A bearings-only-tracking framework based on the EKF and UKF combined algorithm

Abstract

Bearings-only tracking (BOT) is with the common interest in array signal processing society and is actually a nonlinear estimation process which acts as a posterior one of beamforming. Extended Kalman filter (EKF) which is commonly used in BOT inherits Kalman Filter (KF)'s advantage in having good computational efficiency, but often leads to unstable estimations. Particle filtering (PF) and Unscented Kalman filtering (UKF) are recently suggested for stability improvements, and UKF is suggested more suitable for real-time applications than PF. This paper delicately compares both of the computing burdens and performances of EKF and UKF in a two-dimensional BOT scenario, and finally proposes a new time-efficiency framework that combines EKF and UKF together to fulfill the estimation process. Simulation results and theoretical analyses are included for presenting the new framework.