Wireless Sensor Networks have been applied to many application fields, such as building automation, assisted living, health care, positioning and tracking. One of the most active research area is the relative localisation, that is localising the nodes of the network with respect to each other but without a fixed, external and global reference. A common drawback of any relative localisation algorithm is that the solution can be only be found up to an isometric transformation. In this article, we propose a solution based on multidimensional scaling that identifies and adopts the minimum set of information to remove the geometric ambiguity and, unique in this respect, uses only ranging measurements. The characteristics of the algorithm are investigated via Monte Carlo simulation, validating the proposed system with an experimental test. In addition we propose a compact device prototype, based on ultrawide-band technology, used for the experiments and to practically show the technical viability of the solution.