Take a look at some of the free vendor software
available. I've used Agilent AppCAD and Cascade32
from Spectrum Microwave, both of which I believe
are linked through RF Cafe. I've also used an Excel
spreadsheet passed down through co-workers with
changes I've made. It's created from the basic cascaded
noise figure and intermod equations. Also, I believe
Kirt has a spreadsheet for a small fee. Doing your
own spread sheet would be very instructive. Use
one of the other programs mentioned as a double
check.
Keep in mind that a mixer will add
out of band noise unless that is filtered out of
the applied signal.
Filters can be handled
by allowing for a bandwidth term in the spreadsheet.
There will be some manual tweaking of the equations
but it will assist you in seeing the effects of
a 5 MHz BW filter and a 20 MHz BW filter.
None of the mentioned software does BER, however,
that is dependent upon the SNR and SFDR you have,
so you should be able to get close.
Also,
be aware that the simulations don't typically take
into account component to component variations and
also rely on ACCURATE data at all frequencies involved.
This means a lot of bench testing of simple amplifiers.
(You haven't lived until you've taken intermod data
on a mixer over temperature and LO drive for ten
units and then find there is no pattern to the distortion
products or losses that can easily be quantified
or described in any equation.)
Temperature
drift and stage to stage mis-match need to be taken
into account also.
To sum up, your best bet
is to get one of these software packages and start
using them with component manufacturer's data, building
small functional blocks, always comparing simulation
with measured data.
GOOD LUCK!!