De slimme meter (Algemeen)

Uit het NRC
Hoe slimme meters ontdekken wat jij op tv kijkt. http://www.nrc.nl/nieuws/2014/06/17/hoe-slimme-meters-ontdekken-wat-jij-op-tv-kijkt
Jordan Roberts:

Slimme meters geven inzicht in hoeveel stroom je gebruikt. Maar dankzij die informatie krijgen energiebedrijven ook veel andere gegevens binnen. Ze kunnen bijvoorbeeld achterhalen welke tv-programma’s je bekijkt, blijkt uit onderzoek.

Smart meters may be getting too smart for our own good.

As businesses look for new ways to gain insight into consumers, utility meters that wirelessly transmit energy-usage data are increasingly drawing attention because of what they can reveal about our behavior at home, such as when and how often we use certain appliances.

Last month, a unit of WPP, the world’s biggest advertising agency, announced it was teaming up with London-based software company Onzo to study ways to collect smart-meter data on household energy use. Onzo CEO Joel Hagan told Bloomberg News that the information has the potential to “open the door of the home.”

But unlocking the front door is just the beginning. The next stop for big data could be on the sofa next to us as we watch TV. Information flowing through smart meters can be mined to determine users’ viewing habits — not just that people are watching TV, but which programs they’re watching, down to individual scenes at specific times, according to a little-known study by the University of Applied Sciences in Steinfurt, Germany.

The research, which was published in 2012, measured how much power it takes to display certain programs on a television screen. Looking at seven movies and two television shows on five different brands of TV sets, the researchers found that each program had a unique power signature based on how much electric current was needed to show the images on the screen. Among the programs used for the tests was “Star Trek.”

Once the programs’ signatures were identified, the researchers found they could then match that information with data coming out of a smart meter. That meant a power company or other entity could mine this data to determine what a household was watching. The test was comprehensive enough to show the technique is broadly applicable and that there is an urgent need for stronger protections for smart-meter data, the researchers said.

As smart meters roll out around the world, the new technology has been met with anxiety and, at times, protests over security and privacy concerns. Last month, a protester was injured during an altercation with a water-utility employee at a demonstration in Ireland, and citizens’ groups in Australia, the U.S. and Canada have mounted campaigns against the installation of smart meters.

By harvesting smart-meter data, companies can identify the brands of appliances people are using and create detailed marketing profiles, such as which homes have children and which have laborious mealtime preparation routines, said Ulrich Greveler, a researcher on the TV project.

“This is definitely interesting data for advertising agencies but also a huge invasion of consumer privacy,” Greveler wrote in an e-mail.

Utilities, however, have touted the potential of smart meters to lower energy bills, reduce energy consumption and prevent blackouts. One utility uses data gleaned from smart meters to alert customers in Delaware and Maryland to spikes in demand and offer discounts for immediately reducing consumption.

Smart meters “enable two-way power and information flows between the utility and the consumer” and the data they provide “improve the efficiency and reliability of the electric system,” said Lisa Wood, executive director of the Institute for Electric Innovation, part of The Edison Foundation, an organization focused on the electric industry, in a report on the subject.

This commercial-privacy trade-off isn’t new. In exchange for getting to use a search engine such as Google for free, Web surfers pay by surrendering personal information that is used to target ads. By using in-car transponders to electronically pay bridge tolls, motorists speed their commute but submit to having their routes tracked.

Smart meters are just the latest example of the compromise consumers may face. Even amid the technology’s potential cost savings to us, there’s still no free lunch. Especially if you’re cooking it in that Internet-connected microwave.

http://www.bloomberg.com/news/2014-05-11/wpp-unit-onzo-study-harvesting-smart-meter-data.html
http://www.nds.rub.de/media/nds/veroeffentlichungen/2012/07/24/ike2012.pdf

Jaqueline,
Waar haal jij je onbegrensde vertrouwen in de slimme meter toch vandaan? Het hele WWW staat vol met onafhankelijke onderzoeken naar de mogelijkheden en effecten van de van de slimme meter. Deze onderzoeken bewijzen vaak dat de consument gewoonweg met oneigenlijke argumenten de slimme meter opgedrongen krijgt. Jij probeert op dit forum steeds weer opnieuw aan te tonen dat de energie boeren het zo goed voor hebben met hun slimme meters. Elke keer probeer jij gebruikers die tegen de slimme meter zijn, ongeveer voor gek te verklaren. Waar heb jij zoveel kennis vergaard dat jij met zekerheid kunt zeggen dat de slimme meter geen spion in ons huis is, dat er geen gezondheidsrisico’s zitten aan de gsm / gprs functie zitten? Waar komt dat eindeloze vertrouwen van jou vandaan Jaqueline? Dat jij altijd als het er kritische berichten worden geplaatst in dit forum “onopvallend” steeds weer de kant van de netbeheer kiest…
Hier een klein stukje uit een universitair rapport/ onderzoek:

De volgende tekst komt uit een Europees universitair onderzoek:
Titel “Het bewerken van slimme meter data om huishoud apparaten te identificeren”

A LGORITHM E VALUATION AND L IMITATIONS
In order to analyze the performance of the AppliSense
algorithm, we installed the whole system in a laboratory en-
vironment. For the evaluation, we used a controlled set of
appliances which typically occur in a student’s household.
Table II provides an overview of the appliances, their real
power consumption stated on the manufacturer label, their
verified real power range in operation (measured by a sepa-
rate power monitor), the appliance category (O for ohmic, I
for ohmic-inductive, and C for ohmic-capacitive), and the
real power that is obtained as part of the power signature
using our smartphone application. All devices were connect-
ed to the same phase over the whole evaluation. Some of the
appliances have power consumptions within the same range.
However, if belonging to different categories, we should still
be able to differentiate the corresponding events.

