New Part Day: Indoor Location Systems

GPS is an enabling technology that does far more than the designers ever dreamed. If you want a quadcopter to fly to a waypoint, GPS does that. If you want directions on your phone, GPS does that. No one in the 70s or 80s could have dreamed this would be possible.

GPS, however, doesn’t work too well indoors. This is a problem, because we really don’t know what is possible if we can track an object to within 10cm indoors. Now there’s a module that does just that. It’s the decaWave DWM1000.

This module uses an 802.15 radio to track objects to within just a few centimeters of precision. It does this by sending time stamps to and from a set of base stations, or ‘anchors’. The module is also a small, and relatively high bandwidth (110kbps) radio for sensors and Internet of Things things makes it a very interesting part.

Some of the potential for this module is obvious: inventory management, and finding the remote and/or car keys. Like a lot of new technology, the most interesting applications are the ones no one has thought of yet. There are undoubtedly a lot of applications of this tech; just about every ball used in sports is bigger than 10cm, and if ESPN ever wanted even more cool visuals, just put one inside.

If you’d like to try out this module, decaWave has an eval kit available through distributors for about $600. Somehow, there’s also a Kickstarter for a board that uses the same module, Arduino compatible, of course.

Thanks [Roy] for the tip.

86 thoughts on “New Part Day: Indoor Location Systems

  1. All these technologies can go to /dev/null. 90s called and want their business model back. Only an open and dirt-cheap standard will succeed and enjoy any significant deployment.

    1. I remember I commented about this somewhere… ah, notes on my Google Plus account from early April. Anyway, they quote up to 6.8Mbps for a peak data rate, and they’re available on Digikey for $32 for the module pictured. The ranging method is built into the updated 802.15.4-2011 UWB standard. This is just the first module you can buy for that.

      1. Yes, but all of these research team are well aware of the fact that WiFi with its max bandwidth of ca 40MHz sets a serious limit on the temporal resolution of things (effectively, 1/40MHz = 25ns) and hence, without a lot of additional effort (more antennas, phase-coherent and -calibrated receivers) on spatial resolution (basically, 25ns * 300,000km/s = 25e-9s * 0.3e9m/s = 7.5m). Only direction of arrival, combined with a lot of intelligence, can make that better, and it’s not even respecting anything like noise or the indoor multipath channel, in which you can never tell whether the most of your signal comes directly from the source (quasi Line Of Sight) or is bounced over a conductive surface, or whatever effects lead to it.

      2. WiFi based positioning is crap and suffers from major capacity issues due to the limited BW as opposed to GHz worth of BW UWB has. Add to that 10x longer range, better multipath performance and lower power and you have a clear winner over WiFi.. WiFi and other time division techniques will fail for IoT and UWB as used by Decawave will win out. It’s purely a matter of time

    2. DWM1000 on PCB with antenna, in stock at Digikey, for $32 in single quantity. Not “dirt” cheap, but quite reasonable for a new technology. Datasheets and software, including sample RTLS apps, appear to be downloadable freely with registration (though I haven’t registered, so I’m not 100% positive on that). So you should be able to quickly and easily roll your own RTLS system using the $32 version, without buying an expensive devkit; especially if you omit the supplemental sensors (accelerometers, gyros, magnetometers) and sensor fusion code.

      And it *is* based on an open standard, IEEE802.15.4-2011 UWB. The only parts of decaWave’s implementation that are subject to patent relate to their own silicon size and power optimizations, not functionality. So eventually there will be competition, and the price will go down. Way down once we start seeing Chinese chipsets.

      Unless you meant the fantasy-based definition of “open”, where schematics and firmware are available to build your own 3.5-6.5Ghz UWB transceiver entirely from scratch, thus ensuring there are no backdoors for the Illuminati. ;)

      1. The DWM1000 has “6 frequency bands supported with centre frequencies from 3.5 GHz to 6.5 GHz”.

        But is this ultra-wideband approach in this frequency range going to fit into the RF spectrum allocation regulatory system, eg. FCC part 15 for a consumer product for unlicensed use, as well as the regulatory system in different countries? Aside from 5.8 GHz, there are no ISM bands within that range.

