Hi Nasi,
when I checked the fading models supplied in GNU Radio 3.7 the other week they seemed to be generating nasty glitches in their output. You could instead take a look at 'my' custom block for doing simple single path configurable doppler Rayleigh fading. The output of this has been statistically validated against the theoretical Rayleigh distribution. I can provide further details of the validations that have been done if required. There is a version of the block for both GNURadio 3.6 and 3.7 API's. This block works fine when instantiated singly, however if you try and build a multipath fading model out of it by instantiating more than one of these blocks (with delays) at the python level then the whole thing tends to seg fault, for as yet undetermined reasons.
You are welcome to give it a try however.