1 Generic Sensor Model
For proposals and early design phases the sensor is best modeled with a simple linear approach. In later design phases where the nonlinear aspects of the sensor response cna be modeled. This is frequently unnecessary unless the limits of your sensors and controllers are being pushed.
The simple model starts with a transfer function that represents the frequency response of the sensor. In addition to the frequency response of the sensor noise is modeled (simple band-limited white noise or PSD derived time history noise). Sensors are usually designed (and electronics or software added) to provide a linear response.
1.1 Sensor Noise
Most vendors provide noise measured on the sensor's output. (Be careful some vendors provide noise on the sensor input.) Noise will limit system performance. Noise passes straight through your control system to your output.
1.2 Nonlinear Models
Nonlinear models of sensors typically include Quantization effects and hard nonlinearities from software. Software thresholds can cause abrupt step changes to the output or to a control signal; these are what I am calling hard nonlinearities. Obviously these hard nonlinear steps can cause abrupt transient errors in the system. The abrupt steps will cause transients which need to settle before the system perofrmance will return to the expected behavior.
Nonlinear models are typically much more difficult to model. Validation is even more difficult and time consuming. Save these for later development stages where lots of test data can be taken and you have weeks or months to really dive into the data.