Passive WiFi On Microwatts

A lot of you use WiFi for your Internet of Things devices, but that pretty much rules out a battery-powered deployment because WiFi devices use a lot of juice. Until now. Researchers at the University of Washington have developed a passive WiFi implementation that uses only microwatts per device.

Working essentially like backscatter RFID tags do, each node has a WiFi antenna that can be switched to either reflect or absorb 2.4 GHz radiation. Your cell phone, or any other WiFi device, responds to this backscattered signal. All that’s missing is a nice steady signal to reflect.

passive_wifi-shot0008A single, plugged-in unit provides this carrier wave for multiple WiFi sensor nodes. And here’s the very clever part of the research: to keep the carrier from overwhelming the tiny modulated signal that’s coming from the devices, the plugged-in unit transmits off the desired frequency and the battery-powered units modulate that at just the right difference frequency so that the resulting (mixed) frequency is in the desired WiFi band.

If you’re a radio freak, you’ll recognize the WiFi node’s action being just like a frequency mixer. That’s what the researchers (slightly mysteriously) refer to as the splitting of the analog transmission stage from the digital. The plugged-in unit transmits the carrier, and the low-power nodes do the mixing. It’s like a traditional radio transmitter, but distributed. Very cool.

There’s a bunch more details to making this system work with consumer WiFi, as you’d imagine. The powered stations are responsible for insuring that there’s no collision, for instance. All of these details are very nicely explained in this paper (PDF). If you’re interested in doing something similar, you absolutely need to give it a read. This idea will surely work at lower frequencies, and we’re trying to think of a reason to use this distributed transmitter idea for our own purposes.

And in case you think that all of this RFID stuff is “not a hack”, we’ll remind you that (near-field) RFID tags have been made with just an ATtiny or with discrete logic chips. The remotely-powered backscatter idea expands the universe of applications.

Thanks [Ivan] for the tip!

30 thoughts on “Passive WiFi On Microwatts

  1. I find this intriguing and sort of a prerequisite to the whole IoT hype. Imagine a passive WiFi version of e.g. ESP8266!

    Would also be great if these continuous wave “beacons” could be set up in using the same infrastructure as GSM/3G but for low power applications to utilize. E.g. smart watches that want to focus on powering the screen, not the communication.

  2. This tech would be utterly useless in our home. Built a century ago, the stucco on the interior walls is held up by metallic lath, essentially a steel screen, which effectively makes every room a Faraday Cage – not good ones mind you – but good enough to make using low power wireless anything a series of small annoyances. Nor are we alone, most of buildings in this part of our town were built this way and suffer the same issues.

      1. For some nearfield applications I suppose, but then that is what Bluetooth is for. Internet of Things, as I understand it, has devices controlled over at least a home-wide area and as wifi is effectively blocked past more than one room over from the router, this will be an issue for this building if IoT becomes the norm.

  3. while this is good for portable devices, overall this actually uses more power if you have a 2.4 GHz transmitter constantly active. this could be improved by using an active low frequency/power request signal to activate the transmitter for X seconds.

  4. The way I see it, it seems like a lot of trouble just to have the passive device speak 802.11b naively. Don’t get me wrong, it is still a very interesting technique but I fail to see why the node needs to speak Wi-Fi directly. If you read the paper the illuminator device (i.e. the one supplying the RF power to the passive devices) is doing a bunch of stuff just to keep it Wi-Fi compatible but what is the real advantage of that? You could use the same technique (i.e. backscatter passive communication) but with a proprietary protocol instead that could be further optimized for low power (because it does not have to be .11b compatible) and simply use the illuminator device as a gateway between the two networks (it already does something like that anyway). In the current state of affairs the passive devices need to be polled by the iluminator and don’t get to send their info as they please anyway, so latency is not the real issue here.

    1. I wonder this every time I see a WiFi / IoT device that doesn’t have the bandwidth requirements, etc. that WiFi provides. Why didn’t they go with an optimized radio? Or BLE? Or?

      The answers I’ve got are: b/c it’s a standard and there’s a ton of WiFi devices out there, or “because IoT!” But I’m sure that I’m missing something. Anyone?

      Still, awesome hack.

      1. But the key difference is that this device requires a device aware of he local WiFi available nevertheless, so no new device needs to be added. In your scenario using a proprietary protocol would mean adding a new device for no other use than translating between the protocols.

        Certainly more applications could make use of the BLE though. If enough devices chose to use a single standard, there could be standard LAN-BLE bridges for accessing them from the Internet.

      2. Yes, this is pretty much the thing. Discussing with non tech people, most think radio=wifi and shown some home automation kits don’t understand why there needs to be a gateway.

        The new sub ghz wifi standard could solve this problem, but i would not expect it to be mainstream in less than 5 years.

      3. The reason is that each standard (or proprietary protocol) comes with a different set of tradeoffs. Oversimplifying it a bit, you can either have high bandwidth (Wi-Fi) or low power (BLE, ZigBee, others) but not both. In the article described you almost get the impression that you can actually have both however, in reality, the passive devices (i.e. low power ones) are not actually doing full Wi-Fi they are merely parroting Wi-Fi like frames so any regular Wi-Fi device can pick on them. The point being that much of the heavy lifting is done by a third device that additionally to providing RF power for the backscatter communications, it also tries to fool the Wi-Fi access point into believing that the simple devices are actually talking to it (performing associations, reacting to ACKs, etc). My point being with all that complexity why not just let this third device simply be a gateway between two distinct networks: the sensor node network optimized for low power; and the Wi-Fi network that is so popular and well supported?

