International Journal of Engineering and Technology

Sensing of spectrum holes in a frequency spectrum is one of the important concepts in implementing a cognitive radio system. Cognitive radio provides a way to use the band width effectively and efficiently by identifying the spectrum holes in a particular spectrum. The presence of cyclostationary features indicates the absence or presence of primary users. The presence of signal or noise can be determined by calculating the threshold of a signal by using cyclic cross-periodogram matrix of the corresponding signal. To circumvent the difficulty in estimating the accurate threshold (statistical techniques were used by other researchers), an artificial neural network has been trained by extracted cyclostationary feature vectors which have been obtained by FFT accumulation method. 70% of extracted data has been used for training and the rest 30% has been used for testing the efficiency of the network in estimating 99% accurate prediction of the threshold. The regression plot clearly indicates the superiority of the proposed scheme in estimating the threshold. Similar threshold samples derived from the data (other samples) have also been experimented in this scheme, which provided consistently good results with reduced MSE.