Few, or no RF switches are going to let you commutate
at 2GHz. They try to isolate RF from the control ports.
But they will pass / block 2GHz (50 ohm system).
An old-school analog switch, forget it.
There are only a few topologies - simple, T, Pi. I think
the simple switch (pass only) may suffice since the
input signals are probably not moving very fast (T is
for HF isolation w/o leakage, Pi for isolation and low
port impedance).
You will want Ron*Cmsb to be << 1/freq or that will
become your settling-time bind for sample rate. You
will want to design for minimum charge injection to
further tamp down settling time.
Beyond that there is nothing special about 2GHz on the
control pins, you either have the intrinsic speed in your
process or you don't. Might just see how fast one inverter
can drive another and still have full rail-rail swing. If that
isn't 2GHz, then you're probably hosed.