Localizability with Range-Difference Measurements
The physical position is crucial in location-aware services or protocols based on geographic information, where localization is performed given a set of sensor measurements for acquiring the position of an object with respect to a certain coordinate system. In this paper, we revisit the longstanding localization methods for locating a radiating source from range-difference measurements, or equivalently, time difference-of-arrival measurements from the perspective of least squares (LS). In particular, we focus on the spherical LS error model, where the error function is defined as the difference between the true distance from a signal receiver (sensor) to the source and its measured value, and the resulting spherical LS estimation problem. This problem has been known to be challenging due to the non-convex nature of the hyperbolic measurement model. First of all, we prove that the existence of least-square solutions is universal and that solutions are bounded under some assumption on the Jacobian matrix of the measurement model. Then a necessary and sufficient condition is presented for the solution characterization based on the method of Lagrange multipliers. Next, we derive a characterization for the uniqueness of the solutions incorporating a second-order optimality condition. The solution structures for some special cases are also established, contributing to insights on the effects of the Lagrangian multipliers on global solutions. These findings establish a comprehensive understanding of the localizability with range-difference measurements, which are also illustrated with numerical examples.
READ FULL TEXT