Channel modelling

Hi,

I'm trying to model a multipath radio channel, I've been doing a bit of
reading and I've got some idea of what I want to do but I'm not really
sure how to go about implementing things.

For example, I've read that I can model a rayleigh fading channel as
two gaussian noise sources, X and Y, and then if I just sum them as
phasors, Z = sqrt(X.^2 + Y.^2), then Z will be a rayleigh distributed
random variable.

But I don't "get" how I implement this - all my books just assume this
is known. I guess that my ultimate aim is to do y(t) = h(t) * x(t) +
n(t), so I want to find the impulse response of my channel. But how do
I go from the equation for Z above to the impulse response h(t)? I
tried setting X and Y to be vectors of gaussian distributed random
variables (ie. X = randn(1, 200); and the same for Y) then summing as
above to give the rayleigh distributed vector Z. But how do I get from
Z to h(t)?

Perhaps I can just use Z as my h(t) coefficients? I tried this and
convolved Z with my complex signal samples x(t) but the result looked
pretty awful, I doubt any receiver could recover this (hence I suspect
that convolving Z with h(t) might not be the correct thing to do). Or
am I wrong here?

If anyone can give me any nudges in the right direction it would be
much appreciated!

Ted.

.