Global and local control of processing using knowledge-based signal processing techniques-an application

Knowledge-based signal processing is a concept that emphasizes the need for selection and tuning of digital signal processing procedures by use of information specific to the domain of the application. The author describes how a long-range goal can be dynamically reduced to a series of subgoals, and how the methods used to reach the various subgoals can be controlled and tuned. The propositions put forth originate from experiences gained from a specific signal processing problem, namely that of decomposing a set of superimposed pulse trains into the constituent individual trains. The aim of the project described was the automatic, high-accuracy decomposition of myoelectric signals