An adaptive deconvolution smoothing filter implemented on a multiple transputer network

A multiple transputer system is demonstrated to work in handling adaptive deconvolution smoothing. The cooperation between its four units allow the separate tasks of data I/O, parameter estimation and smoothing filter to proceed in parallel, with the different data items and parameters being passed across the transputer channels connecting the four units. Although the performance of the transputer system was found to be slow compared to the TMS32010 system, the hardware and software flexibility provided by the transputer did allow the easy integration of a supervisor into the system. The use of a supervisor is shown to enhance the robustness of an adaptive signal processor. From the evaluation, the transputer would be well suited to applications within adaptive control. Future research will consider the application of supervisor transputers to adaptive control problems