Abstract
The increasing demand for spectrum resources due to the rapid growth of broadband wireless services, commercialisation of 5G, and development of 6G communication techniques have highlighted the importance of addressing spectrum scarcity and underutilisation. Dynamic Spectrum Access and Cognitive Radio technology, supported by Compressed Sensing, offer promising approaches to improve radio spectrum utilisation and wireless communication efficiency.
This PhD thesis investigates various aspects of Compressed Spectrum Sensing (CSS) systems to address existing challenges and enhance practical implementation. The research contributes to some of the key techniques that enhance the performance of CSS systems in a practical framework, including refining spectrum recovery accuracy, optimising sampling pattern design, and improving reconstruction algorithms.
Key contributions include the development of a block Multiple Measurement Vectors (MMV) model to enhance spectrum support set reconstruction accuracy, addressing the limitations of traditional MMV models. The research also explores the design of sampling patterns for multicoset sampling in noisy environments and proposes two optimisation algorithms to minimise mutual coherence, improving spectrum recovery performance. Furthermore, a double-threshold matching pursuit algorithm is introduced to reduce high false-alarm rates in signal reconstruction.
A significant aspect of this thesis is the emphasis on system implementation. The research presents a software-defined system for CSS on mmWave to accelerate prototype validation and hardware design parameter selection. This system achieves real-time spectrum sensing for a 3.072 GHz bandwidth signal at a 28.5 GHz centre frequency, utilising Bayesian sparsity estimation and data decimation algorithms for robust performance.
Finally, the thesis includes the design and implementation of a multicoset sampler for wideband spectrum sensing, the first on-board realisation of a periodic nonuniform sampling structure. This achievement demonstrates the practical application of the proposed techniques and advancements in CSS system development.