Decimation:

This library implements a polyphase FIR decimation filter wikipedia. The decimation class is used to decimate the input signal by a factor of two.

Low-pass filter:

This matlab command is used to generate the FIR filter:

b = fir1(47,0.43);

the impulse response of the filter in time and frequency domains:

Tests:

Simulated signal with two components: 1- cosine at 1700 Hz 2- cosine at 20 Hz

and sampling rate at 25000 Hz:

In four test cases the output of the code is compared with Matlab's decimate function:

1. 20 Hz cosine with amplitude of 100 + 12000 Hz cosine with amplitude of 10

2. 20 Hz cosine with amplitude of 100 + 6250 Hz cosine with amplitude of 10

3. 20 Hz cosine with amplitude of 100 + 3000 Hz cosine with amplitude of 10

4. 20 Hz cosine with amplitude of 100 + 1000 Hz cosine with amplitude of 10

Matlab code to plot the above figures:

>> signal = csvread('signal4.csv');
>> decsignal = csvread('decimated_signal4.csv');
>> d = moddecimate(signal(2:end,2),2,47,'fir');
>> plot(t, decsignal(:,2), '-.', t, d(12:end-12), ':r', t, decsignal(:,2)-d(12:end-12),'k', 'linewidth', 2);
>> legend('avxDecimate', 'matlab', 'avxDecimate-matlab');
>> xlabel('time (seconds)');
>> ylabel('amplitude');
>> title('20 Hz cosine + 1000 Hz cosine')

moddecimate is a copy of the decimate function with only one change:

original:

line 93: b = fir1(nfilt,1/r);

changed to:

line 93: b = fir1(nfilt,0.43);

to have the desired impulse response (see the second figure from the top).

Needed libraries:

1. Boost::Circular_buffer

Compile:

cmake based