Narrowband process quantization

Summary form only given, as follows. We introduce a technique of tracking and quantizing the complex envelope of a narrowband process, which is a continuous time Gauss-Markov process of arbitrary order. The narrowband process is modeled in a state space framework and several quantization schemes using the in-phase and the quadrature components of the narrowband process are considered. These in-phase and quadrature components are baseband processes, and thus are much more slowly varying than the original narrowband process. The advantage of such a scheme lies in the fact that the sampling in such a system is less sensitive to timing errors. We compare the performance of these schemes with several previously analyzed schemes both with respect to a time averaged smoothed error, and their robustness with respect to the changes in the input spectrum. We compare the performance of the envelope quantization schemes with differential state quantization for a second order narrowband process. The envelope quantization schemes perform better than state vector quantization but worse than differential state quantization. Differential state quantization considers the quantization of a state consisting of the proces and its derivatives and was shown to be a superior quantization scheme for Gauss-Markov processes. We also define a measure of robustness for different source coding systems when the input spectrum changes and compare the performance of this envelope quantization scheme with several other previously analyzed schemes