Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
UW engineers achieve Wi-Fi at 10,000 times lower power (washington.edu)
397 points by jonbaer on Dec 9, 2016 | hide | past | favorite | 67 comments


The low-power devices don't actually generate any signal at all thus don't have to pay that expensive power cost. Instead a passive antenna is modulated to reflect or absorb ambient signal, the reflected signal conforms to traditional 802.11 packets. The ambient signals are created by a relatively high-powered device that is plugged into wall power.

Here is a 2 minute video showing an early prototype of their work from 2014:

https://www.youtube.com/watch?v=snZWgAZqo2c


I've looked at the paper [1], but I can't figure out how this is different from backscattered RFID with 802.11 on top. The paper says RFID requires a full-duplex radio while the WiFi chipset is orders of magnitude less expensive. Are they modulating differently or is there some RF magic?

[1] http://passivewifi.cs.washington.edu/files/passive_wifi.pdf


The technology is ~50 years old, and was originally invented by the Soviets to eavesdrop on the US embassy in Moscow.

https://youtu.be/mAai6dRAtFo?t=9m35s

https://en.wikipedia.org/wiki/The_Thing_(listening_device)

TL;DR Using a passive antenna to modulate a directed microwave emission.


When will they reevaluate those twelve wing aeroplane designs


Wow that's interesting, how did they find "The Thing"?



Passive RFID is powered by the incoming radio waves that is 1000x stronger than the signal transmission energy from the RFID tag.

There is no statement to the power requirement for passive WIFI from an active WIFI power source. The active WIFI transmitter may require outputting radio waves many times more powerful in order to receive a signal transmission from the passive wifi device.

edit: For that reason, I'm not that excited about this, and this is likely why it wasn't pursued earlier due to the power requirements of active <-> passive signalling. If it pollutes the EM spectrum with powerful radio waves that are magnitudes more powerful than what we're outputting today, then I hope it gets abandoned. If it's the same power consideration, then it's worth pursuing imo.


Come to think of it, rather than following the usual inverse square law, power should drop off with the fourth power of distance (just like with radar) - there'd be one inverse-square dropoff on the path from active transmitter to transponder, multiplied by a second inverse square loss on the return path. That's got to hurt performance.


The maximum transmitting power is regulated. In the article they state that it can use some kind of crafted wifi packet to be compatible with any existing devices.

Edit: After looking at the pater, it seems that I may have misunderstood the article, it needs a plugged RF carrier tone generator at at fixed frequency.


Imagine coming up with the frequency scheme in an apartment situation; if the building were running the network it would be one thing, but I would expect individual units to foster their own connection.


Is it the sort of tech used by some devices in the NSA ANT catalog? It is useful for spying because they are undetectable if they are not subjected to a carrier frequency. There was also The Thing[1]

[1]https://en.wikipedia.org/wiki/The_Thing_(listening_device)


Did they just say enabling internet for billions of battery free devices? Is it like a push method? So I guess I could imagine state updates, or task pushes to autonomous agents.


Their video from 2013 shows some pretty cool applications of an earlier version of this tech that looks essentially like passive NFC using ambient RF rather than using 802.11.

Very cool stuff!


As far as I can tell, it seems to work the same way that passive RFID tags work. Pretty neat, I guess.


Almost. Technically it's the same as active (not passive) RFID tags: both reflect RF signals, and both need a source of battery power is needed to open and close the antenna circuit.

Passive RFIDs harvest power from the RF signals.


This article is a few months old. The technology is now being commercialized by the same authors, as Jeeva Wireless: http://www.jeevawireless.com/


In case someone is wondering 'Jeeva' (pronounced as Jeev) means living thing in Hindi.


Живой ( jee-voi ) is also a Russian word that means alive. Or in case of Жива ( jee-va ) it is a feminine gender of the present tense.

Gyvas ( Gee-vas ) is the same cognate in Lithuanian. And same as in Russian, Gyva is the feminine gender of the same word.

The root of the word comes down to us all the way from PIE

[1] https://en.wikipedia.org/wiki/Russian_language [2] https://en.wikipedia.org/wiki/Lithuanian_language [3] https://en.wikipedia.org/wiki/Proto-Indo-European_language


Jeeva is a telugu[1] word, which is actually pronounced 'jee-vaa' athough Jeev does mean a living thing in Hindi. It's easily to get confused with less popular Indian languages as being Hindi words because it's more popular, although they're similar because they're all derived from Sanskrit.

[1]:https://en.wikipedia.org/wiki/Telugu_language


To add, Jeeva means living thing in Malayalam, another language from the south India. In fact, Jeeva is a common name in Kerala and Tamil Nadu, two states from the south India.

[1] https://en.wikipedia.org/wiki/Malayalam


Very interesting, thanks. I know a South African family of Indian descent with that last name. They're the only Muslim family I've met with that name so I've wondered about its roots.


