An example of a fair comparison between controller structures

This paper describes a way to quantify the potential advantage of adding a simple undelayed feedback path to a sampled-data control system, thereby bypassing delays associated with sampling, holding, and computation. The output regulation problem is studied. The achievable performance limits with different controller structures are used to make comparisons between the structures (rather than individual controllers). The controller structures are non-standard; they have multiple feedback paths. One feedback path uses sampled-data control with an associated delay, while the other path is undelayed. The regulated plants are modeled as stochastically perturbed continuous-time linear systems. Performance measures are the RMS control signal magnitude and the RMS regulation error - as in the linear quadratic Gaussian (LQG) problem framework. An example of this technique is presented.