backtop


Print 100 comment(s) - last by William Gaatje.. on Oct 16 at 11:43 AM


An artistic rendering of the new light-driven wireless network in action.  (Source: Boston University)
Could the networks of the future run on light

Solid-state lighting is one of the hottest topics in the tech industry, and with good reason.  The Department of Energy is sponsoring a $20M USD "L Prize" for advances in LED lighting, a type of light which uses solid-state components (diodes).  The research is a big deal as lighting currently consumes 22 percent of the electricity in the U.S.  If the DOE accomplishes its goals of reducing lighting energy use by 50 percent, it would save billions of dollars and reduce environmental impact.

New research from Boston University's College of Engineering, funded by a National Science Foundation grant, indicates that LEDs may be not only the integral lighting component of the future, but may also form the backbone of future wireless networks.

BU Engineering Professor Thomas Little describes the new research, stating, "Imagine if your computer, iPhone, TV, radio and thermostat could all communicate with you when you walked in a room just by flipping the wall light switch and without the usual cluster of wires.  This could be done with an LED-based communications network that also provides light - all over existing power lines with low power consumption, high reliability and no electromagnetic interference. Ultimately, the system is expected to be applicable from existing illumination devices, like swapping light bulbs for LEDs."

The primary goal of the research is to develop LEDs that do exactly that -- transmit information wirelessly via controlled blinking. 

Little continues, "This is a unique opportunity to create a transcendent technology that not only enables energy efficient lighting, but also creates the next generation of secure wireless communications.  As we switch from incandescent and compact florescent lighting to LEDs in the coming years, we can simultaneously build a faster and more secure communications infrastructure at a modest cost along with new and unexpected applications."

Professor Little and his colleagues imagine LED lighting in the room being hooked up to computer circuitry, which uses existing lighting to implement a wireless network which provides data to computers, personal digital assistants, television and radio reception, telephone connections and thermostat temperature control.  Prototypes of the new network design, according to Professor Little, should start at around 1 to 10 Mbps.  Better yet, bandwidth would be greater than in existing radio frequency (RF)-driven networks.

In the new network, each LED light bulb would act as an access point.  Another perk of the new design is beefed up security.  Unlike RF networks, the new signal would not pass through walls or other opaque objects.  This would help prevent snooping and connection theft.  The new system would also use much less power than RF, as solid state lighting is energetically cheaper to the strong radio signals needed for wireless internet. 

The flickering which drove the network would be performed so fast the human eye could not see it.  The network would ideally be able to operate outdoors as well as indoors.  The first test deployment may be outdoors, with a likely candidate being car interiors.  Professor Little continues, "This technology has many implications for automobile safety.  Brake lights already use LEDs, so it's not a stretch to outfit an automobile with a sensor that detects the brake lights of the car in front of it and either alerts an inattentive driver or actively slows the car."

While the technology seems very promising, one quandary is how to make the communication bidirectional.  Professor Little and his team have not elaborate on this tricky point yet in the initial press.  In order to send data requests, you would need a means of receiving light from devices such as cell phones or laptops, however, you ideally would want to avoid having to have a bright blinking transmitter on your device walls covered in sensors. 



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

RE: Another problem
By mindless1 on 10/7/2008 4:32:45 PM , Rating: 2
It's unlikely that it will depend on wavelengths, even a random batch of LED vary in the wavelength enough that there has to be some consistency to maintain a visible light - necessary since they're talking about leveraging visible light room lighting for implementation - that maintains a pleasant hue.

Ultimately the solution will be high frequency flickering, and yet this requires a more complex power supply so it drives up costs. Right now one of the best features of LED lighting besides the reduction in power and maintenance costs is the versatility in power sources, basically pick among many different designs to fit the application, so long as the current is kept where it should be.


RE: Another problem
By William Gaatjes on 10/7/2008 4:52:20 PM , Rating: 3
Well, afcourse the data is hidden in the modulation of the light. But you can use different wavelengths for tx and rx purposes. These wavelengths are afcourse not 1 single number but just 2 narrow bands of the used spectrum. This would allow for deviation by production processes. You could use more small bands for multiple devices. 3 devices would equal to 6 narrows bands. 3 * (tx + rx) for example.


RE: Another problem
By mindless1 on 10/9/2008 12:50:01 AM , Rating: 2
No, you can't use different wavelengths for TX and RX, if the goal is using the same LEDs for room lighting, unless you mean including two different specific wavelength LED in the same room lighting fixture but even then, their wavelengths would be too similar to be effective if they were both part of the room lighting benefit.

