Outlier Identification Based on Clustering for Joint Positioning Systems

Working in the complex or even hostile underwater circumstances, robustness and stability are of great significance for underwater positioning systems. This paper presents a method to identify outliers in joint underwater positioning based on clustering. Firstly, using the distance and direction of arrival (DOA) measurements of the system, preliminary estimates of the target position are calculated. Secondly, clustering algorithm is employed to analyze the consistency of preliminary estimates. Finally, on the basis that outliers and other measurements are incompatible, outliers can be identified through estimating state variables, and then elimination of the localization deviation caused by the outliers can be achieved. Simulations are conducted to verify effectiveness of the method. Results show that compared with the current distance-based robust localization methods, the proposed method fully utilizes the distance measurements and DOA measurements, and the outlier identification performance is more satisfactory.