We propose a graph-based process calculus for modeling and reasoning about wireless networks with local broadcasts. Graphs are used at syntactical level to describe the topological structures of networks. This calculus is equipped with a reduction semantics and a labelled transition semantics. The former is used to define weak barbed congruence. The latter is used to define a parameterized weak bisimulation emphasizing locations and local broadcasts. We prove that weak bisimilarity implies weak barbed congruence. The potential applications are illustrated by some examples and two case studies. τ − → M ′ (unobservable transition). As the first theoretical result, we prove that the two semantics describe the same behaviours.Thirdly, two kinds of behavioural equivalences for GCWN are developed. We first adopt the concept of barb [10] to define a weak barbed congruence without location information. Barbed congruence is natural to describe that two networks are identical if they exhibit the same barbs during their reductions in any context. However, barbed congruence is hard to handle directly, because one has to consider all possible contexts