Bias Estimation in an Association-Free Nonlinear Filter

Abstract Musick, S., and Kastella, K., Bias Estimation in an Association-Free Nonlinear Filter, Digital Signal Processing 12 (2002) 484–493 Previous nonlinear filtering research has shown that by directly estimating the probability density of a target state using a track-before-detect scheme, weak and densely spaced targets can be tracked, and data association (in which reports are associated with tracks) can be avoided. Data association imposes a heavy burden on tracking, both in its design, where complex data management structures are required, and in its execution, which often requires many computer cycles. Therefore, avoiding data association can have advantages. However, a concern exists that data association is essential for estimating and correcting additive sensor biases, which are nearly always present. This paper demonstrates that target tracks and sensor biases can be estimated simultaneously using association-free nonlinear methods. We begin by defining a state consisting of target locations and a slowly drifting sensor bias. Stochastic models for state dynamics and for the measurement function are presented. A track-before-detect nonlinear filter is constructed to estimate the joint density of the state variables. A simulation that emulates estimator behavior is exercised under low signal-to-noise conditions. Simulation results are presented and discussed. This work extends the useful range of nonlinear filtering methods in tracking applications.