Robust State Estimation for Time-Difference-of-Arrival Localization Systems Under Measurement Loss

In this paper, we investigate the problem of robust state estimation in time-difference-of-arrival (TDOA) localization systems under scenarios involving random measurement loss. Traditional TDOA localization methods typically rely on the extended Kalman filter (EKF), which performs local linearization of the nonlinear system model using a first-order Taylor expansion. However, the EKF relies on a first-order linearization of the system model, which inevitably introduces approximation errors and compromises the accuracy of state estimation. In addition, when the fusion center is subject to Denial-of-Service (DoS) attacks, communication interruptions, or data tampering, measurement data may be lost or corrupted, further compromising the stability and reliability of the state estimation process. To address these challenges, we augment the EKF cost function with a penalty term on the innovation sensitivity to modeling errors, thereby proposing a robust state estimation algorithm with enhanced resilience to disturbances and system uncertainties. Under certain assumptions, it is theoretically established that the proposed robust estimator guarantees bounded estimation error. Numerical simulations indicate that, in three-dimensional TDOA-based localization scenarios, the proposed algorithm achieves improved estimation accuracy compared to the standard EKF.