Fast Interpolation of Complex Baseband Signals

This contribution contains 3 custom and easy to use interpolation routines that outperform the matlab functions resample.m and interp.m.

I provide 3 new interpolation (by a K factor) routines:

1. interp11.m is fast-convolution based and basically convolves the input signal with a sinc window.
2. interp21.m is based on zero-padding in the frequency domain.
3. interp31.m is a more elaborate zero-padding technique originating from a SAR Signal Processing book by professor M. Soumekh.

The following assumptions were made:

1. All input signals should be of even length, preferrably a power of 2.
2. All input signals should be in baseband.
3. The first sample x(1) corresponds to x[0].

It has been proven that for the provided chirp signal, all three routines run faster than matlab-provided routines resample.m and interp.m

Also, it has been proven that interp11.m is also the most accurate of all.

Run the interpdemo5.m to see the results.

Note: The interpolation is not performed on a given grid but is based on a factor K. This means that if the length of input signal x is Lx, the length of the output signal y will be Ly=K*Lx.