Project description
Sensing and communications systems produce data that are irregularly sampled in time or space. When these differences are significant, traditional Fast Fourier Transform (FFT) techniques are difficult to apply. The Non-Uniform Fast Fourier Transform (NUFFT) provides a framework for processing such data and has applications in passive sensing, medical imaging, astronomy and wireless communications. This project will investigate algorithms and implementations for efficient processing of non-uniformly sampled signals using the NUFFT. Possible directions, depending on student interests, include:- Development and analysis of NUFFT-based algorithms for signal detection, estimation or localisation.
- High-performance implementations that could consider parallel computing, or GPUs.
- Investigation of numerical accuracy, stability and computational trade-offs.
- Comparison of NUFFT-based methods with conventional interpolation and FFT approaches.
- Consideration of an application of interest.
Assumed knowledge
This project includes aspects from signal processing, applied mathematics, scientific computing, optimisation and electrical engineering. Students should have strong interests in at least one of these areas. Experience and interest in mathematical or "low-level" coding is preferable.
Note: You need to register interest in projects from different supervisors (not a number of projects with the one supervisor).
You must also contact each supervisor directly to discuss both the project details and your suitability to undertake the project.