Indeed I was. TMYK


Their best goto-market is probably licensing and embedding in various existing consumer devices: battery-less TV remotes, external temperature sensors, cyclometers, etc. Otherwise, distribution and advertising becomes an uphill climb to typical startup doom.


I wonder if you could use this technique to make a completely passive wi-fi device. Ie, no battery or plug of any kind, just using the power of the wifi emitter. I guess it would be similar to passive NFC, but at a longer range and driven by a separate device from the one that does the receiving. Or would the power requirements be too high?

Edit: I just realised that the earlier prototype video from the comments indicates that's the goal, but it's not mentioned in this more recent one. Instead they just talk about higher power efficiency. I wonder if that means they tried it and it wasn't feasible.

Edit 2: Aha, got it. There are three separate projects:

1. Ambient Backscatter (2013): no external power, no base station (uses ambient frequencies like TV), not wifi, low data rate (1Kbps) and short range (<1 metre) [http://abc.cs.washington.edu/]

2. Wi-Fi Backscatter (2014): no external power, base station, kinda wifi (encodes data in the CSI/RSSI metadata and presence/absence of packets), low data rate (1Kbps) and slightly less short range(<3 metres) [http://iotwifi.cs.washington.edu/]

3. Passive Wi-Fi (2016): low external power, base station, actual wifi (802.11b packets), high data rate (11Mbps) and long range (<30 metres) [http://passivewifi.cs.washington.edu/]


I doubt that a microprocessor will be able to operate solely off ambient RF, but I could see a low power microprocessor being combined with this and a betavoltiac for impressively long life. Using this with the new diamond battery that they have in the labs at the university of Bristol would enable the creation of wireless devices that last over 5000 years.


PoWIFI


I invented a coffee machine which works using 10000 times less power: it works by deferring the actual coffee making to another machine you plug into another wall socket.


Your analogy is flawed: let me fix it for you. "I invented a battery powered coffee machine which works using 10000 times less power: it works by deferring the actual coffee making to another machine you plug into a wall socket. Now you can stop worrying about recharging the battery on the first coffee machine, and still enjoy your coffee just like before!"

With WiFi, the advantage is that battery powered devices like your smartphone will consume less energy. Sure, your router will draw more power from the wall socket, but you are effectively reducing power consumption where you actually are using WiFi the most, mobile devices that run on batteries.


Ty for the analogy, wasn't sure if I really figured what they were saying in the paper


It's funny when you put it like that, but that's how coffee was made for a long, long time. No electricity is actually needed, just a source of heat.

https://en.wikipedia.org/wiki/Coffee_percolator

https://en.wikipedia.org/wiki/French_press

https://en.wikipedia.org/wiki/Vacuum_coffee_maker


Yeah but can it fill your cup from the next room?


Depends on the velocity.


Maybe a coffeemaker that uses way less power, but it gets the hot water from another source.

The coffee maker still makes coffee, but it doesn't do the expensive heating of water.


I connected all the office coffee machines using a hot water tube, use just one to heat the water and distribute this to the others. Now I reduced energy consumption of almost all coffee machines 10,000 times, also I removed their water heaters to make them much lighter and cheaper. And still everyone here can enjoy (almost) as fresh and convenient coffee in every room, as the machines are still close to us.


Hey I invented a one bit compressor, which consist of the file.


10,000 times lower power, but achieving speeds up to 11Mbps

Being named passive wifi. Quite neat and could be useful for IoT considering also approx 100 ft distance


100' is a very low distance for real IoT applications I've run across


A directional antenna at the active wifi radio would enable longer range. One interesting application would be the ability to dogleg a wireless connection between points A and B where there is no LoS to each other, but both have LoS with point C. It would be a low throughput link, but it would allow for an awesome implementation of a private two way communications system. I.e. Several miles without direct LoS between the two end points.


This is already a thing, however data rates tend to be slower


Yeah but it's easier to pair down the antenna and underpower the radio than to go the other direction :)


but 30m is way longer than I need for my IoT. It also makes my devices more secure from various attacks on my network, as less people can see it.


It'd be enough to achieve significant power reduction for the <100ft region though; pretty sure wifi modules would be able to support both passive and active modes depending on availability.


You won't be downloading OS images very well, but it's still a lot.

That's two HD videos simultaneously, with some to spare.


Only if you have the whole channel available for that one device. Wi-Fi is a shared medium - devices at slow bitrates use the same amount of airtime that could be used by a faster device to send much more data. So a) if you're in a dense wi-fi environment, every bit sent with this system takes away much more than 1 bit worth of time from faster devices, and b) assuming the array of sensors mentioned in the article is around a dozen or two devices, we're down into the sub-Mbps range per device.

Still better than bluetooth, and very useful for the kinds of data that low-power sensors would send, but not the kind of performance you'd want for pushing media files or really high-detail sensor data.

(Hence the performance benefits for large greenfield networks of dropping support for low bitrates - both preventing old clients from hogging the channel, and obviating the need to send control traffic at the slower bitrates.)


