On optimal sensor placement for time-difference-of-arrival localization utilizing uncertainty minimization

In this paper we investigate optimal sensor placement for time-difference-of-arrival localization. The minimization of estimation uncertainty is used as the optimization criterion, which is equivalent to maximization of the determinant of the Fisher information matrix. It is shown that for equal sensor noise variances equiangular sensor separation is an optimal localization geometry irrespective of the emitter range from sensors. For six or more sensors the optimal geometry is not unique. If the sensor noise variances are different, equiangular sensor separation does not always correspond to an optimal configuration. These results are illustrated with numerical examples.