# Calculate the minimum data rate required for transmission of EEG signals (0.4-64.4Hz)

I'm a little bit confused here.

I assumed that the transmission is going to be PCM, so I calculated this way:

I assumed number of bits (n) and sampling frequency (64.4 * 2) then I said that the minimum data rate is n*sampling frequency.

Is this right or is the minimum data rate the same as the sampling frequency?

• How did you calculate the sampling frequency? Jan 27, 2022 at 15:13
• You need a minimum of 2x 64.4Hz to represent a 64.4Hz waveform as a squarewave. If you want any kind of detail of that waveform at it's max frequency, the sample rate will need to be much higher, like 10-20x depending on how much resolution you want. Jan 27, 2022 at 15:23

The data rate depends on the data width/resolution (number of bits per sample), whether or not the data is transmitted serially or in parallel, how it is encoded, etc.

The sampling rate depends not just on the maximum pulse rate, but the maximum frequency of interest (harmonic) necessary to recreate the waveform accurately enough to be useful.

Where did the 64.4 Hz number come from?

• (0.4-64.4) is the EEG frequency range and I assumed the number of bits to be something like n=3, so the data rate 3 * 64.4 * 2 am I right? Jan 27, 2022 at 15:43
• Where did you get 3 bits as the amplitude resolution? Jan 27, 2022 at 17:03
• If I did not assume the number of bits,how would i solve this Jan 27, 2022 at 18:05
• What is the amplitude resolution you need? For example, if the positive peak signal is 1.0 V, so you need to know that +/-0.1 V, +/-0.01 V, etc. Another approach is to start with how much harmonic or waveshape distortion you can tolerate. Why are you digitizing the EEG waveform? What is the end application for the digitized data? Jan 27, 2022 at 18:31

From Shannon's theorem you need 128 samples per second to represent the 64Hz between 0.4Hz and 64.4Hz

If you want to make it easier to recover the waveform you'll want to exceed that by some amount.