        There are probably other advantages in doing it the way the authors described, I am just not seeing it. Anyone?

  5. Very similar, but off by 90 degrees or something.

    The seal bug was just taking in energy on the antenna, and re-radiating directly at the higher frequency that the cavity was (variably) tuned to. The cavity was resonating at the carrier frequency, which was changing with the microphone — giving FM.

    These researchers are chopping the reflected wave on/off at a difference frequency. No device is directly producing the desired WiFi frequency, if I understand it right. The WiFi signal is a mixture product of the “passive” device’s modulation and the powered device’s carrier.

    1. Theremins bug modulated a remote carrier to produce a side band that was twice the original carrier frequency. Because the (upper) side band was a harmonic of the carrier there was no lower side band.

      So in effect you had the carrier frequency (normally from the “Local Oscillator”) being transmitted at power then the bug formed the (AM) modulator / frequency mixer and the resulting upper side band was received by sensitive receivers. This scheme needed a large separation of frequencies and hence the use of a harmonic. Yhis is sometimes called “out of band” carrier.

      This wi-fi solution is different as modern filtering techniques (like DSP) allow much closer frequencies.

      The powered Carrier Generator is actually generating an intermediate (IF) frequency. The Wi-fi cards are not passive so they can make much more complex changes to the signal. Through Amplitude Modulation (absorption or reflection of the carrier) the simple air is used as a frequency mixer and the resulting upper side band is the correct frequency for the wi-fi receiver.

      In theory the remote low power unit can’t output more than it absorbs without battery power so if 12 dB is lost in the air path from the carrier generator then the return loss would be the same – ie a 24 dB loop loss. So this sort of thing is only useful for short distances or where the carrier generator and receiver are separate unit and perhaps equally spaced.

  6. Great work, but like other wifi powered devices we have seen before they are rather limited in the range and complicate things a lot. Both this system and the one shown before would probably fail majestically in a place which is crowded with wifi nets

  7. I would think that their approach would also shift other signals they receive by the same amount, and may result in an increase in RF interference in that local area. If the “passive” low power device received a strong nearby signal, such as from a TV broadcast transmitter or a nearby cell tower, this might cause noticeable interference. I hope they use a band-pass filter to limit the range of frequencies which they affect and to keep their emitted signal within the constraints of FCC part 15.

  8. I for one am getting tired of the justification of whether a particular post is a true hack or not. I’m not blaming the article writers but the people who seem to find it their personal mission to point them out.

    The name of the website is Hack A Day, can we not agree that if just one article per day is considered a hack by most that they are living up to their name?

  9. I am using ESP8266 for IOT modules using UDP and CoAP. This allows the wifi to fire up, send a signal and shut down in around a second thus conserving power. As battery powered IOT objects usually only need to send a small amount of data infrequently, that should be good enough to allow long battery life and maybe several years of use if some energy harvesting is added.

  10. The thing this lets you do is deploy a large number of sensors in a system without having to run ANY wires, not even power. That’s a big deal if you want to instrument a machine for example. It becomes practical to encapsulate the battery with the sensor making the sensor usable in harsh environments.

    This is also the WiFi version of the NSA bugged USB cables and Theremin’s great seal bug.

    Certainly the coolest RF technique I’ve encountered in a long time. I hope someone starts selling their chipset or something similar.

  11. Well, that’s… somewhat amazing!

    And the passively-powered stations manage to do the frequency-hopping too? Amazing what you can do with a few microwatts. These are chips more computationally powerful than entire computers from a few years ago, now running and communicating on the power you can suck in from an aerial. Insane!

    I wonder if you could get an LCD powered from something like that? An unpowered wristwatch would be good. Have a capacitor onboard, to either just run the RTC, or possibly keep the whole thing going constantly. Give it 5 minutes sat near any old radio transmitter, or maybe a Wifi station in particular, and you’ve got a permanent watch.

    Turns out wireless power is possible after all, as long as you use almost bugger-all power in the devices you run. Would be fun to have a whole room of almost zero-power stuff, all powered from one little discreet (not discrete) transmitter in a corner.

  12. I have a related question about antenna configurations for the WiFi power module. If I was to place this module in the middle of my home but at the highest internal point of the roof space, what shape antenna do I need so that all of the signal goes down and the strength of the signal is biased to favour more distant locations to give them as good a signal as something directly below the power module? It is not a trick question, I really don’t know, particularly the second requirement to bias the signal toward the horizon, but only to a limited extent, no point powering your neighbour’s devices for them is there?

    Also if I had a house full of these tiny micro-powered devices and somebody compromised my WiFi key, will I have to find them all and reconfigure each one?

    1. First there is no transfer of power, even if there where there wouldn’t a downside from your neighbors using the signal you radiate. The radiation pattern. Multiple floors would require more planing to be stating the obvious

  13. There may be a reason the terminology used was chosen, but it doesn’t add up IMO the device called passive devices are active devices. While the parasitic devices may make use backscatter waves they are just as likely to use the wave that emanate from the host energy source.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.