[Help-gsl] rayleigh vs complex gaussian

[Volkan KUMBASAR]

dear sir,

i am new -a newbie - in this mailing list , so first i will say hello to
everybody.

my question is simple:
i need to create rayleigh random variables with the variance of 1, to
provide
fading coefficients for my communication system.
fortunately the gsl library provide a function for my aim - the
gsl_ran_rayleigh( r, sigma ).
i use this function in my communication system and get good results.
then i try  to use instead of this function this structure:

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

// rayleigh = abs( gauss + j * gauss )  , gauss ~ N( 0 , 0.5 )

complex<double> alphaTmp( gsl_ran_gaussian( r, sqrt(0.5)  ) ,
gsl_ran_gaussian( r, sqrt(0.5)  )  );
double alpha = ABS( alphaTmp ); // ABS() is a function which calculates the
absolute value.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

if i implement this code in system, i get bad results :(

so for your opinion ; which rayleigh generator structure should i use in my
system ? or
more specifically : can i trust gsl's gsl_ran_rayleigh ?

thank for helping and have a nice day.
--
vk