I take it you've never used 802.11b. You were lucky to get a single YouTube video at 480p to load without downgrading, let alone two HD streams.


I'm (un)lucky enough that my WiFi has never been slower than my internet connection


5Mbps is an interesting definition of HD...


Also, not to mention, current best AC3200 wifi is rated at 1Gbps+, however, actual speeds are max 320mbps, or roughly 30% of the rated speed (for a single client right next to the antenna).

So, based on that type of math, I expect to see only 4mbps actual throughput to a single client; less if shared by multiple clients. These radios are likely not MIMO, so the antenna will have to "context switch" between multiple clients, so expect bandwidth to be significantly less in multiple device scenarios.

All that being said, I think this is fascinating, and quite perfect for the hundreds of various environmental sensors that will inevitably end up in all of our houses to help us live more automated lives. All hail our new IoT sensor streaming, data driven, fully automated smart homes...


Naturally it depends on compression, bandwidth variability, etc. It's what Netflix recommends, and they have zero incentive to lowball that.

https://help.netflix.com/en/node/306

The biggest issue is that people think they're getting 20mbps but they're lucky if they get 8. And that not a technology problem.


Yes it is. RF does not go in nice straight lines. The noise floor has been steadily rising, making getting a good signal harder and harder and leading to more packet loss. Add to that the fact that most TCP implementations don't deal well with lossy channels like WIFI, and you get a problem.


This is great research!

10,000 times lower power is only for the "passive wifi" devices, but these passive wifi devices require a "heavy-lifting and power-consuming plugged-in device" nearby.

Assuming that device consumes no more than a traditional wifi device, then the overall system would consume:

    [usual consumption of 1 wifi device] + [# of passive devices] * [usual consumption] / 10000
So, if we want to calculate per passive device it would consume around

    [usual consumption of 1 wifi device] / [# of passive devices]

In other words, the more of these devices, the better. Too few devices might not offset the plugged-in device energy cost.


It's not really the overall power savings that is the key advantage of the system though. Rather it's being able to have a multitude of low power devices that you don't have to run wall power to. They can either run for quite a long time off of battery, or even harvest power from RF signals.


I am just wondering if the plugged in part needs to transmit at full power all the time. This could seriously increase the overall energy consumption.

Plus, I have a hard time deciding what it would transmit at full power all the time...


I'm not an RF expert, but it sounds like it would be transmitting an empty carrier wave, not a real signal. Think of it as a big light source and the passive transmitters as mirrors used to flash signals back and forth with reflected light.

(Of course you'd also need a traditional wifi router for everything to connect through, and you might as well build that router into the same shell as the emitter thing, so it gets confusing. But the basic carrier wave wouldn't necessarily carry a signal.)


It looks like this can only transmit, not actually receive and interpret data-- and if you listen carefully, this is the kind of applications they're suggesting.

That's good for sensor nodes reporting status, but not so good for interactivity.


I don't think that's right, except in an extremely narrow technical sense. This specific technology they've developed is only for low-power transmission, because receiving wifi signals can already be done with little power. Device manufacturers can pair the new transmitter with any off-the-shelf wifi antenna for reception.


Any source for it being unidirectional?


Interesting idea. Things like this generally need a bigger antenna, since they're getting power from it. It probably means the base station has to run at high power most of the time, rather than throttling down for short range transmissions. If it went all the way to full wireless power, like some RFID tags, it would be more useful.

High data rate capability probably doesn't matter. Anything that's receiving or sending a substantial amount of data is probably using more power doing something with it than sending it.


Love to see someone makes watch/heart rate monitor with this technology that last months/years or maybe forever with tiny solar panel like what's on solar calculator.


It would be great if this could be used in (new) mobile phones and tablets in a low power mode when you don't need high data rates.


>Now, paper towels, coffee or sugar can communicate their status using WiFi.

Internet of edibles?

God help us all :)


You say that but if we could easily swallow or "tattoo" cameras/sensors, there are many medical upsides.


I wonder why there isn't a WEP-like protocol using AES-CCMP. (of course, without the problems of WEP)


old news. I guess I remember this news in February here at HN


This seem another wifi achievement only replicable in a desert environment, during a solar maximum, on the dark side of the moon, while holding a black cat.

If you are using a "wifi" antenna then that antenna is going to be picking up all sorts of little things. Normally these aren't much of an issue, but at these sensitivities suddenly that phone down the hall, the one not running at a 10e-3 power setting, is a real issue.

Also, February 2016?


There's nothing wrong with posting older material to HN if it hasn't had attention yet. I'm not sure if this story had, though.

The "new" in "Hacker News" is more like a second-hand clothing store in my neighbourhood that used to be called "New to You".


I'm not seeing anything to indicate that the passive transmitters are extra-sensitive. The whole wifi system is still running at full power and producing full-power signals; you've just shifted the power load from the remote transmitters (which are dependent on tiny batteries) to the central transmitter (which has a nice fat power main to draw from).




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: