Gigabit Ethernet Through the Air

There are a couple of really great things about transmitting data using light as the carrier. It can be focused so that it doesn’t spill all over the neighborhood like radio signals do — giving it both some security against eavesdropping and preventing one signal from stepping on another’s toes. And while you can modulate radio signals up nearly to the carrier frequency, the few gigahertz we normally use for radio just won’t cut it for really high bit rates. Light gets you terahertz.

The Koruza project is an open-source, “inexpensive” system that aims to transmit 1 Gb/sec over distances around 100 meters, using modulated infrared light. The intended use-case is urban building-to-building communication at speeds that would otherwise require laying fiber-optic cables. Indeed, the system piggy-backs on existing fiber-optic equipment to get the job done, but the hard part is aligning the units to get maximum signal from point A to point B.

koruza-spec-info

Koruza does this by including motorized lenses on the 3D-printed chassis. You make a rough alignment with a visible green laser, and then fine-tune the IR beams from a web console where you get immediate feedback on how the received signal strength is changing. Both Koruza boxes have a Raspberry Pi inside and use normal networking for calibration and signal-strength statistics. It’s a really neat system, and it’s fully DIY’able except for the commodity fiber-optic bits.

We’ve always had a soft-spot in our heart for transmitting data over light beams. The Ronja project has been doing so since 2001, and over longer distances, with completely DIY hardware, if at a slower bitrate. And now that Li-Fi seems to be getting traction, we might see an unfocused equivalent running inside our homes.

Thanks [Pavel] for the tip!

68 thoughts on “Gigabit Ethernet Through the Air

  1. ” urban building-to-building communication at speeds that would otherwise require laying fiber-optic cables.”
    Unfortunately this isn’t true. 1Gb/s is way less than what fiber optics is capable.
    Btw.: Commercial units have a heater on the window (like in a car rear window) to avoid condensation. It would not be acceptable, that telephone and internet fail just because of bad weather.

    1. You’ve got it reversed, the claim wasn’t if you could lay fiber you can get the speed with this the claim was normally you’d need to lay fiber to get this speed.

      Commercial unit or not I wouldn’t rely on this method for primary link. Every time a bird came through or the rain/snow got to heavy you’d drop signal.

      1. Indeed, we had this as a primary link to our office at some point as a temporary solution. (they where digging new cables). It dropped out or lost a lot of speed during rain. And the real fun came with fog.

        Birds are no real problem tough. Small interruptions of a signal also happen with WiFi.

  2. I was connected with Ronja 10mbit. Works perfectly with almost zero lag on 600m distance. There were one big issue. Doesn’t works at all in foggy winter. Really inexpensive, under $100.

    1. Where did you get a RONJA for under $100? that’s not the cost I remember when I researched this around 2009. Is Ronja so cheap now?. Did they simplify the electroniccs? I remember someone was working on putting all of thr RONJA circuitry on a FPGA, but never came across a final/open project.

      FC

    2. Back in 2011 I came across someone who did his graduation thesis about converting the RONJA to FPGA, he even made a prototype.
      https://www.vutbr.cz/en/studies/final-thesis?zp_id=42762
      I tried emailing him in 2011 but never heard back. Visited the RONJA web page today thinking I’d see improvements but basically the news section is all self-pandering about mention on RONJA on different sites and books, but little substance. Hardware-wise, the RONJA project seems not to have moved one nanometer since I last checked out the project in 2009, SEVEN years agoi. Saddened. :-(

  3. Very nice to my project featured here, it is also on https://hackaday.io/project/10083-koruza. Happy to answer all the questions you folks may have.

    @Martin, so far we have not seen any problems due to condensation, because the unit is not air-tight and equalizes to outdoor temperature already. There is a mosfet output on the control board, such that one can easily install a heating element next to the lens, which might be most useful if the unit gets snowed in.

    In terms of reliability of the system, I am working with my team to properly quantify this through deploying a number of this units globally and figuring out their reliability, currently most significant problem is units mis-aligning due to thermal expansion of the KORUZA unit, mount and building itself, see more on the following blog:
    http://irnas.eu/koruza/2016/03/01/link-attenuation-due-to-temperature-variation

    Rain and snow generally do not have a significant effect on the system operation at 100m distance, however fog has a noticeable effect. We are working on properly quantifying this in a controlled environment and outdoors, see http://irnas.eu/test-facility for the testing setup.

