Calculating SNR of a Matched filter

[mordak]

Hey guys,

I'm going to implement a matched filter, but faced a problem with the basic idea behind that. When we apply a signal to the input of the matched filter, say a sine wave, and assuming we have knowledge about the amount of the noise at the input, we can easily calculate SNR of the input signal. However, when this signal passes through the matched filter, the output is just one number (amplitude of the original signal), so how the output SNR would be defined for this kind of filters?

Thanks!

 

[crutschow]

It's determined by the noise spectrum and the noise bandwidth of the matched filter. The first must be determined and the second can be calculated from the filter transfer function.

 

[D.A.(Tony)Stewart]

Noise Sources:
Internal

Random noise in resistors and transistors
Mixer noise
Undesired cross-coupling noise
Power supply noise

External
EMI, Co-channel, Adjacent channel, cosmic noise, Power Line transients
etc.

SNR= Signal Power/Noise Power

Details

 

[mordak]

It's determined by the noise spectrum and the noise bandwidth of the matched filter. The first must be determined and the second can be calculated from the filter transfer function.

Thanks for the comment. Can you please walk me through this or give me a ref with a numerical example in it? I guess the transfer function of the low pass filter in the matched filter would be a sinc, so the input noise would be multiplied by that, but I'm still not clear with the whole analytic calculation of it.

Noise Sources:
Internal

Random noise in resistors and transistors
Mixer noise
Undesired cross-coupling noise
Power supply noise

External
EMI, Co-channel, Adjacent channel, cosmic noise, Power Line transients
etc.

SNR= Signal Power/Noise Power

Details

Thanks for the reply. I'm totally familiar with the concept of SNR, but I have problem with calculating output noise of a matched filter, since output is a number, not a signal.

 

[D.A.(Tony)Stewart]

Read link in Details... Both signal and noise are numbers for power, some of which can be calculated, others must be measured.