Dynamic graphs are being used to a large extend in the modeling of lots of dynamically changing and uncertain environments such as complex networks or mobile ad hoc networks. These models are specialized and dedicated to the application they handle. To your knowledge no unified model handles the ideas of dynamic graphs regardless of any applicative background. Moreover, problem solving in these dynamic environments is a key problem. In this work a unified formulation of dynamic graph is proposed. We then study the resolution of problems in these dynamic graphs with the naturally dynamic and distributed methods that are Swam Intelligence approaches. The proposed models are validated in the field of mobile ad hoc networks. We proposed an original approach aimed at constructing and maintaining communication paths under several constraints. Another problem successfully addressed is the construction and the maintaining of a spanning forest in a mobile ad hoc network.