GMTI radar track segment association and out-of-sequence measurement processing

Date of Completion

January 2011


Engineering, Electronics and Electrical




In this dissertation two problems are investigated in the area of target tracking: GMTI radar track segment association and out-of-sequence measurement (OOSM) processing. The latter consists of four subproblems: tracking with multisensor OOSMs with residual biases. OOSM processing for particle filters, update with multiple OOSMs with arbitrary arriving order and optimal removal of OOSMs from tracks. ^ In a real tracking system, track breakages can occur due to highly maneuvering targets, low detection probability, or clutter. A track segment association (TSA) approach was developed for an airborne early warning (AEW) system to improve track continuity by "stitching" broken track segments pertaining to the same target. However, this technique cannot provide satisfactory association performance in tracking with a GMTI radar ground moving targets employing evasive move-stop-move maneuvers. To avoid detection by a GMTI radar, targets can deliberately stop for some time before moving again. Since a GMTI radar does not detect a target when the radial velocity (along the line-of-sight from the sensor) falls below a certain minimum detectable velocity (MDV), the move-stop-move maneuvers of the targets usually lead to broken tracks as a result. A new TSA technique is proposed which can effectively stitch both "regular" broken tracks and broken tracks due to targets' move-stop-move maneuvers.^ In multisensor target tracking systems measurements from different sensors on the same target exhibit, typically, biases. Furthermore, measurements from the same target can arrive out of sequence. A novel approach is presented for the combined problem of handling biases from multiple sensors when their measurements arrive out of sequence. The major benefit of this new approach is the significant improvement of filter consistency.^ The OOSM problem is considered in a situation where the filtering technique used at the tracker is the particle filter (PF). First, an exact Bayesian algorithm for updating with OOSMs is derived. Then, the PF implementation of the exact Bayesian algorithm, called A-PF, is developed. Since A-PF is rooted in exact Bayesian inference, if the number of particles is sufficiently large, A-PF is the one (and the only one) that is able to achieve the optimal performance obtained from the in-sequence processing. Also, it is shown that the performance of A-PF is always superior to previous (heuristic) PF-based algorithms with the same number of particles.^ Numerous algorithms have been proposed in literature for state update with an OOSM optimally or suboptimally. However, what one faces in the real world is the multiple OOSMs, which arrive at the fusion center in arbitrary order, e.g., in succession or interleaved with in-sequence measurements. A straightforward approach to deal with this multiOOSM problem is by sequentially applying a given OOSM algorithm; however, this simple solution does not guarantee the optimal update under the multiOOSM scenario. The difference between the single-OOSM processing and the multiOOSM processing is discussed. A general solution to the multiOOSM problem, which is called the complete in-sequence information (CISI) approach, is proposed. It is shown that the CISI approach guarantees the optimal update with multiple OOSMs with arbitrary arriving order.^ In real-world tracking systems some earlier measurements that have been used to update a track might be reassigned to other tracks and then there is a need to remove such measurements from the track under consideration. Since in most real-world systems the past measurements are not stored, refiltering is usually not an option for removing an earlier measurement. An optimal removal algorithm is developed which only requires the storage of the state estimates and covariances. The proposed algorithm yields significantly better results than the previous (suboptimal) "one-step" solution, especially when the measurement to be removed is (nearly) an outlier. ^