Positioning systems have undeniably progressed. Currently, in an outdoor environment, the accuracy reaches a few centimetres under certain conditions: open space, clear sky, specific measurement techniques, etc. Nevertheless, the problem of positioning in an indoor environment remains persistent: multipath, attenuation, etc. Different systems for indoor positioning have been developed, using technologies such as UWB, WiFi or Infrared. These systems provide interesting results that could allow to reach one meter accuracy. But, this accuracy is related to many criteria: infrastructure, technology, calibration, technical computing, etc. To reduce these constraints, we propose a new approach for positioning. Our approach utilizes all the network equipments present in an environment. The approach is based on two fundamental steps: the study of visibility and the development of geographical links. The study of visibility estimates the visible equipments in the environment. We have studied several models of visibility and we carried out a comparison of the results. A three-dimensional graph is build using the study of geographical links between equipments. This graph allows us to visualize the distribution of equipments and to estimate the geographic positions of each device. To implement our approach, we developed a simulator in Matlab. The simulator first studies the visible equipments for the unknown device. Then, it estimates the distances between the device and the visible equipments. Finally, it constructs a graph and calculates the geographical positions. Simulation results are presented to validate our approach. Our approach is a positioning system capable of operating without additional infrastructure in an indoor environment