You can't just divide up narrow bands when LEDs themselves deviate more than this in aveage production, it would require hand-picking and testing each LED which drives up costs a lot more when we're in an era that the adoption of LED lighting becoming widespread requires reduction rather than increase in cost.

Someday it will be more viable. It just won't be within the next few years for cost-conscious average consumers.


RE: Another problem
By William Gaatjes on 10/10/2008 5:18:41 AM , Rating: 2
When i look into several datasheets, led's are very consistent when it comes to produced wavelengths. I do not think that is the issue. And assembling multiple leds into a light has to be done anyway. Natural light consists of a very wide spectrum. Ever used a led light on different materials, you might not see the colour that is actually there because of the narrow spectrum of led's, even white led's have this problem although they are getting better with every generation. Besides white led's are already a combination of red green and blue led's. That's already 3 seperate wide bandwiths.

1 Led cannot produce the entire spectrum of natural light since the radiaton produced by the led is depended on the bandgap of the used materials. And we want natural light lighting. Trust me on that.

Every element has it's own colour. By mixing elements and other advanced tricks you can make most colours but not a wide spectrum. So in the end you can use different spectrum's for TX and RX. Because a led lamp that produces a similair spectrum as for example the good old edison lamp with it's glowing wire needs multiple led's , or a led that is build up from multiple emitting materials. Afcourse the handheld devices themselves don't need "natural light" TX led's , just there specified spectrum led's.


RE: Another problem
By William Gaatjes on 10/10/2008 6:58:44 AM , Rating: 2
I must mention that most white led's are blue led's with a phosphor coating. The phosphor coating is exited by the radiation of the blue led and the phosphor itself emit's a lower part of the spectrum. Here the improvement must be found in the coating since these led's are still a bit blue.


RE: Another problem
By mindless1 on 10/12/2008 2:00:59 AM , Rating: 2
They are fully able to coat to reach less blue hue, but it incurrs a loss of efficiency. Ultimately we're talking about consumers so some studies should be done to decide what the most blue light they are willing to accept is and build a according to that standard.


RE: Another problem
By mindless1 on 10/12/2008 1:59:22 AM , Rating: 2
No, most room lighting LEDs are binned for their wavelenght, any comprehensive datasheet should mention this binning.

Yes they have to be assembled into one light but that is why you have to know and pick a bin to achieve some particular wavelength.

White LEDs are not already a combination of red green a blue, they are blue LEDs with a coating that results in somewhat nearer "pure" white, the further away from blue it is the less efficient the LED is in producing light.

You can't use very different spectrums because you then escape the whole point which was that the *light* used was visible light within a desired room lighting specturm else the power was wasted and they might as well have used separate LEDs from the room light LEDs.

It is possible to have TX and RX offset enough in spectrum to accomplish this, if the RX has long enough on-cycle time to offset the TX LEDs, but adding to the complexity in this manner seems unproductive since the ultimate limit in range will be the client devices which are not using their TX/RX LEDs at anywhere near the power level the mains powered lighting is so for practical purposes the benefit is lost versus using an equivalent LED TX/RX pair as the hosts use. IOW, separate optical networking modules instead of integration into room lighting.


RE: Another problem
By William Gaatjes on 10/12/2008 7:49:43 AM , Rating: 2
Picking and binning is no issue. I have explained how led's produce the radiation. This is pretty constant.

And i already explained some led's are blue led's with a phosphor coating. These led's give unnatural lighting and is not pleasant to be in all day. Ask you fish for example what good lighting does for them. visible light is a big spectrum.


RE: Another problem
By mindless1 on 10/15/2008 4:59:20 AM , Rating: 2
Picking and binning is CRUCIAL, because we're talking about wavelengths that result in the desired ambient light hue by leveraging the light produced to also be a comm means.

You can't just ignore wavelength, the whole point was these wavelengths were power used to light the room desirably, not just picking among the whole spectrum.

Visible light is NOT a big spectrum when the goal is a desirable hue. The spectrums have to be very close or else you will have a pulsating effect noticable to the human eye.


RE: Another problem
By William Gaatjes on 10/16/2008 11:37:12 AM , Rating: 2


The modulation frequency is much too high for the human eye to notice. If we would be able to pick up light modulations in excess of 50 HZ for example, television would have never worked.


RE: Another problem
By William Gaatjes on 10/7/2008 5:04:55 PM , Rating: 2
And indeed the current through the led is important. The best led powersupplies are current regulating and not voltage regulating. The voltage over the led is fairly constant.


