you miss the point that intensity of the reflections is orders of magnitude lower than a direct signal. I wouldn't even bother with a 3rd order reflection except as a homework problem for students.

You can also encode the signal using different wavelengths. They have been doing this in optical fiber systems for decades. For visible light you need more spacing, but it is not hard to design a 4 color LED system (red, amber, green, blue) for lighting. You can buy them today. With 4 channels, you have a lot more flexibility in how you encode a bit, so 50 Mbits/sec does not have to equal 50 MHz modulation.
LED Guy
not rated yet Oct 07, 2012
This was all proposed over a decade ago, it is just now that the technology is advancing to the point where you can do this at reasonable data transfer rates. LED prices have also dropped by orders of magnitude so it is all much closer to reality.

5 years would be a reasonable estimate for developing something that could be purchased by a typical consumer. It will take a least a year or two just to develop an industry standard for data transmission. After that it all becomes a matter of economics of scale.
Eikka
not rated yet Oct 09, 2012
you miss the point that intensity of the reflections is orders of magnitude lower than a direct signal


I thought that was implied. Furthermore, the signal is not just weaker, but it's also distorted due to the way different materials react to light. They don't merely reflect it, they absorb and re-emit it, which is what gives objects their color.

For visible light you need more spacing, but it is not hard to design a 4 color LED system (red, amber, green, blue) for lighting.


The problem is that the light quality of such a lamp would be poor. Good LED bulbs use blue and red LEDs, and a yellow-green phosphor to generate a wide continuous emission to fill the gap in between. Bad LED bulbs omit the red LED, and no LED bulb leaves out the phosphor because the result would be just horrible.

The difference of using only LEDs is a CRI of 50, compared to 85 for the typical CFL bulb, and 100 for the incandecent. Good LED bulbs can do 90 with the phosphor.
Deathclock
not rated yet Oct 09, 2012
Flashing lights? The supporters of this idea need to do their homework to show that their flashing frequencies have no negative side effects such as seizures...


What the fuck are you talking about? You obviously didn't read even the first paragraph of the article.
Deathclock
not rated yet Oct 09, 2012
In existing computing, Speaker A needs to know who has the answer ahead of time, in most cases, or you need complicated software and hardware (routers) to figure that out.


No... this is not true.

You always go on and on yet you almost never know what you are talking about. Information is broadcast to all receivers, only the receiver that it is intended for does anything with it, the rest ignore it.

Do you actually think your wireless router somehow targets individual devices to send data only to that device? No, the signal propagates outward as an expanding wave from the antenna, ALL receivers in range receive the signal, the data contains an address and only the receiver who's address matches the one encoded in the data responds.