T ABLE II. A PPLIANCES USED FOR ALGORITHM EVALUATION
Appliance Light bulb Kettle Heater CD player Fan Notebook Fluorescent
lamp Wii Labeled Power 75W 200W 2000W 13W 50W 72W
Power Range Category 6W to 2000W:

During times when only a single appliance was active,
the algorithm identified the on/off events of all devices ex-
cept the CD player correctly. Every device was turned on
and off at least three times. The edges caused by the CD
player were not recognized neither when being turned on nor
when being turned off. This can be explained through the
limitations introduced by the filtering. Using a window size
of 5 samples in our test scenario leads to a lower boundary of
10VA for edges that can be recognized. The CD player has a
relatively high standby consumption of 6W compared to its
3 – 7W in operation. While the median filter does not influ-
ence the signal, the constant 3W during operation result in a
Figure 11. Device recognition: Comparing a detected power edge to
a known signature in the Euclidian dQ/dP space.
step-wise increase of 0.6VA after the application of the mean
filter. This increase is too small (<<2VA) to be detected as
an event using the chosen median/mean filter.
Next, we combined the use of multiple devices in a ran-
dom order. Although the CD player cannot be recognized,
we operated it and other devices with unknown signatures
from time to time to vary the base line consumption and to
have more appliances concurrently running. Over a time
span of several hours, we documented 80 switching events of
which 77 were identified correctly.
Figure 12 shows a sample labeling output of the algo-
rithm (for a simulated office environment). After the note-
book has been turned on, different devices were concurrently
used and a kettle was operated. However, the red circle high-
lights a moment at which the office lamp is turned on but the
event is not detected. This is due to the oscillations caused by
a device that was operating at the same time. A second prob-
lem (not depicted) occurred when operating the notebook.
Due to the different battery levels, the power consumption
had varied compared to the one registered in the appliance
signature database. This led to the correct identification of
the edge, but no appliance signature could be matched to the
detected event.
These two examples outline limitations of the current im-
plementation. Oscillations caused by operating devices can
mask the consumption, especially of low power drawing ap-
pliances. This could in particular be a problem in larger
households (e.g., family houses) with lots of appliances and
activity. In addition, the algorithm cannot detect devices that
do not have well-defined operation states but have a continu-
ously changing consumption. This is due to the initial as-
sumptions regarding the algorithm design and the tradeoff
for relying on a single sensor system with a 1Hz sampling
frequency. In the conducted laboratory study we observed
that the appliance signatures recorded with the smartphone
application were very reliable. That is, the delta vectors ob-
tained with the measurement function when turning an appli-
ance on/off are stable and reproducible over time. However,
this may be different in a more dynamic home environment –
there the algorithm may need several (slightly different) sig-
natures per device to reliably recognize appliances.
Overall, the evaluation shows promising results. We gen-
erated 144 device-switching events in our test scenario.

Nou Jacqueline ben benieuwd hoe je het bovenstaande gaat weerleggen. Of afdoen als onzin.
(uiteindelijk valt de slimme meter discussie uit in twee kampen netbeheer, leveranciers en overheid aan de ene kant. De gebruikers aan de andere kant)

Groet,
Calimero…
Enexis is groot ik ben klein en ik wil helemaal geen slimme meter…

http://tweakers.net/nieuws/107270/software-update-moet-storing-slimme-gasmeters-binnenkort-verhelpen.html
Software-update moet storing slimme gasmeters binnenkort verhelpen

Door Krijn Soeteman, maandag 11 januari 2016 15:58, 105 reacties • Feedback

De storing waardoor bepaalde slimme gasmeters hun gasmeterstanden niet meer doorgeven, lijkt opgelost. Meterleverancier Iskrameco test de update deze week. Als alles goed gaat, verspreidt de leverancier de nieuwe software vanaf zaterdag.

Om de update binnen te krijgen, hoeft de klant niets te doen, zo meldt Netbeheer Nederland in een e-mail aan Tweakers. Wel kan het enkele weken duren voordat alle 400.000 getroffen adressen de update binnen hebben. Pas als de nieuwe software binnen is, kan de meter weer zelf actuele gasmeterstanden doorgeven aan apps en aan de energieleverancier.

Omdat de meting enkele weken niet is doorgegeven, kan het zijn dat de energieleverancier bij facturatie in sommige gevallen een schatting gebruikt. Ook is het mogelijk dat de leverancier een klant vraagt handmatig de gasmeterstanden door te geven. De gasmeters hebben wel het verbruik gemeten tijdens de storing.

Op vijf januari bracht Netbeheer Nederland naar buiten dat elektrameter Iskra-type 382 DSMR 2.2+ samen met een Flonidan- of Landis&Gyr-gasmeter de gasstanden niet meer doorgaf sinds 1 januari van dit jaar. Het softwareprobleem betrof volgens de beheerder een ‘verlate millenniumbug’.