15

I've recently seen some commotion online about "smart meters," electronic power consumption meters that are replacing the mechanical meters in some areas. My understanding is that these meters have radio transmitters that broadcast each household's power consumption (I'm not sure where these transmissions are received) once or several times per day (the effectiveness of this is discussed in this question).

One example of such commotion is this YouTube video, which claims that smart meters are an invasion of privacy, because:

  • They can see when you turn appliances on and off. This seems to imply that smart meters transmit real-time data. I would think that they would transmit a running total at intervals, rather than the power consumption on a per-minute or per-hour basis.
  • They can measure the power consumption of individual appliances. I'm extremely skeptical of this. Electricity just doesn't work that way. If the point above is true, then at best they could guess at what kind of appliance is on just after it was turned on, based on the increase of consumption. But that wouldn't be accurate at all.
  • They record your personal living patterns. This follows from the other two, and is the ultimate conclusion that smart meters will be used as surveillance devices.

Is the breadth of information visible to a smart meter anywhere near this level of accuracy? It defies my understanding of how electricity works in a home, but that understanding is far from complete. From what I've seen, the people spreading this warning are unusually calm and well spoken, and it's causing me to second guess myself.

Community
  • 1
Carson Myers
  • 2,755
  • 1
  • 16
  • 16
  • 5
    "Sure, grow-ops are illegal, and if you're not doing illegal things, you've got nothing to worry about." -- Pretty much any violation of rights and draconian action can be justified this way. Whether or not they are a violation of rights, this is the worst defense of a law. By that reasoning we should put jaywalkers to death. After all, if you're not doing anything illegal you have nothing to work about. – Russell Steen Jul 29 '11 at 05:54
  • @Russell I completely agree, but I don't know if claiming a violation of rights is valid here -- it feels to me like saying ISPs are evil because the logs they keep can incriminate you. I think the smart meter data has the side effect of being supporting evidence if it's subpoenaed, much like any other information about you kept by various companies and agencies, so calling it draconian is a bit of an overreaction. – Carson Myers Jul 29 '11 at 06:14
  • Hi Carson, can you please limit this question to one verifiable fact please? As it is it will tend to generate a bunch of speculation... – Sklivvz Jul 29 '11 at 06:18
  • @Sklivvz updated to include only three points, which are related to each other. – Carson Myers Jul 29 '11 at 06:26
  • Interesting extra privacy issue: the security implications if someone were to hack into these (I have seen this demonstrated) they could see when you were out, on holiday etc... – Rory Alsop Feb 27 '12 at 23:12

1 Answers1

17

The IEEE Spectrum magazine has discussed this issue in the past. Apparently it is clear from the data generated by these meters that certain appliances can be easily distinguished from others with reasonable accuracy and certainty. They reference a study from the NIST which illustrates

frequent meter readings may provide a detailed timeline of activities occurring inside a metered location and could lead to knowledge about specific equipment usage

And they use this graph to demonstrate: NISTIR 7628 Volume 2 Figure 5-1: Power Usage to Personal Activity Mapping

So basically because they sample the total energy flow at a fairly high frequency, they can estimate which devices are being used at which times based on the characteristics of those devices, which for a utility company should be easy to determine. It isn't perfect, but it's accurate enough especially if it can be correlated with smart meters for water, etc, because then they can map water usage to dishwashers, etc, and more accurately determine which electrical loads correspond to which major appliances.

Mr. Shiny and New 安宇
  • 512
  • 4
  • 8
  • That's interesting, I didn't realize the peaks had such definite shapes. – Carson Myers Jul 29 '11 at 21:57
  • 1
    I don't see how they can apart toaster and kettle. – vartec Jul 31 '11 at 09:34
  • @vartec: The toaster has an immediate drop-to-zero when finishing, the kettle maintains a slightly larger consumption for a few minutes after finishing - probably because electric kettles have lights that remain on even while not actively heating. –  Jul 31 '11 at 10:22
  • 1
    @Joe: I don't buy it. Only some models have any lights at all, and ones that do, have LEDs, which are have power consumption so low, that it wouldn't be noticeable on this graph. – vartec Jul 31 '11 at 16:54
  • One caveat to this research (which you can see from the graph) is that it requires sampling at a *very* high frequency; by the looks of the graph, this is picking up 1-minute intervals. Most hardware tracks 60- or 15- or sometimes 5-minute intervals, which gives you *way* less fidelity and you can't really see the peaks. So this is more of a theoretical invasion of privacy than a practical one. – Aaronaught Aug 05 '11 at 13:48
  • 1
    @Aaronaught: That may be true now (I'm not sure and a quick scan of the NIST report didn't reveal what hardware they used nor its installed base size) but I think it's a safe assumption that as utilities find a use for higher-frequency sampling they'll deploy the meters for it. – Mr. Shiny and New 安宇 Aug 05 '11 at 14:54
  • Perhaps, but you have to understand how much that costs. The utility industry moves at a glacial pace - even the most "cutting-edge" systems tend to incorporate a lot of legacy COBOL. With 1-minute intervals we are talking hundreds of terabytes of data and a whole lot of dedicated, distributed hardware to run it (keep in mind that all of this data actually has to be validated before any money can be attached to it!). Anything may happen in the distant future, but in the *near* future, this isn't going to be deployed nationwide. – Aaronaught Aug 05 '11 at 16:48
  • @Aaronaught: Oh, I know that it costs a lot to deploy meters nationwide. But the thing is that once the overall infrastructure has been upgraded from dumb to smart, making it smarter is relatively easy. Billing will probably not happen on a per-second basis but rather per-billing period. How many billing periods are needed? We have 4 in ontario (peak, 2 mid-peak, and off peak). So those old cobol systems can be fed aggregate data even if the rest of the system collects more. And those smart meters can probably be configured to send more or less data as needed. – Mr. Shiny and New 安宇 Aug 05 '11 at 17:04
  • There's only one mid-peak band in Ontario (may happen twice on some days, but it's the same rate). Time-Of-Use/Time-Of-Day billing can be accomplished with hourly data or even non-interval, banded register meters - however, the hardware generally only supports 4 buckets. So you can feed the billing systems aggregate data, but you still need an MDM to aggregate it, and if it doesn't store the original intervals, then you've lost all this ad-hoc data-mining capability. Plus there's a tangled web of regulatory and communication issues to work out. It's not as simple as just making them smarter. – Aaronaught Aug 05 '11 at 17:15
  • @Aaronaught: I'm just pointing out that just because the deployed hardware or software doesn't right-now collect all that data doesn't mean that parts of the system can't already do that, and that the system won't be upgraded more later on. Like how telephone systems had 90% of what was needed for caller-id display decades before it was made available in phones. – Mr. Shiny and New 安宇 Aug 05 '11 at 18:13
  • I understand, and I'm not trying to claim that any part of your answer is incorrect. I just wanted to highlight the distinction between the *distant potential* for an invasion of privacy and the *practical likelihood* of such an event. The question refers to "smart meters" in general, and this answer refers to a specific (and still somewhat rare) configuration of one. – Aaronaught Aug 05 '11 at 18:25