    The results from an upcoming paper summarized:
    Experimental results, with respect to the KORUZA system, approximate the fog density in terms of visibility as experienced by the human eye, required to limit the system performance in the operational range of 100 m, to be significantly less than 50 m. In particular, the 23 dB optical margin of KORUZA system, maintaining operational link up to 230 dB/km attenuation for 100 m link, which corresponds to less than 20 m visibility conditions.

    Fire away with questions or see more blog posts on the topic at: http://irnas.eu/blog

    1. The article says this could be used for communication across building tops. Have you considered the fact that modern steel buildings can grow a few inches during a hot day due to heat expansion or that the tops of some structures can sway up to a few inches in a strong wind?

        1. Why not put a light ring on each end? A ring of low-cost IR leds to act as a signal guide to help aim the other side’s laser (And of course, a corresponding, cheap, IR sensitive camera). As for isolation (from, say, another unit placed nearby) you could used timed pulses to identify the other end.

          Admittedly, this could increase the costs pretty quickly, but it would simplify setup and help mitigate migration from thermal expansion somewhat.

  4. Very nice to my project featured here, it is also on https://hackaday.io/project/10083-koruza. Happy to answer all the questions you folks may have.

    @Martin, so far we have not seen any problems due to condensation, because the unit is not air-tight and equalizes to outdoor temperature already. There is a mosfet output on the control board, such that one can easily install a heating element next to the lens, which might be most useful if the unit gets snowed in.

    In terms of reliability of the system, I am working with my team to properly quantify this through deploying a number of this units globally and figuring out their reliability, currently most significant problem is units mis-aligning due to thermal expansion of the KORUZA unit, mount and building itself, see more on the following blog:
    http://irnas.eu/koruza/2016/03/01/link-attenuation-due-to-temperature-variation

    Rain and snow generally do not have a significant effect on the system operation at 100m distance, however fog has a noticeable effect. We are working on properly quantifying this in a controlled environment and outdoors, see http://irnas.eu/test-facility for the testing setup.

    The results from an upcoming paper summarized:
    Experimental results, with respect to the KORUZA system, approximate the fog density in terms of visibility as experienced by the human eye, required to limit the system performance in the operational range of 100 m, to be significantly less than 50 m. In particular, the 23 dB optical margin of KORUZA system, maintaining operational link up to 230 dB/km attenuation for 100 m link, which corresponds to less than 20 m visibility conditions.

    Fire away and other questions or see more blog posts on the topic at: http://irnas.eu/blog

  5. How many degrees of freedom do the motorized controls give it? I see something that looks like pitch and zoom, but no left-right control. Wouldn’t you need that also?

      1. nah, only really need high speed in one direction, so dont need a high power transmitter, could use existing simple wifi/2.4ghz solutions to get any tracking info back from the VRhead set/controls

  6. To avoid problems with rain, fog, condensation, interlopers & pigeons, I would use inexpensive, easy to assemble, 100% waterproof PVC pipe.
    I have never tried it, but I would bet that the pipes will do a dandy job even if there are bends in the path.
    Slightly pressurize the pipe, stick a pressure sensor in there and you have a cheapo tamper alarm for your private communication line.

  7. I don’t understand why anyone would deploy this as opposed to some sort of point to point wireless such as the systems available from Ubiquiti for an order of magnitude lower price?

    1. Where possibly one should do exactly as you suggested, except in highly populated urban areas where a large number of WiFi devices congests the RF spectrum and thus throughput is limited. When using wireless optical systems that have a collimated beam, it does not congest the neighbouring links. Thus we can reliably achieve 1Gbps or 10Gbps across the street.

  8. How’s the performance on that simple lens? Is it good for this purpose or just acceptable? I wonder about spherical aberration any time I see a one-element lens used. I wonder if an alternate version that can adapt an old SLR lens would improve or hinder performance?

    1. spherical aberration shouldn’t be an issue since your only working with a single wavelength. Aberration is a problem when using multiple wavelengths, where each color travels slightly differently through the lens, based on the lens materials refractive index.

