Models Without Parameters: Step Response and Correlation Analysis

The viewer can (1) read static gain and speed off a step response and say why noise limits it, (2) explain the weighting sequence h(k) and the convolution sum, (3) derive E[y(t)u(t−τ)] = σ²h(τ) for white-noise input, and write the estimator ĥ(τ), (4) state the remarks: inaccurate with little data, gain off, describes input→output only.