Adaptive Tracking System Study.

This is the Final report on phase three of the Adaptive Tracking System Study, which considers the application of adaptive filtering to passive sonar bearing estimation. The study was initiated because the adaptive filter structure seemed well suited to bearing estimation problems involving unknown input statistics and dynamic targets. In particular, a an adaptive filter does not require a priori power spectral information on signal and noise fields, b because it is iterative, it can track non-stationary inputs while preserving its underlying minimum mean square optimality criterion, and c since all the correlation information between adaptive filter inputs is contained in the filter weights, the potential exists to perform both broadband and narrowband tracking simultaneously. In summary, there are some genuine advantages in performance and perhaps even in implementation offered by the adaptive filter tracker over conventional trackers. The performance benefits come when signal or noise statistics are not known to the designer of the front-end filters on a conventional system. Implementation benefits can occur when sampling rates at the beamformer outputs are much slower than the time resolution required to achieve a specified angular estimation accuracy. The adaptive tracker interpolates between tap values rather than having to sample at a higher rate.