EE504 ADAPTIVE SIGNAL PROCESSING
| Course Code: | 5670504 |
| METU Credit (Theoretical-Laboratory hours/week): | 3 (3.00 - 0.00) |
| ECTS Credit: | 8.0 |
| Department: | Electrical and Electronics Engineering |
| Language of Instruction: | English |
| Level of Study: | Graduate |
| Course Coordinator: | Prof.Dr. TOLGA ÇİLOĞLU |
| Offered Semester: | Spring Semesters. |
Course Objectives
1) Develop different adaptive filtering methods along the two main tracks: steepest descent and Newton's methods.
2) Form a view of two basic functional aspects of adaptive filtering as a signal estimation method and as a parameter estimation method.
3) Settle the significance of autocorrelation and cross correlation matrices in the context of Wiener filter as an adaptive filter.
4) Introduce stochastic gradient estimation methods.
5) Introduce least squares methods.
6) Contrast stochastic gradient estimation based methods and LS methods
7) Introduce performance assesment measures, analyze transient and steady state performances of adaptive algorithms.
8) Relate and contrast transversal AF, IIR AF, recursive AF/Kalman filtering
9) Regarding the design and performance of an adaptive filter, settle the interaction among the role of filter length, step-size and spectral content of filter input signal
10) Study computational aspects, introduce computational varieties of adaptive algorithms.
Course Content
Overview of discrete-time stochastic processes. Wiener filter theory. Linear prediction. LMS algorithm and its variants. Frequency domain adaptive filtering; RLS, QR-RLS algorithms and their connection to Kalman Filtering. Order recursive adaptive filters; QRD-LSL algorithm and its variants. Analysis and discussion of adaptation algorithms and their convergence properties. Computational complexity considerations. Filter structures and algorithms for fast adaptation and real-time processing. Numerical stability of fast algorithms. IIR adaptive filters. Applications of adaptive filtering.
Course Learning Outcomes
1) For a given linear adaptive estimation problemand its requirements, choose appropriate adaptation methods.
2) For a given linear adaptive estimation problem and its requirements, choose appropriate filter length.
3) For a given linear adaptive estimation problem, identify relevant signals, express adaptation and filtering operations.
4) Write adaptive filtering codes and compare the performances of adaptation methods.
5) Correctly choose or decide on the strategy about the step size parameter according to the nature of the problem and/or computational environment.
6) Propose ways to reduce computational load of algorithms.
7) Propose ways to improve numerical stability of algorithms.
Program Outcomes Matrix
| Contribution | |||||
| # | Program Outcomes | No | Yes | ||
| 1 | Depth: Our graduates acquire in depth knowledge in one of the various specialization areas of Electrical and Electronics Engineering, they are informed about current scientific research topics and they implement innovative methods. | ✔ | |||
| 2 | Breadth: Our graduates get familiarized in other subspecialty areas related to their specialization in Electrical and Electronics engineering and/or relevant areas in other disciplines. | ✔ | |||
| 3 | Research: Our graduates acquire the skills to conduct and to complete scientific research by accessing contemporary knowledge in their specialty areas. | ✔ | |||
| 4 | Life-long learning: Our graduates develop their life-long learning habits. | ✔ | |||
| 5 | Communication skills: Our graduates concisely communicate their ideas and work related results in written and oral form. | ✔ | |||
| 6 | Ethics: Our graduates internalize rules of research and publication ethics as well as professional ethics. | ✔ | |||