Dynamic Communication Systems (DCS) are characterised by a dynamically
changing and potentially unbounded number of processes operating in a
varying communication topology. We present abstraction refinement
methods for Data-Type Reduction, a special form of finitary abstraction,
of DCS. The refinement substantially increases the precision of the
abstraction and allows to automatically solve a rich class of
first-order quantified temporal properties. The refinement itself is
based on examination of communication sequences of a given DCS.