Module description
The last three decades witnessed major advances in signal processing, most notably in the area of signal expansions localised both in time and frequency. The ability of such transforms to represent signals sparsely has led to step advances in tasks underpinning a wide range of applications, in areas ranging from consumer electronics, entertainment, wireless and mobile communications, through biomedicine, seismology, finance, to warfare, security and space exploration projects. One of the aims of the module is introduce fundamentals and principles of such time-frequency localised representations, including wavelets, filter banks, and short-time Fourier analysis.
Much of the data of the 21st century are signals defined on vertices of graphs, irregular, discrete, heterogeneous domains, and are of very large scales; examples include social, transportation, neural, biological, trade, sensor, communication and other networks. The emerging field of signal processing on graphs is combining concepts from graph spectral theory and computational harmonic analysis to extend existing methodologies of signal processing towards making sense of network data. Introducing signal processing on graphs is another objective of the module.
Assessment details
Written examination/s; coursework
Educational aims & objectives
Syllabus
1. Introduction
2. Fundamentals of signal decompositions
3. Discrete-time expansions and filter banks
4. Wavelet series expansions
5. Continuous wavelet transform
6. Signal processing on graphs