Skip to main content
Topic: Electrical network frequency analysis (Read 7794 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Electrical network frequency analysis

This looks pretty amazing if true. Comments on the article are not all believing.

Quote
ENF relies on frequency variations in the electricity supplied by the National Grid. Digital devices such as CCTV recorders, telephone recorders and camcorders that are plugged in to or located near the mains pick up these deviations in the power supply, which are caused by peaks and troughs in demand. Battery-powered devices are not immune to to ENF analysis, as grid frequency variations can be induced in their recordings from a distance.

At the Metropolitan Police's digital forensics lab in Penge, south London, scientists have created a database that has recorded these deviations once every one and a half seconds for the last five years. Over a short period they form a unique signature of the electrical frequency at that time, which research has shown is the same in London as it is in Glasgow.

On receipt of recordings made by the police or public, the scientists are able to detect the variations in mains electricity occuring at the time the recording was made. This signature is extracted and automatically matched against their ENF database, which indicates when it was made.
Source: The Register - Met lab claims 'biggest breakthrough since Watergate'

There's a link to the research in the comments too - Applications of ENF Criterion in Forensic Audio, Video, Computer and Telecommunication Analysis. Catalin GRIGORAS, PhD, Forensic Examiner (1.1MB PDF)

Electrical network frequency analysis

Reply #1
That looks like a very impressive technique ! Thanks for sharing the info.
I wonder how reliable it is (iow, if it's possible to cheat by modifying the signal).

Electrical network frequency analysis

Reply #2
That looks like a very impressive technique ! Thanks for sharing the info.
I wonder how reliable it is (iow, if it's possible to cheat by modifying the signal).

If you have the time to wade through the noise, there's some discussion of that in the article's comments. Replaying an old recording over speakers & re-recording it with a microphone is a simple suggestion to give a recording a newer timestamp.

Electrical network frequency analysis

Reply #3
That looks like a very impressive technique ! Thanks for sharing the info.
I wonder how reliable it is (iow, if it's possible to cheat by modifying the signal).

If you have the time to wade through the noise, there's some discussion of that in the article's comments. Replaying an old recording over speakers & re-recording it with a microphone is a simple suggestion to give a recording a newer timestamp.


There's a very  good chance that both "time stamps" would be preasent, and/or there would be obvious evidence of tampering.

Electrical network frequency analysis

Reply #4
There's a very  good chance that both "time stamps" would be preasent, and/or there would be obvious evidence of tampering.


So the basic theory is that any recording has a little hum?

If one had access to the database, one could make an original recording and dial up whatever time desired.

Electrical network frequency analysis

Reply #5
There's a very  good chance that both "time stamps" would be preasent, and/or there would be obvious evidence of tampering.


So the basic theory is that any recording has a little hum?


With modern tools, amazingly small amounts of hum can be reliably dtected and measured.

Quote
If one had access to the database, one could make an original recording and dial up whatever time desired.


FWIW

Electrical network frequency analysis

Reply #6
What about the use of an uninterruptible power supply (UPS) which filters out power line variations?

Electrical network frequency analysis

Reply #7
What about the use of an uninterruptible power supply (UPS) which filters out power line variations?


In the first post, it is claimed that you can detect this on battery powered devices that are not connected to the power line at all via reactive coupling.  Presumably if isolation is insufficient, then incomplete isolation provided by a UPS will also not be sufficient.

Electrical network frequency analysis

Reply #8
That looks like a very impressive technique ! Thanks for sharing the info.
I wonder how reliable it is (iow, if it's possible to cheat by modifying the signal).

If you have the time to wade through the noise, there's some discussion of that in the article's comments. Replaying an old recording over speakers & re-recording it with a microphone is a simple suggestion to give a recording a newer timestamp.


There's a very  good chance that both "time stamps" would be preasent, and/or there would be obvious evidence of tampering.


Although you could probably make a very convincing fake if you notch filtered first.

Electrical network frequency analysis

Reply #9
Put the battery powered audio recorder, and microphone, in a magnetically shielded box (mumetal) so that nothing is exposed to fluctuating electomagnetic fields.  No externally induced ENR would be present.

Electrical network frequency analysis

Reply #10
That looks like a very impressive technique ! Thanks for sharing the info.
I wonder how reliable it is (iow, if it's possible to cheat by modifying the signal).

If you have the time to wade through the noise, there's some discussion of that in the article's comments. Replaying an old recording over speakers & re-recording it with a microphone is a simple suggestion to give a recording a newer timestamp.


There's a very  good chance that both "time stamps" would be preasent, and/or there would be obvious evidence of tampering.


Although you could probably make a very convincing fake if you notch filtered first.


Notching out the ca. 60 Hz tone without leaving any artifacts is in actual practice, pretty hard to do. You end up leaving a noticable dip in the noise floor.

Electrical network frequency analysis

Reply #11
Notching out the ca. 60 Hz tone without leaving any artifacts is in actual practice, pretty hard to do. You end up leaving a noticable dip in the noise floor.


Well, 50 hz in this case.  I don't know that they do, but I could see filters for voice comms blocking all the bass (to reduce ambient sound energy even if not compressing) which would include at least fundamentals of the hum.

Electrical network frequency analysis

Reply #12
I'm skeptical...  There are no statistics in the research paper (proposal?) indicating the probability of a false positive or false negative.    I can see how this might work sometimes, but I'd like to see the results of some blind tests with a variety of recordings with a variety of equipment under a variety of real-world conditions. 

