Real-Time Recognition of Manual Morse Telegraphy Using Nonlinear Estimation and Viterbi Decoding.

摘要

This report describes the analysis and design of a sophisticated real-time detector and decoder for manually-sent Morse code signals transmitted via high-frequency radio circuits. The system software is designed to run if required in a distributed computer network such as the DCN and can provide simultaneous multicircuit decoding capabilities, using a multiplexed analog-digital converter and multiple terminals, in a single machine. The system uses the principles of optimum filtering and receiver design, including matched filtering and Viterbi decoding. The statistical structure of the Morse process is encapsulated in a Markov model which includes a-priori probabilities and transition probabilities reflecting the entire alphabetical structure of the code. A key component in the design is a new algorithm, called ratio-weighted estimation (RWE) which provides extremely fast adaptation to the time-varying Morse code parameters typical of multioperator circuits.