| Spectrum sensing is one of the key technology of cognitive-radio. Spectrum sensingof High accuracy can better prevent the user from interference. However, multi-antennasbase-station can cooperate with the cognitive multi-users around, which will greatlyimprove the performance of spectrum sensing. So that collaborative networks can get thebest detection performance. This thesis focuses on studying the multi-antennasbase-station and the cognitive radio network of the surrounding cognitive multi-users andproving the necessity and reliability of the cooperation of multi-antennas base-station andcognitive multi-users around.The chief content of this study is mainly about spectrum sensing in cognitive radionetworks. Taking the average power as test statistic, the paper highlights the analyticalexpressions for the probabilities of detection and false alarm over AWGN channels. Onthis basis, the expressions for detection probability are deduced over various fadingchannels. Then, analyze the detection performance of soft fusion and hybrid fusion overRayleigh channel and draw a series of new results. The main research and contribution ofthis thesis can be listed as following:1. When the test statistic of cognitive radio users is sent to the fusion center ofbase-station and collaborate with the test statistic of multi-antennas in base-station, we callthis kind of fusion scheme as soft fusion rule. This thesis first analyze the performance ofthis fusion scheme in difference cognitive radio networks, and then compare with theperformance of OR criterion of cognitive radio users and with the performance of equalgain combining of multi-antennas in base-station in available literature. The thesisanalyzes the detection performance of the above three cooperation by simulation andcompare its analytical expression with the results of Monte Carlo simulation, which findsthat the performance of spectrum sensing of soft fusion rule is always the best in anydifferent radio network.2. In this thesis a new fusion scheme is presented. Because of the bandwidthlimitation of the control channels, the cognitive radio users around the base-station can only compare the test statistic with the local threshold and send the resulting decision ofthe comparison to base-station (fusion center). Then base-station will quantify theresulting decision to soft information (quantized power) and cooperate with the teststatistic of multiple antennas in base-station to get the final result. We call this kind offusion scheme as hybrid fusion scheme. The thesis explore the fusion scheme and analyzeits performance to draw the conclusion as follows:(1) In the different cognitive radio network, by means of Neyman-Pearson criterionto make the local threshold of best performance of hybrid fused spectrum sensing close toa fixed value.(2) On the premise of the local threshold of the optimal value, in order to verify thevalidity of the hybrid fusion scheme, we compare and analyze performance of spectrumperception of the hybrid fusion rule, OR criterion and the equal gain combining rule. Wefind that:A. When there are few cognitive radio users around the base station, no matter theSNR(signal-to-noise ratio) of the cognitive radio network is high or low, and no matter inbase station the antenna number more or less, compare with the other two fusion rule, theperformance of hybrid fusion cooperation is always the best;B. When there are many cognitive radio users around the base station, change of theother two parameters (signal-to-noise ratio, the number of antenna in base station) ofcognitive radio network, the performance of the hybrid fusion cooperation is still the mostoptimal; The more cognitive users are, the more obvious advantage of the performance ofthe hybrid fusion scheme is;C. When the SNR is low, hybrid fusion scheme still has great improvement ofperformance, the larger the SNR is, the more obvious performance improvement is.(3) On the premise of the local threshold of the optimal value, analyze the effectwhich the base station antenna under different cognitive radio networks have on spectrumsensing performance of hybrid fusion scheme, we can know that the number of the basestation antenna have little effect on the performance of cooperation under hybrid fusionscheme.(4) In hybrid fusion scheme, when the cognitive radio network parameters (such as signal-to-noise ratio, the number of cognitive radio users and the umber of multi-antennasin base-station) change, it firstly proves that the local threshold which makes hybrid fusionperformance optimal close to a fixed value. Thus the fusion center can make use of thefixed value to get the quantized power of different resulting decision so that the quantizedpower won’t change with cognitive radio network parameters to make the hybrid fusionscheme easy and practicable. Then according to the amount of different resulting decisionsent and the quantized power of quantized cognitive radio users, we will get the resultingdecision under the cooperation with total power of the testing of antenna in the basestation.3. In hybrid fusion only when the multi-antennas base-station cooperate with thequantized power of optional local threshold, the performance of hybrid fusion will becomeoptimal. Thus firstly we should analyze what effect the cognitive network parameters willhave on the optional local threshold. Then the total power of fusion center need tocompare with the global threshold to get the final resulting decision. Thus we shouldfurther validate how the changes of signal-to-noise ratio affect the global threshold.Making use of the above two results to determine the robustness of the hybrid fusionscheme, conclusions can be drawn as follows:(1) Effects that cognitive network parameters have on the optimal local threshold:A. When the number of the antennas of base station become fixed value if thenumber of the cognitive users is different, the section of the optimal local threshold arealso different. When there are few of cognitive users, the section of the optimal localthreshold in hybrid fusion will be small. And if the local threshold is changed it will havegreat effect on the performance of hybrid fusion scheme. However, if there are manycognitive users, the curve will be smooth so that the sections of the optimal local thresholdof hybrid fusion scheme will become wider and the small changes of the local thresholdwill not affect the spectrum sensing performance a lot. But no matter the number of thecognitive user is small or big, the interval value of optimal local threshold will around1.14.B. When the number of the cognitive and the signal-to-noise ratio are certain, thebigger the signal-to-noise in the network of spectrum sensing is, the wider the section of the local threshold. In different signal-to-noise ratio the interval value of optimal localthreshold will still around1.14.C. when the number of the cognitive and the antennas of base station are certain, thebigger the signal-to-noise ratio in spectrum sensing network is, the widder the section oflocal threshold is. Even optimal local threshold under different signal-to-noise ratio theinterval value is still around1.14.According to the above three points we can conclude that when cognitive radionetwork parameters change, all the sections of the optimal local threshold have onecommon point1.14. We can take this value as the optimal local threshold. Thus thecognitive radio network parameters will not change the value of the optimal localthreshold.(2) When the false alarm probability, the cognitive users, and the number of basestation antennas are fixed value, the local threshold will be set to1.14, and thesignal-to-noise ratio has little impact on global threshold.According to conclusion (1) and (2) we can know that the hybrid fusion scheme is ofstrong robustness.(3) Finally comparing the spectrum sensing performance of the two different fusionrule of multi-antennas base-station and the cognitive multi-users around, and combiningwith the implementation of these two fusion rules, we can draw a conclusion that hybridfusion scheme is the optimal spectrum detection method. |
PDF Full Text Request