Characterization of waveform performance in clutter for dynamically configured sensor systems

摘要

In this paper, we consider the problem of dynamic selection of frequency modulated waveforms and their parameters for use in agile sensing. The waveform selection is driven by a tracker that uses probabilistic data association to track a single target in clutter, employing measurements derived from a nonlinear observations model. We present an algorithm that performs the selection so as to minimize the predicted mean square error which is computed using the unscented transform. We compare and analyze the performance of several trapezoidal envelope frequency modulated waveforms with different time-frequency structures using the Cramer-Rao lower bound. The simulation results indicate that the dynamic selection of waveforms and their corresponding parameters improves the tracking performance.