Efficient Ambiguity Resolution in Wireless Localization Systems
Abstract: Wireless communication is widely used for node localization in various automation domains. Schemes based on time difference of arrival (TDOA) are a good compromise of achievable accuracy and implementation effort. Due to restrictions on possible reference sensor positions in industrial environments, the geometric conditions between reference and mobile nodes locations are often difficult. Consequently, iterative algorithms can fail or result in high computational effort. Exact direct methods introduce no ad additional ditional errors, but can result in a valid pair of solutions. Existing direct methods thus either rely on a priori information to sort out one of those two solutions, or revert to nonexact linearization approaches of the underlying hyperbolic equations. In this work, we present a direct approach that applies exact direct methods, and resolves the ambiguous pairs of solutions without a priori information. Its divide and conquer structure and the high computational efficiency of the available exact direct methods hods makes it a very good candidate for fast parallel computing in distributed sensor networks. The performance is compared to the Cramer Cramer-Rao Rao lower bound, to the iterative Gauss-Newton Newton method and to another direct method. The new method is presented for th the e TDOA scheme, but can also be applied to time of arrival, or any other range-based based scheme.