Theory and applications of digital smear filters. (1994)
AuthorsRolls, Andrew Johnshow all
This thesis investigates the theory and applications of digital smear filters. A smear filter can be defined as a device that disperses the energy contained in a wide bandwidth pulse in time. A desmear filter performs the inverse operation to the smear filter by compressing the smeared pulse in time. The first part of this thesis, consisting of chapters 2-4, presents a theoretical basis for digital smear filters and develops three methods for designing these filters. One of these methods is an extension of the window method used to design linear phase FIR filters; another is an extension to the frequency sampling method; the third design method has no linear phase FIR filter counterpart, and we have called it the iterative Wiener method. The second part of this thesis, consisting of chapters 5-8, investigates novel applications for these smear/desmear filters. These applications include using smear/desmear filters to: • Compress the dynamic range of speech. (A 3-4 dB reduction in the peak-to-rms ratio can be achieved.) • Destroy the dependence between a signal and its quantization noise. (Smear filters can increase both the subjective quality and the intelligibility of hard limited speech.) • Encode data for transmission over a noisy channel in order to protect the data from bit errors. (A novel method for channel coding that uses smear filters is described, and the principles of this method are substantiated.) • Scramble a signal for privacy. (Smeared speech signals can be made unintelligible for delays as short as 400 ms.) Although smear/desmear filters are shown to be extremely versatile devices, they have one serious drawback: They introduce delay. Some of the applications described above require very long delays to approach their theoretical bounds of performance.