Increasing the Speed of Control Algorithms

This paper discusses two complementary approaches for increasing computational speed of control and signal processing algorithms. The first approach is concerned with the development of algorithmically specialized computational structures. As illustrative examples we describe structures to implement Riccati equations and polynomic transformations. The second approach achieves the speed increase by constraining the algorithm to a computationally efficient structure. The approach recognizes that even with state of the art technology it may become necessary to find an optimum compromise in the speed/performance characteristics. The technique, termed array constrained design, is illustrated with a signal reconstruction problem.