Instructor: Panagiotis Tsakalides, Research Assistant Professor
Department of Electrical Engineering -- Systems
Signal and Image Processing Institute, Room EEB 428
University of Southern California
Los Angeles, California 90089-2564
Office Hours: Monday and Thursday 1:30-3:00 pm or by appointment.
Administrative Contact: For administrative questions, contact Ms. Kathi Collins, phone: 213-740-0115, fax: 213-749-3289.
Required Text: Higher-Order Spectral Analysis , C. L. Nikias and A. P. Petropulu, Prentice Hall, 1993.
Prerequisites: Introductory courses in digital signal processing (e.g., USC's EE483) and random processes (e.g., USC's EE562a.)
Course Description: Definition of moments/cumulants, power spectrum, higher-order spectra for both stochastic and deterministic signals; conventional methods, polyspectra methods, non-parametric methods, parametric methods based on AR, MA, and ARMA models, direction finding methods, adaptive filter methods.
Course Objective: On completion of this course, the student will have a firm basis of conventional and modern techniques for higher-order spectrum estimation, their complexity, performance comparisons and applications. These algorithms are widely used in signal processing applications, in both academic and industrial environments.
Grading Policy: The course-work will consist of written assignments and two programming projects. For the two projects, you can use any programming language on any machine available to you. In addition, you will need access to some kind of visualization software (use of MATLAB in an engineering workstation is recommended.) In particular, the grade will be based on the following:
|Activity||Percent||Seven homework sets||10%|
|Mid-term exam (February 23 or March 1)||20%|
|Project I (due March 22, a week after spring break)||20%|
|Project II (due May 3)||20%|
|Final exam (last day of class, April 26, 9:00-11:40 am, open books, open notes)||30%|
Detailed Course Outline:
PART I: BASIC DEFINITIONS AND PROPERTIES
EE 683 Home Page