Using Optimal Pulse Radio-Signal Receivers for Speech Filtering

Nonlinear filtering theory is viewed in the Gaussian approximation to synthesize optimal pulse radio-signal receivers for speech filtering with various forms of modulation. A Markov-process speech model is considered; it is described by a system of two linear first-order stochastic equations. The signal (a train of radio pulses) is considered against an additive white-noise background. The mean-square filtering error is determined for amplitude, pulse-width, and pulse-time modulation.