On the design of quantized control using randomization

In this paper we describe an approach to the design of quantized feedback control making use of randomization. The methodology described here provides a basis for designing uniform or non-uniform quantization suitable in situations where fine quantization is not possible. We show that a suitable inhomogeneous finite state Markov chain whose transition rates depend on measurements of the error, can be used to generate low information rate regulators capable of achieving basic control objectives. One attractive aspect of this approach is the possibility of using classic feedback control design methods to determine the properties of the quantizer.

Full Text (PDF format)

Published 18 November 2015