Design of a filter for exctracting the unwanted, unknowed part of a mixed signal

[amixich]

Hello to all,

i'm newbie to digital signal processing area and therefore i hope i will find someone, to help, to advise me for my actual problem.

In my project I sample a vital parameter with 200Hz. When a body movement occur, the signal of the vital parameter is mixed with this movement. I have the possibilities to measure the movement trough an accelerometer.
Now, I would like to remove the unwanted signal (movement) from the vital parameter. I suppose, i will need to design a stopband. But the question is at which frequency? The movement can comprise more than one frequency.
Which types of filter should be used for such a challenge?

Any comment is welcome.

A. Mixich

 

[KlausST]

Hi,

the movement usally is from 0 to a max. frequency. But we can´t know this frequency.
So either you need to ask somebody related to your project who knows this information... or do a frequency analysis ... or just decide on your own.

For me it´s impossible to know if you are talking about a jetcraft, an ultrasonic transducer or a snake...

Klaus

 

[amixich]

Hi,

the movement usally is from 0 to a max. frequency. But we can´t know this frequency.
So either you need to ask somebody related to your project who knows this information... or do a frequency analysis ... or just decide on your own.

For me it´s impossible to know if you are talking about a jetcraft, an ultrasonic transducer or a snake...

Klaus

Hi Klaus,

i agree with your afirmation, but thinking deeper i got the idea to find maybe the correlation coefficient between the mixed signal and the movement. With a subtraction i can eliminate the part of the unwanted signal, or not? What is your position, to that?

Alex

 

[BradtheRad]

I am used to a medical technician telling me 'Now stay absolutely still during the test'. The technician takes readings. I suppose if I move once or twice, the technician recognizes it as an irregular reading, and ignores it. Your device needs to be smart enough to reject irregular readings. A simple bandpass (or bandstop) filter may not be sufficient.

Software could do the job, since it can take readings, and measure time from one normal reading to the next. This allows it to be flexible. (A simple filter cannot easily be made flexible.) When your software encounters a reading which is early, or too high or low, it rejects it. Or else it could keep a record of irregular readings. If normal readings were to stop then your device sends an alert saying 'abnormal readings' or 'bad connection', etc.