Feedback Control for Computer Systems

Chapter 5. Identifying Input and Output Signals

Initially, it can be difficult to see how feedback methods can be applied to situations other than the “classical” application areas treated in textbooks on control theory. The way control engineering decomposes real-world systems into abstractions often does not easily align with the way those systems appear to others. In this chapter, I want to step through a handful of examples and show how they could be approached from a control-theoretic point of view.

Control Input and Output

The essential abstraction in any control problem is the plant or process: the system that is to be controlled. From a controls perspective, a plant or process is a black box that transforms an input to an output. It is usually not difficult to recognize the plant itself, but identifying what to use as control input and output can be challenging.

It is essential to realize that the terms “input” and “output” here are used only in relation to the control problem and may be quite different from the functional inputs and outputs of the controlled system. Specifically:

• The input or control input is a quantity that we can adjust directly. By adjusting the input, we hope to influence the output in a favorable way.

• The output or process output is the quantity we want to control: we want the output to track the reference value (the setpoint).

These two observations should help to identify the quantities to use as control signals, either as input or as output. Just ask ...