An Improved Extended Kalman Filter for Localization of a Mobile Node with NLOS Anchors

Tracking a mobile node using a wireless sensor network under non-line of sight (NLOS) conditions, has been considered in this work, which is of interest to indoor positioning applications. A hybrid of time difference of arrival (TDOA) and angle of arrival (AOA) measurements, suitable for tracking asynchronous targets, is exploited. The NLOS biases of the TDOA measurements and the position and velocity of the target are included in the state vector. To track the latter, we use a modified form of the extended Kalman filter (EKF) with bound constraints on the NLOS biases, as derived from geometrical considerations. Through simulations, we show that our technique can outperform the EKF and the memoryless constrained optimization techniques. Keywords—Extended Kalman filter; localization; non-line of sight; ultra wideband.