Needed: Hosptial technicians,signal processers.

[eigen]

I was wondering how difficult it would be to take a commercially available eeg sensor? and have it communicate with a standard desktop. What sort of equipment would it take? I would like there to be no signal processing done before the data was written to disk.How feasible? But more important how much?

Reasons: I am gonna use the data as seeds for a prng. I am not satisfied with any of the other novel collection means such as http://www.wired.com/wired/archive/11.08/random.html as these are not random.


I found this interesting as well http://www.cs.colostate.edu/eeg/

 

[Geniere]

I toyed with EEG devices cc. 1965-1984 so my knowledge is a little dated. The frequency bandwidth is very low ranging from near DC to about 40hz. I?m not sure what you mean by ?signal processing?. To me that term is associated with converting analog to its digital representation. You will defiantly need to amplify the signal before input to the computer.

To acquire the EEG signal without use of a commercial machine is difficult unless you have a great deal of design experience in small signal amplifiers. You will have to pre-amplify voltages as little as a few microvolts and include a circuit to eliminate common-mode noise.

After amplification, the rest may be easy if you can get a sound card for your PC that is DC coupled, not AC coupled and has a flat response from DC to 30-40hz. Considering the great number of ?purist? audiophiles, such a device must be available. If not available, there are a slew of laboratory type analog data acquisition PC cards; some may have the necessary amplification built in. Some will include the analysis software.

In brief, via appropriate electrodes in conjunction with shielded cables connected to high gain amplifier introduce the signal to the computer via a DC coupled audio card. Afterwards you will need the necessary software for display and analysis.

 

[eigen]

Originally posted by: Geniere
I toyed with EEG devices cc. 1965-1984 so my knowledge is a little dated. The frequency bandwidth is very low ranging from near DC to about 40hz. I?m not sure what you mean by ?signal processing?. To me that term is associated with converting analog to its digital representation. You will defiantly need to amplify the signal before input to the computer.

To acquire the EEG signal without use of a commercial machine is difficult unless you have a great deal of design experience in small signal amplifiers. You will have to pre-amplify voltages as little as a few microvolts and include a circuit to eliminate common-mode noise.

After amplification, the rest may be easy if you can get a sound card for your PC that is DC coupled, not AC coupled and has a flat response from DC to 30-40hz. Considering the great number of ?purist? audiophiles, such a device must be available. If not available, there are a slew of laboratory type analog data acquisition PC cards; some may have the necessary amplification built in. Some will include the analysis software.

In brief, via appropriate electrodes in conjunction with shielded cables connected to high gain amplifier introduce the signal to the computer via a DC coupled audio card. Afterwards you will need the necessary software for display and analysis.

Woah, Thanks alot that really helped me think about some of the problems I beleive I will encounter.

 

[f95toli]

One question: Why would you like to REMOVE noise in order to get a "random" seed?
But I assume this is mostly for fun?

As I am sure you know there are many other wats to get a signal which is as "random" as an eeg signal (besides, AFAIK EEG signals are more or less periodic so I am not convinced it is any better than for example thermal noise from a resistor).

 

[eigen]

Originally posted by: f95toli
One question: Why would you like to REMOVE noise in order to get a "random" seed?
But I assume this is mostly for fun?

As I am sure you know there are many other wats to get a signal which is as "random" as an eeg signal (besides, AFAIK EEG signals are more or less periodic so I am not convinced it is any better than for example thermal noise from a resistor).

How is it that I am removing noise?Am I missing something.

Yes this is for fun. I think it will be extra nerdy cool to sit at my dek with a sensor attached to my head.


The problems I have with the other methods I linked to is that they are not random but merely chaotic and in fact I think over a long enough period of time I believe that they will converge to the manldebrot set. On the other hand I dont think brainwaves do this.

 

[f95toli]

The whole point of the carefully designed electronics Geniere described is to be able to aquire very small signals which means getting rid of as much noise as possible, and noise is (almost) always random.

So if you want something random you might as well just amplify just about any small and noisy signal 100 000 times or so, just make sure the spectrum is reasonably white.
One way of getting reasonably good seeds is to amplify the signal from a reverse biased diode, I think this was actually used in an old computer, the ABC 80
http://www.old-computers.com/m...uter.asp?st=1&c=37

One way to get truly random numbers would be to measure radioactive decay (which is TRULY random).
I think it should be possible to DIY this by rebuilding an old smoke detector. However, I am just guessing and I have not actually tried.

