Dynamicity Aware Graph Relabeling Systems and the Constraint Based Synchronization: a Unifying Approach to Deal With Dynamic Networks Arnaud Casteigts and Serge Chaumette LaBRI, Universit´ e Bordeaux 1, 351 Cours de la Lib´ eration, F-33405 Talence, France. {arnaud.casteigts,serge.chaumette}@labri.fr http://www.labri.fr/ Abstract. Many research projects are being done in the domain of wireless mobile and/or ad-hoc networks to provide tools, algorithms and applications that make it possible to handle or use their dynamic characteristics. The purpose of most of these projects is to solve a specific problem, in a specific context. Most often, the models and formalisms that are used are also specific and the results are therefore difficult both to understand and to use in another context. We believe that what is needed is a general model offering a very high level of abstraction, in order to define and characterize what is feasable or not feasable in a dynamic network, depending upon some of its characteristics. In this paper we define such a model and its associated formalism. They are adapted to the study of dynamic networks and to the modeling of algorithms in a dynamic context at a high level of abstraction. The proposed model (Dynamicity Aware Graph Relabeling Systems) derives from what has been achieved in the area of local computations, and that produced useful results in the context of static networks. Our contribution comprises a model and the associated formalism, plus an original synchronization mode between nodes that allows to seemlessly adapt an algorithm to different mobility contexts. All the concepts are illustrated and discussed through the example of a document propagation algorithm, with a resume feature. The work presented in this paper is carried out at LaBRI (Laboratoire Bordelais de Recherche en Informatique) and more precisely in the SOD (Distributed Systems and Objects) team. It is partly achieved within the framework of the Sarah (Asynchronous services for mobile ad-hoc networks) project supported by the ANR (Agence Nationale de la Recherche). 1 Introduction The research activity in the area of wireless and pervasive networks keeps on growing. The emer- gence of numerous broadband wireless standards such as 802.11/16, bluetooth or 3G/UMTS that offer many new possibilities enforces the trend. While many research projects focus on routing layers [9], or on a specific problem in a given network, we study these different kinds of networks from a theoretical point of view in terms of what basically makes them similar or different, and what can be done (or not) in each of them. This requires a well adapted model, which makes it possible to take into account the dynamicity of the network, and the locality of the communications. To understand this last point, let us examine more precisely one single device in such a network. The only thing this device can basically be aware of is its own state, and the state of its direct neighbors. The only thing it can be sure of is that, at a given instant, it can communicate with these neighbors, and that this communication link can have disappeared at the next instant, because one of the devices has moved or has been turned off. Another property of the desired model (in addition to locality and dynamicity) is to be at a high level of abstraction. It should be possible to ignore physical implementation details when designing an algorithm. Our first contribution is the DA-GRS model (Dynamicity Aware Graph Relabeling Systems). It has been inspired by the area of local computations, in which most of the desired properties