Indoor positioning appears to be a real challenging topic and many techniques have been proposed these last ten years, ranging from High-Sensitivity GNSS to sensor networks, through Assisted-GNSS or WLAN based approaches. No ultimate solution seems yet to provide the answer to the problem, probably crucial to Location Based Services, of continuity of the availability of the positioning from outdoors (where GNSS are such good candidates) to indoors. Current works are mainly oriented towards hybridisation techniques. Many interesting results are provided with accuracies down to a metre sometimes. But this performance is always linked to associated constraints such as specific deployments, calibration phases, non-mature technologies, etc. We propose another approach based on the use of all available wireless networks and all available connections between mobile terminals. Our approach is based on two fundamental steps: the study of the so-called "visibility" and the estimation of geographical links. The visibility estimates which equipments in the environment have the ability to be radio connected to the terminal and a three-dimensional graph is build according to relative distance estimates between terminals. This graph enables the visualisation of the geographic positions of each device. Various models of visibility have been implemented and evaluated through comparisons of positioning in several elementary scenarios. MatLab simulation results are presented to validate this approach. The positioning accuracy ranges typically from 2 to 10 metres, depending on the environments, density of terminals, percentage of known location terminals and dynamic scenarios. One should keep in mind that these accuracies, even if they are not among the best published, are achieved with no additional infrastructure and is potentially implementable on all today communication devices. Furthermore, this is bound to work outdoors as well as indoors.