Localization Solely based on Ambient Signals

We present two approaches for the localization of passive nodes in a communication network using ambient radio or sound signals. In our settings the communication nodes have unknown positions. They are synchronized but do not emit signals for localization and exchange only the time points when environmental signals are received. The signals occur at unknown positions and times, but can be distinguished. Since no anchors are available, the goal is to determine the relative positions of all communication nodes and the environmental signals. The first approach, Iterative Cone Alignment, deals with arbitrary positions of signals. It solves iteratively a non-linear optimization problem of time differences of arrival (TDOA) by a physical spring-mass simulation. Given a sufficient number of signals the approach converges towards a unique solution. The second approach, Ellipsoid TDOA, demonstrates a closed form solution assuming the signals originate from far distances. The TDOA characterize an ellipse from which the distances and angles between three network nodes can be inferred. Both approaches are tested in numerous simulations and in a real world setting where the relative positions of notebooks are determined utilizing only the sound produced by a girl with a glass bottle and a toy gun.