Resumen en español

This PhD thesis presents the development of new algorithms to perform Wideband Spectrum Sensing (WSS) in Cognitive Radio (CR) by using the sparse Fast Fourier Transform (sFFT) algorithms. On the one hand, the sFFT is a new mathematical tool that computes the Discrete Fourier Transform (DFT) of a sparse signal in frequency domain faster than traditional Fast Fourier Transform (FFT) algorithms for dense signals; this is achieved by using Random Sampling (RS) and ltering windows with small support, in order to obtain a set of samples that is smaller than the original signal. On the other hand, CR is a new paradigm in communication systems oriented to improve the management of the radio electric spectrum by developing smarter radios that are aware of the available spectrum, to use it and share it in a reliable way; the accomplishment of such tasks is supported by the WSS function, inside the CR, that can detect spectral holes or occupied frequency bands in a large bandwidth of frequencies. This PhD thesis addresses the design of a suitable sFFT algorithm that allows building a reliable WSS system using low sampling rates. This goal was achieved by carrying out several research and development tasks as follows: rst, an in depth analysis of the sFFT algorithms was made by considering all the mathematical issues. Second, after performing the analysis, a sFFT algorithm was chosen, and it was modi ed and software-implemented on a multicore computing platform. Third, the modi ed sFFT algorithm was implemented on a hardware/software platform using a hardware accelerator, which can be used to speed up other sFFT algorithms. Fourth, a novel WSS system with low sampling rates was designed using the modi ed sFFT algorithm with new improvements. And fth, a novel sFFT algorithm was designed and hardware-implemented, this algorithm allows the implementation of Cooperative WSS (CWSS) systems with low sampling rates and avoiding the usage of RS, which requires complex analog circuits.