Waveform time-frequency characterization for dynamically configured sensor systems

In this work, a waveform agile sensing method for the target tracking problem has been described. By dynamically selecting the transmitted waveform from a generalized FM chirp signal library with different phase functions and chirp parameters, the tracker can improve the performance dramatically. The waveform design algorithm is based on the myopic optimization of a cost function, which is the predicted mean squared error. The cost function is approximated using the CRLB combined with unscented transformation. Then the generalized FM phase function and the corresponding parameters are configured according to a grid search over all the phase function candidates and allowable parameter values. Due to the nonlinearity of the measurement model in the tracking system, particle filtering is applied to track the state of the target. This approach was demonstrated by tracking a target moving in a 2-D cluttered environment with two active sensors. The simulation results showed that the mean squared error of the tracking was dramatically reduced by adaptively adjusting the transmitted signal.