# matched filter position

I have seen that matched filter should be used in TED (gardner loop) after resampler

resampler --------->   matched filter  ---------> down sample ---------> loop filter -> interpolation control for resampler
(4 samples/symbol)            (4 samples/symbol)     (2 samples/symbol)


is this right?

Can matched filter be used with carrier recovery through costas loop?

• This post may help you. The Gardner uses two samples per symbol and has better performance BEFORE the matched filter unless you use prefilter techniques prior to timing recovery to minimize zero crossing jitter. dsp.stackexchange.com/questions/31517/… I would approach it as resample to 2 samples per symbol, use Gardner to determine timing error on the complex samples prior to the matched filter and then use matched filter prior to symbol decision to optimize SNR for best symbol decision. What is good for symbol decision is often not good for timing Dec 18 '19 at 2:20