Downsampling Downsamp pling trades efficiency with accuracy. Two downsampling schemes are introduced I Frequency domain perspective(Input downsampling) u[n]-AA filterDH DT plant yIn LPF Downsampling Ss Interpolation 2. Time domain perspective(Output downsampling): L DT plant Lifting SS Interpolation large step transition small step transition tput(state)traj Interpolation Many interpolation schemes are available. Newlsim chooses MATLAB command interp. m Downsampled by 4 expe showing Sinusoidal signals are downsampled nd th Lineal inte pol with 1000 Hz original sampling rate f frequency domain method is fairly accurateDownsampling u[n] y[n] AA filter D DT plant I 1. Frequency domain perspective (Input downsampling): LPF Downsampling SS Interpolation u[n] y[n] L DT plant I Lifting SS Interpolation 2. Time domain perspective (Output downsampling): • Downsampling trades efficiency with accuracy. Two downsampling schemes are introduced: output (state) trajectory small step transition large step transition Interpolation • Many interpolation schemes are available. Newlsim chooses MATLAB command interp.m. • A small experiment showing accuracy: • Sinusoidal signals are downsampled and then interpolated, with 1000 Hz original sampling rate. • Frequency domain method is fairly accurate. 100 101 102 103 10-8 10 -6 10-4 10 -2 100 10 2 104 f Linear intepolation Cubic spline Frequency domain interpolation Downsampled by 4