Short-Term Measurements What is the point of this lecture? Recursive 50 Hz LPF Time-Domain Windowing Let denote a sequence to be analyzed. Let's limit the duration of to samples: where is a rectangular window and is defined as . The Fourier transform of is given by: The transform of is given by: . This introduces frequency domain aliasing (the so-called picket fence effect): Improvements Via Better Windows Popular Windows 1. Rectangular: 2. Generalized Hanning: 3. Bartlett 4. Kaiser 5. Chebyshev: 6. Gaussian There are many others. The most important characteristics are the width of the main lobe and the attenuation in the stop-band (height of highest sidelobe). The Hamming window is used quite extensively. Recursive-in-Time Approaches Define the short-term estimate of the power as: We can view the above operation as a moving-average filter applied to the sequence . This can be computed recursively using a linear constant-coefficient difference equation: Common forms of this general equation are: (Leaky Integrator) (First-order weighted average) (2nd-order Integrator) Of course, these are nothing more than various types of low-pass filters, or adaptive controllers. How do we compute the constants for these equations? In what other applications have we seen such filters? Relationship to Control Systems The first-order systems can be related to physical quantities by observing that the system consists of one real pole: can be defined in terms of the bandwidth of the pole. For second-order systems, we have a number of alternatives. Recall that a second-order system can consist of at most one zero and one pole and their complex conjugates. Classical filter design algorithms can be used to design the filter in terms of a bandwidth and an attenuation. An alternate approach is to design the system in terms of its unit-step response: There are many forms of such controllers (often known as servo-controllers). One very interesting family of such systems are those that correct to the velocity and acceleration of the input. All such systems can be implemented as a digital filter. Time/Frequency Properties of Windows