In this paper, we define a model of an ad hoc routing protocol, i.e. the OLSR (Optimized Link-State Routing) protocol. This model handles novel constraints related to such networks and issues new challenges to treat these constraints. In the network community, the practice is based on simulation models that allow to perform performance measures but no formal methods are used. We propose to promote the use of formal description techniques such as the promela and SDL languages that are both well accepted in the community of communication protocols. Stable communications with a right quality of service need to dispose of a correct underlying routing protocol