        1. UWB is a whole new ballgame. The transmit energy of this device is spread out over 500Mhz/1Ghz, depending on settings. So the energy and interference received by any traditional narrowband receiver to which spectrum is allocated by law is much, much smaller than the total transmit power. It seems the FCC has embraced widespread low-power use of it on that basis, according to the Wikipedia article on ultra-wideband:

          “Ultra-wideband refers to radio technology with a bandwidth exceeding the lesser of 500 MHz or 20% of the arithmetic center frequency, according to the U.S. Federal Communications Commission (FCC). A February 14, 2002 FCC Report and Order[18] authorized the unlicensed use of UWB in the frequency range from 3.1 to 10.6 GHz. The FCC power spectral density emission limit for UWB transmitters is −41.3 dBm/MHz. This limit also applies to unintentional emitters in the UWB band (the “Part 15″ limit). However, the emission limit for UWB emitters may be significantly lower (as low as −75 dBm/MHz) in other segments of the spectrum.”

          An online calculator tells me that -41.3 dBm/MHz is equivalent to only 0.00007mW/Mhz! That is such an *incredibly* small amount of power, I assume it’s beneath even the background noise floor of any traditional narrowband receiver.

          The transmit power for the DWM1000 is programmable from -35 dBm/MHz to -62 dBm/MHz. So it can be operated within FCC regulations, and without license. The little bit of extra power available above the FCC limit is probably to compensate for lossy PCBs/antennas, they only care about actual radiated power.

          I can’t comment on regulations in other countries, but I don’t see any reason they should be significantly different, given the unique nature of UWB.

    3. And what, dear RiccoElectrico, entitles you to that opinion? The fact that in the 90s, engineers had to be paid for what they developed? Hasn’t changed. The fact that getting a product out there costs serious money, not only in design, but also in manufacturing, sales negotiations, license fees, marketing costs? Hasn’t changed. The fact that silicon doesn’t come for free, and neither does fab time? Hasn’t changed.

      The only thing that has changed is that, a) your average technological product today is built atop a reference circuit for a chip that is built in billions, so cost is low and b) the innovation depth of “nearly free” products is often not very high. And of course, c) people, who don’t rely on being paid by the company that employs them as engineers think they deserve cheap components.

      Seriously, this is emerging technology. As much as you wish, unless the company behind it is really really big and willing to take a risk (which, for good reasons, is most of the time frowned upon nowadays) you can’t expect dumping prices on new hardware. This isn’t your average “let’s produce another ARM cortex, and hit the market with 5$ eval boards”; this is application specific, home-grown, medium volume, high-investment risk hardware and someone will have to pay for this.

      so: 32$ is cheap, and everyone knowing the least about the technology involved will acknowledge that. Also, 32$ is cheap, and everyone that has the least bit of imagination can find a lot of uses for this, even if it costed 200$ a piece, so even the simples market considerations can’t justify a price lower than this.

    4. I agree with several of the bloggers points, especially the “lack of documentation”, I think I could write decent (enough) docs/man pages for some programs, but to get the coder to sit down and explain it is difficult, if for no other reason that they have moved on to their next project.

    1. If you don’t mind investing in some infrastructure, you can get away with passive UHF RFID for that kind of accuracy. Checkout the Impinj xArray. It keeps the end devices dirt cheap, but costs a bit to deploy sadly.

  2. So that GoPro drone that follows 10 feet above, 5 feet to the left and about 2 feet behind and films everything you do? Done.

    Selfie sticks just got a whole lot more interesting.

    1. Not exactly. You still need multiple stations to determine position and orientation. However, since the positioning is based off a timed sequence (or accurate time stamps synched across the array) things getting in the way will not make your drone veer sharply to the left because that side’s signal suddenly got a lot weaker.

        1. Actually, that doesn’t sound like a bad idea. Mind if I steal that from you for a drone beacon circling idea? A plane or quadcopter (whichever style is more appropriate for your needs) would fly in a pattern, keeping a constant distance or position and recording an area / doing something remotely / functioning as an amplifying antenna. It could carry and drop beacons, or they could be pre-set.

  3. Not just handy indoors – my (commercially bought) robot lawnmower could benefit from this. Currently it just does a random walk, but sticking these at the corners of the yard would let it mow nice straight lines. Other benefits: plan its trip more efficiently, find its way back to the charger more efficiently (both mean can mow longer and leave less ‘reserve’ for finding its way back to the charger), and learn to avoid that one place where it keeps getting stuck.

