COGNITIVE RADIO SPECTRUM SENSING WITH IMPROVED EDGE DETECTION OF FREQUENCY BANDS

摘要

A spectrum sensing method for cognitive radio wherein spectrum holes are detected in a wireless environment having spectrum scarcity. First, a cognitive radio user (CR) determines the power spectral density (PSD) of a wideband signal and detects subbands within the wideband using wavelet transforms (WT). WT coefficients are calculated by convolving the PSD with first derivatives of wavelet smoothing functions. The extrema of the WT coefficients demark frequency subband edges. Detecting subband edges becomes more robust against noise by median filtering the PSD before calculating WT coefficients, summing over WT coefficients with different scale factors, and suppressing WT coefficients below a noise threshold. After identifying subbands, the CR determines subband availability by measuring the subband power and signaling the power to a fusion center receiving power measurements from multiple cooperating CRs, and final decisions are based on data and decision fusion.

主权项

1. A method for spectrum sensing for cognitive radio, the method comprising:
detecting edges of frequency bands, detecting energies in frequency bands, and to deciding whether frequency bands area available to cognitive radio users; receiving a time-domain signal by a cognitive radio user; estimating a power spectral density of the time-domain signal; calculating wavelet transform coefficients at one or more scale factors by convolving the power spectral density with a first derivative of a wavelet smoothing function; obtaining a positive signal by taking the absolute value of all wavelet transform coefficients, and if more than one wavelet scale factor is used, combining wavelet transform coefficients of different scales; comparing the positive signal to a noise threshold and where the positive signal is less than the noise threshold setting the positive signal to zero; detecting one or more frequency-band edges of the power spectral density by solving for peak frequencies, wherein the peak frequencies are frequencies of the local maxima of the positive signal; defining a plurality of frequency bands as all frequencies between nearest neighbor pairs of frequency-band edges; determining a received energy value for a first frequency band of a plurality of frequency bands comparing the received energy value to an energy threshold; and signaling frequency band availability based on the comparison between the received energy value and the energy threshold.