February 24, 2018

Monitoring all the electrical and hydraulic appliances in your house

Dan Wallach recently wrote about his smart electric meter, which keeps track of the second-by-second current draw of his whole house. But what he might like to know is, exactly what appliance is on at what time? How could you measure that?

You might think that one would have to instrument each different circuit at the breaker box, or every individual electric plug at the outlet. This would be expensive, not particularly for all the little sensors but for the labor of an electrician to install everything.

Recent “gee whiz” research by Professor Shwetak Patel‘s group at the University of Washington provides a really elegant solution. Every appliance you own–your refrigerator, your flat-screen TV, your toaster–has a different “electrical noise signature” that it draws from the wires in your house. When you turn it on, this signal is (inadvertently) sent through the electric wires to the circuit-breaker box. It’s not necessary (as one commenter suggested) to buy “smart appliances” that send purpose-designed on-off signals; your “dumb” appliances already send their own noise signatures.

Patel’s group built a device that you plug in to an electrical outlet, which figures out when your appliances are turning on and off. The device is equipped with a database of common signatures (it can tell one brand of TV from another!) and with machine-learning algorithms that figure out the unique characteristics of your particular devices (if you have two “identical” Toshiba TVs, it can tell them apart!). Patel’s device could be an extremely useful “green technology” to help consumers painlessly reduce their electricity consumption.

Patel can do the same trick on your water pipes. Each toilet flush or shower faucet naturally sends a different acoustic pressure signal, and a single sensor can monitor all your devices.

Of course, in addition to the “green” advantages of this technology, there are privacy implications. Even without your consent, the electric company and the water company are permitted to continuously measure your use of electricity and water; taken to the extreme, this monitoring alone could tell them exactly when you use each and every device in your house.


  1. Anonymous says:

    The first question that came to my mind after reading this was how far up the line can these signatures be read with sufficient precision to be an invasion of privacy? I’d guess that transformers might play havoc with them, and as soon as the power line reaches the transformer down the street any given individual’s signature would presumably be lost among the cacophony of everyone else’s.

    But like I said, I’m just guessing (I’m not an electrical engineer) and so am curious to hear the opinion of someone who possesses the relevant technical expertise. Also, there’s probably a related RF emission that would likewise be detectable with the proper equipment, and if that’s the case then this kind of monitoring could conceivably be done remotely with no direct connection to the power line.

    And if transformers do distort or eliminate the signatures (in the wires, not the RF signatures if there are any), then does that mean running our grid power through a UPS would work to achieve the same effect?

    Interesting stuff, thanks for posting it (and thanks in advance to anyone who answers my questions…)

  2. Anonymous says:

    You are right. The line inductance in the wire to the pole top and the transformer itself isolates the homes for each other. The signal will attenuate greatly.

  3. Seth Finkelstein says:

    > Patel’s group built a device that you plug in to an electrical outlet …

    Cool. But off the top of my head, I can think of a few issues.

    How well does this work for multiple lines? That is, I assume you need a device for each line in the house. And does it ever get confused between tracking “on” and “off”? Is it smart enough not to think a lightning storm is a crazy appliance?

  4. Anonymous says:

    While this implementation is certainly novel, the idea of individual signatures of electric/electronic devices and monitoring for that is not new at all.

    High-intensity discharge lamps, such as metal halides or high-pressure sodium lamps have been used in the indoor pot growing scene for decades, and block-monitoring for their signatures has been used as a tool to find these growers for many years as well.

    The electric companies have abetted this, by identifying “unusual” electric consumption, and the monitoring the actual draw.

    I point this out to show just how much of another tool of big-brotherism this will turn out to be, especially as it is being hailed for its “green” usefuleness.
    GREEN POLICE usefullness more like?

    “you left the room lights on in your TV room even while you had gone to bed, that’s against the LAW”, and so on.

    The growers resorted to DC-uncoupling the lamps (akin to a battery-back-up) to avoid signature detection, I wonder it these profile-devices can see through that?
    (or stealing power from the local distribution hub, not because they didn’t want to pay, but because they didn’t want to appear suspicious by drawing 10kwh and more all day).

    As the poster above has said, we will need firewalls now, hopefully that will be developed as well

    • Anonymous says:

      This raises another enviro-issue: Drug prohibition is “brown” (that is, eco-hostile). Legalize pot, and growers won’t have to hide their gardens indoors and then burn God knows how much fossil fuels running high-intensity lamps.

      It is a general rule that over-regulation by government causes inefficiency and waste. Drug prohibition, copyright law, and the like are all cases in point, and all seem to be at the behest of particular industries (Drug prohibition can be traced back to massive propaganda campaigns in the 20s directed, ultimately, by lumber, pharma, and tobacco companies. The main target: pot. Because not only is the plant a potential tobacco substitute, but has pharmacologically active properties that can’t be patented, AND the same plant can be used to make a lot of materials, including paper, that the lumber industry supplies.

      All of this over-regulation causes waste. First, the up-front waste from obeying the regulations. If the regulations have any effect at all it is because a free market would seek a different equilibrium. Usually the equilibrium sought by a free market is more efficient. Ergo, to the extent the regulations have any effect at all they cause waste.

      Second, the enforcement of the regulations itself causes waste.

      Third, the regulations usually exist to let some business or industry extract rents. To some extent this overlaps with the waste from distortion of the economy, but it can go further if businesses become enormously powerful (e.g. Microsoft) that otherwise wouldn’t have, and then throw their weight around; or create inefficient defacto standards (Windows, DRM, HDCP/HDMI, etc. — how much electricity is wasted running overpowered computers to keep bloated, unstable Windows from running like a pig in molasses? How much wastefully decrypting, and then encrypting, and then decrypting again, frames of Avatar, whose contents are public knowledge rather than top secret?)
      — though arguably some of this waste (the DRM-induced waste, particularly) overlaps with the second category (expenditures in the enforcement of unnecessary regulations).

      Bottom line: over-regulation harms the environment as well as the economy.