We present a novel approach to the approximate modelling of symmetric
systems where users' behaviours are highly dependent. The model follow
s the behaviour of a tagged user in an exact manner, and the rest of t
he users' behaviour is approximated by mimicking that of the tagged us
er at steady state using an iterative approach. The benefit of such an
approach is demonstrated in the modelling of collision-avoidance star
local area networks (CASLAN)-a multiple access local communication ne
twork based on an active star topology. We show that the proposed appr
oach is capable of effectively modelling the behaviour of CASLANs as i
t captures the nodal dependency as well as network-parameter dependenc
y. Results show that the model is very accurate compared to simulation
results, and is more accurate than all previous CASLAN performance mo
dels.