An Algorithm Design Environment for Signal Processing

Classically, the computer has been used in signal processing simply for numerical calculations, leaving the majority of the design process has been unsupported: in particular, the analysis of the properties of the selected algorithm and the rearrangement of the algorithm to find efficient, input/output equivalent implementations. This paper explores some of the issues involved in supporting the symbolic rearrangement of algorithms. The paper demonstrates the potential of automatic rearrangement through the design of an incoherent detector for sonar FSK-code signals. An innovative implementation, discovered by the computer environment, for the frequency-chip matched filters is presented: this implementation reduces the computational requirements from order N log N to order N .