Adaptive Processing Filtering and Design of Digital Signals for Realtime Analysis.
BELL AEROSPACE CO BUFFALO N Y
Pagination or Media Count:
The application of adaptive analog filters in the continuous case is investigated in order to derive algorithms for the control of the functional parameters frequency, bandwidth by adaption of the physical parameters gains, multiplying factors. The available transforms from the continuous to the discrete domain are reviewed and the effects of the transforms on the filter parameters are considered in order to minimize the complexity of the resulting digital filters. For the discrete digital filters, the relationships of round-off, truncation and bit accuracy to the filter stability and convergence are investigated. The stability is also examined in relation to the network structure for arrangements of the delays and feedback connections other than the direct canonical form. A discrete processor based upon automata theoretic concepts is reviewed. The formal model is extended for application to noisy environments by use of its inherent flexibility in both the problem definition process and the solution realization phases. Realtime signal processing experiments are reported using the discrete processor for two recognition tasks with audio frequency signals. Several experimental techniques of statistical extension are discussed. Author
- Statistics and Probability