Based On The Cooperation Of Different Fusion Rule Of Spectrum Perception Performance Analysis

Spectrum sensing has been considered as a crucial technology for cognitive radio,and high accurate spectrum sensing can protect primary user from interference moreefficiently. Multi-user cooperation can provide a significant improvement for spectrumsensing performance. Excellent detection performance can be obtained with an appropriatefusion rule. Therefore, selection of optimal fusion rule is a focus of this paper.The chief content of this study is mainly about spectrum sensing in cognitive radio.Taking the average power as test statistic, the paper highlights the analytical expressionsfor the probabilities of detection and false alarm over AWGN channels. On this basis, theexpressions for detection probability are deduced over various fading channels. Then,analyze the detection performance of hard fusion over different channels and draw a seriesof new results. The main research and contribution of this thesis can be listed as following:1. When the number of sampling points is large enough （M>50）, the Gaussianapproximated expressions for energy detection have been used extensively in manyliteratures, the feasibility of which is analyzed in the thesis. Then, the spectrum sensingperformance of the Gaussian approximated expressions is simulated and analyzed. Bycomparing with the results of the analytical expressions and Monte Carlo simulations overAWGN and Rayleigh channels, it comes to the following conclusions:（1） The analytical expressions can provide an accurate spectrum sensing, but theapproximated expressions overestimate spectrum sensing performance in all cases;（2） The approximated expressions can not be applied to the local spectrum sensingwhen a low false alarm probability is required. The lower the false alarm probability, thelarger the differences.（3） The approximated expressions also can not provide an accurate detection forcooperative spectrum sensing using OR rule. The lower the false alarm probability, thelarger the differences.2. It is usually suggested in literatures that OR rule is always taken as a good choicefor cooperative spectrum sensing. With a concreted and deeper exploration，it analyzes and compares the spectrum sensing performance of three kinds of fusion rules （OR rule,AND rule, and Majority rule） in different cases, respectively. When the reporting channelsare perfect, results indicate the following conclusions:（1） Majority rule is the optimal fusion rule over AWGN channels;（2） OR rule is a good choice for Rayleigh channels in small cognitive networks, butMajority rule is the best choice for Rayleigh channels in large cognitive networks;（3） When the fading parameter is small over Nakagami-m and Rician channels,Majority rule is substituted for OR rule to become the best one; when the fading parameteris large over Nakagami-m and Rician channels, OR rule is the best fusion rule for smallcognitive network, and Majority rule also performs best over Nakagami-m and Ricianchannels at low average SNR in large cognitive networks;（4） Majority rule has the best performance in Log-Normal channels with a smallfading parameter σ, but OR rule has the best performance in Log-Normal channels with alarge fading parameter σ.3. When the reporting channels are imperfect, the influence of error probability onthe detection performance is discussed. When the SNR is higher in sensing channels, theselection of optimal fusion rule is investigated in four cases （sensing channels andreporting channels are AWGN channels and Rayleigh Channels, respectively）, it obtainsthe following conclusions:（1） When the reporting channels are AWGN channels, the error probability is low,which has a small influence on communication, the choice of optimal fusion rule isconsistent with the ideal cases:A. When the sensing channels are AWGN channels, Majority rule performs best;B. When the sensing channels are Rayleigh channels, OR rule is a good choice forRayleigh channels in small cognitive networks, but Majority rule is the best choice forRayleigh channels in large cognitive networks.（2） When the reporting channels experience Rayleigh fading, the error probability ishigh, which has great influence on the performance of cooperative spectrum sensing. ANDrule always performs worse. There is blind area in detection using OR rule.A. When the sensing channels are AWGN channels, Majority rule performs best; B. When the sensing channels are Rayleigh channels, there is a point of intersectionbetween the detection performance of OR rule and Majority rule, its correspondingcooperative false alarm probability is indicated by Q_f0, and Q_f0≥min（Q_f）.When Q_f0>0.1,Majority rule proves to be the optimal fusion rule; when Q_f0<0.1and Q_f>Q_f0, OR rule hasthe best spectrum sensing performance; when Q_f0<0.1and Q_f<Q_f0Majority rule is the bestchoice for cooperative spectrum sensing.