Based On The Research On Energy Detection Of Noise Uncertainty

With the development of wireless communications technology, there have been aseverely unbalance between spectrum resources and spectrum utility. The spectrum bandsthat can be used come to saturation, while many spectrum bands which have beendistributed are not used efficiently.It is widely accepted that the emergence of cognitive radio will improve theutilization efficiency of the existing radio spectrum. The main idea of cognitive radiotechnology is “See seam needling”, which is adopted to reuse the distributed spectrumresource. Spectrum sensing is one of the critical technologies in cognitive radio networksfor primary user detection, and energy detection is widely used for spectrum sensing dueto its generality and low complexity. However, noise power level variety with time, whichis also called noise uncertainty, degrades the sensing performance of energy detectionseverely. Although several spectrum sensing methods are proposed to overcome theinfluence of noise uncertainty, according to the latest survey, no results on sensingperformance comparison of hard fusion rules under noise uncertainty has been reported.In the paper, we adopt average power as the decision statistic and illustrate thesystem model of the noise power uncertainty firstly. Then we derive the closed formedexpressions for the average probabilities of detection and false alarm under noiseuncertainty, and discuss the sensing performance under different conditions. Exceptconsidering the impact of the fading effect is a major factor, we also studied the shadoweffect based on the noise uncertainty, deduced the closed form of the expression and theperformance analysis. We mainly study the performance of the three kinds of well knownhard fusion rules (OR, AND and Majority) over AWGN and Rayleigh channels, andidentify the best rule in different conditions. In the paper, we mainly discussed thedifferent SNR and different signals (real signal, complex signal).Extensive simulations indicate that in AWGN channels, when SNR is much higher,whether the network is big or small，Majority rule is the optimal rule of the three rules.When noise uncertainty is low, to obtain the desire performance, OR rule performs thebest performance. With the increase of cooperative users’ number, AND rule has muchmore advantage than OR and Majority rules, AND rule performs the best in large scalenetwork. Moreover, in Rayleigh channels the conclusion is different from in AWGNchannels. When noise uncertainty is moderate or much higher at high SNR, OR rule is the optimal rule, while Majority rule performs the best performance with the increases ofnetwork users. However, when noise uncertainty is low at low SNR, OR rule is muchbetter than the other two rules, Majority rule gains the best performance in much largenetworks. Finally, OR rule always performs the best no matter how many users in thecognitive radio networks when noise uncertainty is moderate or much higher at low SNR.Our research is very helpful and meaningful for selecting the proper hard fusion rule inpractical cognitive radio networks.