And, I doubt this is the "Greatest breakthrough since Watergate".  There have been lots of advancemsnts since 1972-1974, including the availability of desktop computers/software.  Forensic audio analysts have a lot of tools, and I'd be really surprised if this was at the top of their list...    With this "breakthrough", we still don't know what was on that 18 minute gap!

Electrical network frequency analysis

Reply #13
What about the use of an uninterruptible power supply (UPS) which filters out power line variations?
In the presentation, there's a plot of a results from a UPS, and it seems to be effective.

I didn't realise that the frequency of the mains fluctuates in parallel across the whole of Europe. It makes sense, but it hadn't occurred to me before.

Are power generators in the USA in a large scale grid like this, or is it more localised?


Whether it works for most recordings, or not, you've got to admire the genius that spotted / realised this.

Cheers,
David.


Electrical network frequency analysis

Reply #14
Are power generators in the USA in a large scale grid like this, or is it more localised?

There are just three grids in the US: East, West and Texas

Electrical network frequency analysis

Reply #15
Well, 50 hz in this case.  I don't know that they do, but I could see filters for voice comms blocking all the bass (to reduce ambient sound energy even if not compressing) which would include at least fundamentals of the hum.



I should have remembered harder... a typical filter  might have a passband of 300-2400 Hz, and I have a DSP one that in addition notches out constant tones.

Electrical network frequency analysis

Reply #16
I wonder:
1) How many audio recordings are used as evidence, where the time of recording is a concern?
2) With the US's much larger power grid, is the time signature as significant?
3) Do portable analog devices really pick up much/any power line noise?
4) Once the signal is digital, is the power line frequency a factor?
Kevin Graf :: aka Speedskater

Electrical network frequency analysis

Reply #17
You'll probably find the answers here...

Applications of ENF Analysis in Forensic Authentication of Digital Audio and Video Recordings

Small frequency variations in the electrical power network, which are the same through the area serviced by the network, can be used to assess the integrity of audio and video evidence. Small amounts of hum usually leak into a recording, and this provides a unique time signature. By archiving power line frequencies over many years, the time of a recording can be determined by comparing it to the information in the archive database. Similarly, discontinuities in hum frequency or multiple frequency components provide additional forensic evidence.

Author: Grigoras, Catalin
Affiliation: Bucharest S1, Romania
JAES Volume 57 Issue 9 pp. 643-661; September 2009

Electrical network frequency analysis

Reply #18
2) With the US's much larger power grid, is the time signature as significant?


The US grids are actually much smaller, so this probably works better.  You can not only tell when, but also which region of the country something was recorded in.

4) Once the signal is digital, is the power line frequency a factor?


A factor in what?  Changes in the power line frequency and amplitude are what they're trying to measure, so they're obviously very important.  Not sure where you are going with the digital part. 

Electrical network frequency analysis

Reply #19
4) Once the signal is digital, is the power line frequency a factor?

A factor in what?  Changes in the power line frequency and amplitude are what they're trying to measure, so they're obviously very important.  Not sure where you are going with the digital part.

What I was trying to ask was - With a portable battery powered recorder, once the input analog signal goes through the ADC then the AC power frequency would have no impact on a digital signal.  So the only source of information would be the input stage. (more of a thought than a question)
Kevin Graf :: aka Speedskater

Electrical network frequency analysis

Reply #20
2) With the US's much larger power grid, is the time signature as significant?

The US grids are actually much smaller, so this probably works better.  You can not only tell when, but also which region of the country something was recorded in.


I thought the the story was about the GB grid.  In August of 2003 the Eastern North American grid power outage covered an awful lot of territory.
Kevin Graf :: aka Speedskater

Electrical network frequency analysis

Reply #21
So after looking at the Power Point Presentation (I hate PP's without added text or audio) I see that he used the European grid (the GB grid came later) it also shows the output of a DAT with no further details.
Kevin Graf :: aka Speedskater

Electrical network frequency analysis

Reply #22
3) Do portable analog devices really pick up much/any power line noise?
4) Once the signal is digital, is the power line frequency a factor?
Not sure if this is what you were asking, but analogue recording won't work for this - e.g. AFAICT cassette tapes already have far too much speed variation to accurately record the mains frequency.

I'm amazed there's routinely enough interference for this to work, but a sharp filter/FFT can really dig down through noise or "normal" audio to a "hidden" signal like this when the frequency is known.

Cheers,
David.

Electrical network frequency analysis

Reply #23
Not sure if this is what you were asking, but analogue recording won't work for this - e.g. AFAICT cassette tapes already have far too much speed variation to accurately record the mains frequency.

So, do you think that a wow and flutter figure of, say, 0.045% RMS would be sufficient to mask changes in line frequency of +/- 0.2% peak as seems to be the maximum deviation here in the UK?

Bear in mind that many high-end cassette decks used DC servo motors to drive the capstan(s) and commonly achieved this figure if well maintained.

Wouldn't the use of a DC servo motor for capstan drive also mean that long-term speed stability should also be very good?

Electrical network frequency analysis

Reply #24
I have no idea. I don't even have two tape decks that run at exactly the same speed! OK, maybe I do have two that match - but I have more that don't. I also have a few that drift slightly over an hour (e.g. one end of tape to the other).

If you have an analogue recording, a claimed time of recording, and a claimed machine that recorded it, then I suppose you can show there's a match (or not). But if you don't know anything about the recorder, then you can't be sure whether any apparent mismatch is due to the recorder itself, or because the recording is a fake.

Cheers,
David.

 
SimplePortal 1.0.0 RC1 © 2008-2020