RE: Another problem
By mindless1 on 10/9/2008 12:56:03 AM , Rating: 2
Any LED power supply has to be current limiting or it's defective. You could hand-match one constant voltage PSU to one hand-picked LED with a tested voltage:current ratio but this could never be a cost effective manufacturing strategy.

There would necessarily have to be processing of what the data rate vs frequency result was in total light output so the voltage increased as the frequency changed to transmit data. It is attainable, no question about that, but the issue is doing this cost effectively in *every* light module installed when the present disincentive in deploying LED lighting is already the (for most uses) excessive cost per lighting module.

Actually, the voltage over the LED will not be constant within the idea of data TX/RX, it is necessarily going to rise and drop below an emitter threshold at the frequency needed to achieve the desired data rate, at a minimum the voltage will cross a forward voltage/die threshold in the KHz range and probably higher considering their claims of data throughtput rate.


RE: Another problem
By William Gaatjes on 10/10/2008 5:36:56 AM , Rating: 2
Led's are current driven devices. To get the best results you need a current controlled power supply. This means that the voltage varies when the load (the led) varies because of temperature fluctuations. Good old Ohm's law U = I * R.
No handpicking there, just a good led regulator ic.

The led's will be just switched on and off at a high rate afcourse.
With the dataspeeds mentioned in the article i would say it would be megahertz. We have optocouplers that can switch that fast. But i think the switching frequencies of the led's will be much lower and multiple leds with their own part of the spectrum are used to obtain a parallel nature of sending data.

When there is no data to be send , the led's emit there colour at a controlled current set point.


RE: Another problem
By mindless1 on 10/12/2008 2:05:09 AM , Rating: 2
Actually no, the voltage doesn't vary it is current controlled and the heatsink is a constant so the temp should not vary significantly. You could make it more elaborate in an outdoor installation, in some cases that might be reasonable but for practial economical design choices, it won't happen.

They will vary voltage to some extent, to achieve a median brightness value during data TX/RX, to achieve roughly the same average current. That much will be necessary as consumers won't want lights that aren't at constant intensity. The practical applications will depend on budgetizing the design, people won't pay a lot for built in optical networking in every light fixture when a wifi router can be had for $10 total that covers several rooms, and the client adapters equally inexpensive.


RE: Another problem
By William Gaatjes on 10/12/2008 7:51:53 AM , Rating: 2
Please do some reading up on the specification of diode and Light Emitting Diodes. Also, please look at Ohm's Law.


RE: Another problem
By mindless1 on 10/15/2008 4:55:10 AM , Rating: 2
Diodes are not a resistive load, Ohm's law is not applicable. Please do some reading on diodes.


RE: Another problem
By William Gaatjes on 10/16/2008 11:22:38 AM , Rating: 2
Diodes are not resistors, i know they are non linear components. It is for the sake of simplicity i used Ohm's law for example. I don't have to read, i know the characteristics from diodes from memory. I see it simple, to make a led that has a spectrum similair to a ordinary bulb or what is even better day light, You can pick and bin all you like but a led lamp build from multiple led's is easy to calibrate for the desired color temperature. Because of the way led's produce light.


RE: Another problem
By William Gaatjes on 10/12/2008 10:48:59 AM , Rating: 2
There is something called thermal resistance. The junction temperature of these led's are much higher then the temperature of the heatsink. For a common example, CPU heatsinks do not reflect the actual temperature of the hotspots present in the cpu die.

And when the material get's hotter, more current starts to flow. That's why led's are current driven devices. And the more power you draw, the more serious it get's.


RE: Another problem
By mindless1 on 10/15/2008 4:56:02 AM , Rating: 2
I never claimed otherwise, but it is not linear and your basic statement of it has no direct bearing on the conversation.


RE: Another problem
By William Gaatjes on 10/16/2008 11:43:05 AM , Rating: 2
quote:
Actually no, the voltage doesn't vary it is current controlled and the heatsink is a constant so the temp should not vary significantly. You could make it more elaborate in an outdoor installation, in some cases that might be reasonable but for practial economical design choices, it won't happen.


Read your own post please as the quoted section comes from your post.

Please read up on current sources.

Also when looking at the I/U graph of a standard diode, You wil notice that for small forward voltage changes large current changes happen.


"We don't know how to make a $500 computer that's not a piece of junk." -- Apple CEO Steve Jobs














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki