This paper faces the problem of configuration and communication in a distributed radio sensor network, composed of identical sensors randomly placed in a two- or three-dimensional space. The reference is provided by objects with known positions called masters. Two architectures are shown; the first uses one master, the second three masters. The one master architecture makes it possible to identify and locate all the sensors in space and to calculate for each of them the lowest energy transmission path to reach the master. The three-master architecture locates, by triangulation, each sensor when a transmission of information occurs and cannot optimize energy consumption during sensor communication. On the other hand, it is also able to localize moving sensors or to handle dynamically changing sensor topologies. The results show that the three-master architecture is faster, but it implies an energy waste of about 30 times greater than the one-master architecture for a constellation of 50 sensors.