In this paper, a derivation of the macroscopic mean field theory of th
e cellular automaton (CA) model of highway traffic flow starting from
the microscopic dynamical point of view is presented. Starting from an
equation describing the time evolution of the Boolean state variable
at each site of the basic CA model, and using a two-site approximation
for the multi-site correlation functions, a dynamical mapping between
the macroscopic average speeds v(t + 1) and v(t) at different time ca
n be derived. Mean field results consistent with the simulation data a
re obtained by considering the attractors of the mapping and their cor
responding basins.