Signal processing deals with the analysis and operation on signals in engineering and natural systems, with important applications in diverse fields such as electrical engineering, computer science, chemical engineering, physics, bioinformatics and so on. A crucial part of signal processing is the design of the algorithms to perform the analysis and operation on the signals. This project aims to provide the participants with an opportunity to learn and implement one of the most common and powerful tools for signal processing algorithm design. Specifically, the goal of this project is to develop an optimization based algorithm to design digital low-pass filters. Low-pass filter has many important applications including averaging, de-noising and anti-aliasing, and it serves as a major building block for more advanced signal processing techniques. The expected outcome of the project is a piece of MATLAB code utilizing a popular optimization library (e.g., SeDuMi) to design optimal finite impulse response (FIR) filters based on given design specifications. The optimization modelling is primarily based on convex semidefinite programming. The participants will experiment with their design algorithm to study the performance of different filters due to different specifications. The experiment can be carried out for 1-D signals (e.g., sound) or 2-D signals (e.g., picture), depending on the progress of the group.
The participants are expected to be familiar with MATLAB. Also, it is desirable for the participants to have some basic exposure to frequency domain characterization of signals and systems, linear algebra and optimization, otherwise the participants will learn the working concept while working on the project. Students can contact Dr. Kin Cheong Sou for more information if they have concern about the prerequisite and the scope of the project.
Obs! För GU-studenter räknas projektet som ett projekt i Tillämpad Matematik (MMG900/MMG920).
Examinator Maria Roginskaya
Institution Matematiska vetenskaper