Target identification based on state space analysis

A signal reflected from a radar target is modified according to the impulse response function of the target. This interaction may be thought of as a linear filter acting on the incident signal. Filters are not exactly invertible, which means that there is no continuous function between the filter output and its input. Likewise, there is no invertible function between the outputs of 2 different filters driven by the same signal, so there is no function between the responses of 2 different targets illuminated by the same incident signal. We apply a statistic from nonlinear dynamics that describes the probability that there is a function between 2 signals embedded in state space to develop a similarity statistic between 2 signals reflected from different targets. The similarity statistic describes how similar the targets are to each other. We have tested this similarity statistic with numerical simulations and acoustic experiments.