1. The document discusses random signal models, which represent random signals using parameters from probability distributions rather than storing the entire signals. This allows generation, classification, and compression of random signals.
2. Common random signal models include the moving average (MA), autoregressive (AR), and autoregressive moving average (ARMA) models. The maximum likelihood and mean square error methods are presented for determining the model parameters that best represent a signal.
3. An example shows determining the parameters a and b for an ARMA(1,1) model that estimates a signal x from another signal y by minimizing the mean square error between x and the model output. The parameters are calculated from the autocorrelation and crosscorrelation