Parallel algorithms for template and bound generation in robust control

The quantitative feedback theory (QFT) approach to robust controller and prefilter synthesis for uncertain systems is of great practical interest. In this article, we propose parallel computation methods for the two key QFT steps of plant template and bound generation. We explore parallel structures in template and bound generation procedures and develop parallel techniques that are superior alternatives to serial techniques. The parallel techniques are implemented using programming language CUDA on NVIDIA GPU. In an eight parameter aircraft control example problem, the parallel methods are found to be much faster than the existing serial methods by up to 4.5 times in the template generation step and up to 38 times in the QFT bound generation step.