Usually energy-detection based spectrum sensing techniques for Cognitive Radios need, as preliminary operation, the spectrum segmentation in smaller sub-bands, in order to apply test statistic over each sub-band rather than over the entire span, resulting in a sensitivity and selectivity improvement of the output. Despite uniform subdivision is the easiest way, the problem resides on the optimal sub-band width to achieve sensing purposes. In this paper, three different approaches based on wavelet transform are proposed as preliminary stage of a frequency domain energy detection scheme. The advantage in using a wavelet based segmentation is: (i) spectrum is segmented according to its irregularities, (ii) segmentation is not uniform but adaptive to the scenario, (iii) the number of sub-bands is usually less than in the uniform case, (iv) it employs a mathematical tool with fewer DoFs than usual optimization procedures and (v) it can be made completely automatic. As confirmed by obtained results, the considered approaches guarantee different levels of accuracy, even if all of them provide good results for the analyzed figures of merit.
Analysis of different wavelet segmentation methods for frequency-domain energy detection based spectrum sensing
Capriglione, D.;Cerro, G.;Ferrigno, L.;Miele, G.
2017-01-01
Abstract
Usually energy-detection based spectrum sensing techniques for Cognitive Radios need, as preliminary operation, the spectrum segmentation in smaller sub-bands, in order to apply test statistic over each sub-band rather than over the entire span, resulting in a sensitivity and selectivity improvement of the output. Despite uniform subdivision is the easiest way, the problem resides on the optimal sub-band width to achieve sensing purposes. In this paper, three different approaches based on wavelet transform are proposed as preliminary stage of a frequency domain energy detection scheme. The advantage in using a wavelet based segmentation is: (i) spectrum is segmented according to its irregularities, (ii) segmentation is not uniform but adaptive to the scenario, (iii) the number of sub-bands is usually less than in the uniform case, (iv) it employs a mathematical tool with fewer DoFs than usual optimization procedures and (v) it can be made completely automatic. As confirmed by obtained results, the considered approaches guarantee different levels of accuracy, even if all of them provide good results for the analyzed figures of merit.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