    1. Looks like they are still doing positioning based on ye olde, aggregate data on the server side and calculate your position approach, rather than on the client side. So for the client device that wants to know its position, you would still need a data connection to a server somewhere.

      Very novel though. I’d like to see someone hack away at them to see what they are capable of and if you can reverse them somehow (put “tags” all around the place and use an “anchor” as a receiver).

        1. I was only going based on the information on the site:

          “Real Time Location Systems are used to track and identify the location of objects in real time generally using simple, inexpensive nodes (badges/tags) attached to or embedded in objects to be located/tracked. These tags transmit wireless signals to devices called anchors or readers that use the wireless signals received from the tags to determine their location.”

          Based on that information, you would need to transmit the location or distances to the client for it to determine its position. I.e., the client isn’t purely a passive device like GPS.

  4. piffle and tosh.
    ” just about every ball used in sports is bigger than 10cm” where do you come up with this junk.
    Hockey balls, golf balls, tennis balls, table tennis balls, rounders balls, cricket balls, base balls.
    Balls balls balls and all less than 10cm.

  5. It seems you need to sync a 84Mhz between all anchors . But how do you synchronize the clock for the fixed anchors ?
    Can it be done wirelessly using the module itself. In the Posyx project it’s not really explained. You don’t know if you need a way (computer ?) to sync anchors :-/

    1. I have long ago done a project with previous hardware from this company. I was not directly involved in talking to the chip however. There are two modes:
      * Synced mode: here you need to distribute a clock signal, I don’t know the frequency. Basically the same principle as GPS
      * Round trip mode: here you measure a round trip delay to each base station, so the clock can be separate and unsynced.

  6. Talking about cheap… why not create “fake gps satelites” indoor? if gps does not breaks through walls, and this system i see does not also, as 300meters with clear line of sight can be less then wifi range… Why not emulate a gps satelite coordinate system with the correct values indoor? and you get it working out of the box, with the same technology..

    I bet harware cost will be way low, 50$ at max, so let the brain storm..

    1. It’s the regulatory cost associated with transmitting on GPS frequency that’s the problem (this is why GPS repeaters are expensive). Hence why most things only use the ISM bands. They are free to transmit on with certain EIRP values (dpending on the country).

        1. At the moment with the default “standard”, you get about 1-1.5m of accuracy (depending on how far apart they are place and their tx power; a side effect of RSSI having such low precision). I am currently developing two additional methods (you can optionally enabled them), to increase accuracy further. I have the hardware developed and the software is complete for the transmitters, I just need to update the Android receiver API to utilise this.

          I will have an updated video in a few weeks with it all working (busy with work and uni this next week).

  7. Why there’s people here whining about how “not cheap” this is? Your average Arduino-brand board costs 25 €, the average GSM Board costs around 35 €. And I’m talking about damn old stuff.
    Go support the new technology, people.

    1. “If you’d like to try out this module, decaWave has an eval kit available through distributors for about $600. ”
      Did you miss above line? article is about this..

      Do you borrow me $600?

      1. Evaluation kits are always way more expensive than the product itself.
        As it was stated earlier modules are available T $32 (probably less in quantity), if you just want to “play around” with for hobby purposes, you should be able to do that with a module or 4.. sure not china-bay cheap, but not excessively expensive either.

        That kind of evaluation kits are made for companies, who wish to use the product and need a platform where they can easily test the product and software for it. The time saved by not having to design PSU, processor, communication and software around the chip is worth way more than $600.

    1. the pozyx kickstarter actually gives you a full system of 4 anchors and a tag for 449€. So you already need 5 UWB modules for that. On top of that it has other sensors and a powerful microcontroller. I wouldn’t call it crazy expensive.. The evaluation kit of decawave is 600$ for only two modules, so you can’t even do positioning with that..

      1. As it’s not an open source project and a commercial one , it’s seems decent to me. But the same could be probably obtained for 200$ with a 25$ price for the module.

