MULTI-LEVEL COMPUTATIONAL COMPLEXITY REDUCTION FOR CYCLOSTATIONARY SPECTRUM SENSING DETECTOR

Presently the demand for robust wireless communicationwith coordinated use of mobile systems and the internetresult in ineffective use of the spectrum, where aconsiderable portion of the spectrum is neglected.Consequently, Cognitive Radio(CR) is innovativetechnology that raising the spectrum usage performance.Spectrum sensing(SS) is an essential aspect in detectswhether vacant channels are in the spectrum and rendersthem to secondary users. The cyclostationary spectrumsensing(CSS) technique is one of the most effective methodsable to efficiently detects unoccupied spectra with lowsignal-to-noise ratios(SNR). CSS involves the use of a largenumber of samples for detection resulting in highcomplexity, cost, and processing time. In this paper, a lowcomputational complexity CSS approach is proposed basedon using Haar wavelet transform and signed correlator. Theproposed method performance is evaluated over Rayleighflat fading and AWGN channels with low SNR values. Theacquired simulation results indicated the proposed methodachieved reduced complexity to about 80%, on a penalty avery slight reduction in detection probability Pd of about 7%at SNR equals -25 dB. Then the proposed method isoptimized to reduce the detection probability to 6%, and 2%through the use of stratified sampling, and BPF respectively.