  9. 1. Kudos for the makers of this. I’ve long said that FSO is the future, and I knew something better than RONJA would eventually come up.
    2. The cost of the kit needs to come down significantly (I;d say 50-60% for it to be a viable solution down in South America -specially considering import duties of 50-60%).
    3. In my case, I don’t need Gigabit YET, I’d even settle for anything that gives me 100 MBit, and low fiber-like latency but over a slightly longer range (200-250 meters).
    4. Question for project devs: any chance of making this work over 200-250m even at reduced (1/10th) speeds? (=100Mbit) or not at all?
    5. IMHO what the hacker community needs to come up is a way to replace as many components as possible can with readily off-the-shelf components, something users can source locally (eg: the RasPis to begin with), and de-couple them from the kit, thus lowering cost
    These ared my initial observations.
    Kudos,
    Anxiously awaiting comments :)
    FC

    1. We are working on developing the next generation that will be much more cost effective, current KORUZA 1.0 version is the development testbed reliable enough to test out all possible solutions. Agreed on the cost part.

      Lowering the throughput by using 100Mbps SFP modules dues not significantly increase the distance because it gives a few more dB link margin, but reasonably bad fog is about 20dB attenuation for every 100m, so the distance can not be significantly improved. Also at about 200m not only attenuation due to fog comes in play, but also scintillation and other effects, requiring either a fast and expensive tracking system or larger beam=more power.

      Current KORUZA works fine over 200m as well, we have a few links like that, however 100m is the safe/reliable in all cases setup.

      Should anyone want a “barebone” kit with only crucial components please get in touch with us and we happily sort that out.

      1. Thanks for your reply!. Very encouraging.

        You said: “Should anyone want a “barebone” kit with only crucial components please get in touch with us and we happily sort that out.”

        1) What is the preferred contact method? do you have any e-mail address for those kind of requests?
        2) Can’t you just create a “Bare minimum / essentials only” kit and list it on the project/order page?

        Thanks, a lot
        FC

  10. “Why don’t you just use Wireless (RF) Links?”
    “Why don.t you just lay fiber”
    #FACEPALM
    Please, Americans, if you cant grasp the usefulness of this, dont make silly comments. You have absolutely no clue about the myriad of obstacles of linking two near points in crowded cities in LatAm, from regulatory obstacles, to spectrum congestion (2.4Ghz is useless, 5 Ghz wifi band is already crowded, 900 Mhz is unusable, and others EG Ubiquiti Airfiber @ 24Ghz is not a license-free band as it is in the US of A, so that leaves it out of the equation).

    Not to mention that wires must go underground per city regulations (at least down here in BA City), and incumbent telcos do NOT share ducts like in Europe, so each newcomer telco must lat its own ducts, at a roll-out cost (including muni permit delays) of about USD 100/meter (10,000 USD per block).

    So projects like this ARE a lifesaver for us, IF they can get the price further down AND double or triple its range (even if at the expense of 1/10th its current speed).

    FC

  11. Price is and issue but wouldnt it be possible to use a different spectrum laser with lower absorbation in both water vapor and liquid water (two different spectrum charts), to lessen the effect of fog and rain? Increasing power would help, and since the beam is quite big it would still be eyesafe, not to mention that the eye lens becomes opaque and protects anyway above 1400nm?

  12. My dream project would be to take this design (axis pointing and software) and replace as many components as possible with inexpensive, mass-market components already available from sites like DX.com, e.g. visible lasers, using cheap laser pointers, as done in this project.

    http://www.instructables.com/id/Laser-Transceiver/step2/Electronics/

    You could even use cheap laser pointers w different colors positioned at the side of the data link laser to visually help in aiming the two units at each other manually.

    Workable? I’d love comments from this project’s creators…
    FC

    1. Someone here also suggested using a LED ring.
      Why use more than 3 light sources to measure misalignment? Assuming a beam that is no too focused, and having the alignment emitters placed around the main laser as close as possible, this would give enough data for correction, wouldn’t it?

    2. Also, you wouldn’t need/want different frequencies for those calibration parts; they’d just flash their ID in binary (at a far lower rate than the actual data stream, presumably), and clever patter recognition software on the other side would pick up on, and identify those signals. You could even optimize the link without actually using it, that way.

    3. You can’t really use a frequency doubled YAG laser to send high-speed data, it takes too long for it to change state. Also, the shorter the wavelengths (green, blue, violet) loose a lot of power due to Rayleigh scattering, cranking up the power would make them not eye-safe.

  13. Nothing new here. There is a schematic for such an optical Ethernet adapter that has been floating on the web since the 90’s. It works quite well. However, it uses IR LED’s instead of lasers, but that is a moot point as they are essentially drop in replacements to one another, given a few considerations of course.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s