  8. Does anyone know of any implementations of indoor positioning that does not require line of sight? I’m guessing that many of these systems that use RSSI don’t work without line of sight because of obstacles that attenuate the signal and therefore the recieved signal strength is lower and the perceived distance is predicted to be further than it actually is? A possible solution might be to use the time of the arrival of the radio signal instead, although that would require precise timing instrumentation.

    I have a feeling that this type of technology will gain a lot of traction soon, when the main problems are addressed and when it becomes affordable. The applications might not seem apparent immediately, but there are some cases where indoor tracking is desired. Some examples are: For automated robotics in your house to be aware of which room it is in, so that it can avoid rooms it’s already visited or even stay away from “no-noise zones” that you can specify. In a warehouse forklifts (possibly automated ones) can be aware of each other’s position and can each choose a optimal path without congestion.

    1. No positioning system will work 100% reliably without line of sight, unless you accurately model the environment and control everything in it (people moving for example). Reflections and absorption will always be a killer, but you can use averaging (median filters), diversity and aggregation of different positioning methods to reduce the error as low as possible.

      1. I guess that means that the positioning system will need to be adapted for each use case? So if you were to use it on a robot vacuum cleaner like a roomba, would it be possible to reduce error by dead reckoning? Is it possible to also use a computational solution such as mapping out the entire indoor environment?

        1. Not really, as long as it’s “good enough” for most use cases. For things like the Roomba, you can use courser positioning for moving about and you would use additional sensors to create finer grain control (i.e. not running into things and docking).
          In terms of using dead reckoning, you could use that in combination with the positioning system make predictions if you like, but it would be more for path finding logic, rather than positioning itself.

  9. Hi,

    Thanks for your interest in Decawave Technology!
    Just a few extra info:
    – I confirm that the max data rate is 6.8Mbps.
    – 3 different topologies possible
    – point to point distance measurement to create virtual fences/leashes
    – 2D/3D Location/navigation by deploying an infrastructure
    – Mesh to build wireless sensor networks and/or get relative positioning of objects/people one to the other (safety, find my stuff…)
    – Price:
    – $32 is for the module and for a unit price.
    – The underlying chip, DW1000, is using standard CMOS technology like any other RF chip and thus fairly cheap in volume.
    – Size:
    – the chip is a 6×6 QFN so can fit in small designs

    1. I’m interested in the DW1000 to get positioning of a robot mower in a 40x40m garden. I guess I have to choose a TOF configuration where the module of the mower has to send a frame sequentially to each anchor to get distance of each anchor . An then make fusion of the data . Right guess ? :-)

          1. Our evaluation kit – TREK1000- gives a 2D accuracy of 15cm… RAW.
            With a bit of filtering/averaging you can get to the 10cm.
            And some of our customers even get down to 2cm using data from an inertial measurement unit.

          1. The limit of the precision is the bandwidth of the signal. The larger the bandwidth, the better the accuracy.
            Our chip has 500Mhz/1Ghz of bandwidth.
            In comparison, latest Wifi chips using Time of Flight (like our technology) have a bandwidth of 80Mhz. This is why their accuracy is only 2 to 3 meters.

            The drawback is that the larger the bandwidth the higher the power consumption.
            Our technology is an impulse radio so the power consumption remains really low.
            But for technologies like Wifi, a very large bandwidth would imply a very high power consumption.

            Note that current tWifi/BT/BLE embedded in phones and other devices rely on signal strength (RSSI), not Time of Flight. RSSI is very dependent on the environment and line of sight, this is why it has a poor accuracy in most of the cases.

            If you want to learn more we have several application notes on our website explaining the physics of the signal as well as the different methods to design a location system.

      1. +1, I’m looking at picking up a few bare modules for the exact same thing. I couldn’t get good enough resolution/locking with RTKlib GPS in a rover/base station setup, this looks like it could be slick for a relatively small yard.

  10. I see lots of talk here, but I wonder why people are so interested, you don’t really need this kind of thing for a residential home do you? So are you people all managers of large complexes? Because I don’t see this be useful for the normal home tinkerer. If you need to simply know in what room in your house you are there are simpler methods.

      1. Or, you could save a lot of trouble and money and get some vitamin D in the bargain by pushing a cheap mower around.

        •I• want an accurate positioning system for use in interactive art installations.

Leave a Reply

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

You are commenting using your 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