 

[Matthias99]

Originally posted by: f95toli

So if you want something random you might as well just amplify just about any small and noisy signal 100 000 times or so, just make sure the spectrum is reasonably white.

Personally, I'd go with this method -- hook up a broad-spectrum microphone listening to ambient sound to your sound card, crank up the recording volume all the way, and sample the bitstream you get out of it to seed your RNG. You need a sound card and mic that are suitably unbiased over the frequency range you're sampling, but this would be quite random.

Of course, I imagine a similar process might work with an EEG recorder -- but EEG data is decidedly non-random; there are strong frequency bands present, and probably harmonics as well. You can also control EEG output (at least to some extent) via biofeedback. An epileptic having a seizure would give you random EEG data, but at that point you have bigger problems on your hands.

 

[KalTorak]

You could just use the RNG in the Firmware Hub, if you're on an Intel chipset with an FWH...

 

[Vee]

Originally posted by: eigen
The problems I have with the other methods I linked to is that they are not random but merely chaotic

So is everything. The entire universe is exactly organized, and connected. You can have unpredictable (chaotic), but you can't have truly "random", because that's just an imagined concept.

But ok, ok, I see what you're trying to accomplish.

 

[f95toli]

That is not exactly true. As far as we know certain processes are truly random (and not just chaotic), radioactive decay is one example of this but there are many others.

 

[FrankSchwab]

Thought you might be interested in this link that I ran across today:
OpenEEG at SourceForge

Software and hardware for an EEG device.

Good Luck,

/frank

 

[Vee]

Originally posted by: f95toli
That is not exactly true. As far as we know certain processes are truly random (and not just chaotic), radioactive decay is one example of this but there are many others.

Ho ho.
The model which we try to fit to these processes, contains the concept of "random".
I'm convinced "random" doesn't exist outside our minds.
Since chaotic means unpredictable, you have no means to decide what "as far as we know". It could just as well be chaotic as random.

 

[f95toli]

No, there is a difference between random and chaotic. There are many simple (non-chaotic) systems that are random, a simple example is a particle tunneling out from a well (which is also a simplified model for radioactive decay), the hamiltonian is NOT chaotic and you can solve for all eigenstates and eigenvalues but the eigenstates represents probability distributions which means there is no way to predict WHEN a particle will tunnel; the process is complettely random.

And, frankly, nature does not care what you or anybody else thinks about procesees like this, as far as we know the process is random. Wheter or not the universe is ultimately deterministic or not is a question for philosophy , not science.

 

[Vee]

Originally posted by: f95toli
No, there is a difference between random and chaotic. There are many simple (non-chaotic) systems that are random, a simple example is a particle tunneling out from a well (which is also a simplified model for radioactive decay), the hamiltonian is NOT chaotic and you can solve for all eigenstates and eigenvalues but the eigenstates represents probability distributions which means there is no way to predict WHEN a particle will tunnel; the process is complettely random.

And, frankly, nature does not care what you or anybody else thinks about procesees like this, as far as we know the process is random. Wheter or not the universe is ultimately deterministic or not is a question for philosophy , not science.

Obviously, nature doesn't care about what models we create, and try to fit nature.
And yes, I think we approach this from different philosophical viewpoints. The fact that there is no way to predict when a particle will tunnel, doesn't mean that it is random. Only that a random model will be appropriate for now. You're entirely right about "as far as we know the process is random". I'm not really disputing that. The thing is, I would say: "as far as we know, the process could be described by a random model".

Your argument about why it is not chaotic, is interesting. I'll try to figure it out.

 

[jagec]

Originally posted by: Vee
Ho ho.
The model which we try to fit to these processes, contains the concept of "random".
I'm convinced "random" doesn't exist outside our minds.
Since chaotic means unpredictable, you have no means to decide what "as far as we know". It could just as well be chaotic as random.

Maybe you would agree with Cian Dorr on the simplicity of everything? (It's a PDF, his dissertation)

Personally, I'd rather keep my free will, illusionary or not. Time will tell if science agrees with me.

 

[ZUnit]

eigen what you trying to do is completely useless.

 

[Vee]

Originally posted by: jagec
Maybe you would agree with Cian Dorr on the simplicity of everything? (It's a PDF, his dissertation)

Personally, I'd rather keep my free will, illusionary or not. Time will tell if science agrees with me.

I agree. But I believe, that within all frames that the concept 'free will' has any meaning at all, we do indeed have free will.

I'm sure the link is amusing. I'll read it when I get the time. Thankyou.