Abstract: Registration is a prerequisite process for the data fusion in radar networking system to estimate and correct systematic errors accurately. Some classical registration algorithms are all based on the stereographic projection, which introduce errors to the registration of the long distance sensors. In this work, a generalized least squares registration algorithm (ECEF-GLS) is proposed based on the Earth-centered Earth-fixed (ECEF) coordinates. This new approach solves the problem of registration between the long distance sensors, and the covariance of its estimation achieves the Cramer-Rao bound (CRLB) ignoring the errors of the linear model. Simulated data are used to evaluate the performance of the proposed method, and comparisons are made to the generalized least squares (ECEF-GLS) registration algorithm presented by Zhou.

Key words: radar networking, registration, fusion, generalized least squares estimation