On suboptimal detection of 3-dimensional moving targets

The author designates matched filters that are completely characterized by the velocity of the target as assumed velocity filters (AVFs). Like most matched filtering techniques where the signal parameters range in a continuum, the AVF must be implemented suboptimally by partitioning the velocity space. The author investigates the possibility of using a signal-to-noise ratio (SNR) loss factor as the criterion for the partition. The loss factor is a measurement of the loss of SNR at the output of the matched filter due to mismatch of filter parameters. In the scenario of detecting a moving satellite from a ground-based sensor, because of the vast sky the sensor has to search, it is important to keep the number of filters minimal. The author shows that, with a fixed loss factor, the number of filters required for coverage increases linearly as the span of the two-dimensional velocity space increases quadratically. The rate of increase is further reduced when the loss factor is made proportional to expected target angular speed.<
>