Multiresolutional Maneuvering Target Tracking with Average Interpolation and Parallel Implementation

The difficulty for tracking maneuver target is to decide whether a maneuver occurs and when the target begins to maneuver. Since target tracking is a real time problem, the crucial factor is to detect the target maneuver as accurately and quickly as possible. A new approach for the maneuvering target tracking is presented, especially for the target maneuver detection in heavy noise. The algorithm is based on the criterion where the maneuver detection delay is made minimum, which replaces the original measurements by those of the average interpolation to reduce the delay to avoid tracker divergence, and uses wavelet transforms to increase the probability of detection. However, it requires much more computation time, which may be critical in a real-time application. With parallel processing techniques, a parallel implementation of the algorithm is also proposed by multibus computer system with three CPUs. Simulation results show that our scheme has the superior performance than conventional algorithms especially for tracking maneuvering targets in heavy noise, while approaching the same computation time as those. The algorithm has great potential for real-time applications