In the context of wireless instrumentation systems, we proposed the MWAC model to enable reliable communications between sensors. This model, based on a self-organization process, is operational and has been usefully deployed on real world wireless sensors networks. However, for some applications, MWAC presents an important limitation. In fact, for a given emergent structure, the process offers only one path to join the workstation which collects all the data. This paper presents an improvement of this model to increase the path proposal using ant colony optimization.