Research On Multiantenna Based Spectrum Sensing Technology

Cognitive radio(CR)has been widely considered as a key enabling technology to improve the spectrum utilization efficiency in future wireless communication systems.In the CR paradigm,the unlicensed secondary users(SUs)are allowed to opportunistically access the unused licensed spectrum bands without causing harmful interference to the licensed primary users(PUs),thus improving the spectrum utilization significantly.As an essential component of CRs,spectrum sensing(SS)plays a key role in CRs to probe the activities of PUs,which is critical for implementing CR.Due to the extension of the spatial domain,multiantenna SS(MSS)has been an important research topic in CR and has attracted considerable interest in recent years.Thus,this dissertation is mainly concerned with the MSS problem in CR networks,and the main contributions are outlined as follows.Firstly,we discuss the MSS problem in the channel uncertainty environment.Considering a CR network with correlated multiple antennas,in which the fading channel between PU and SU can be either quasi-static or time-varying during one sensing period.Under such a channel uncertainty environment,most of the existing multiantennabased detection methods cannot provide a reliable sensing performance.To address this challenge,a two-stage detection(TD)method based on the eigenvalue moment ratio and weighted-covariance-based methods is proposed.The false alarm and detection probabilities of the proposed TD method are also derived.Simulation results illustrate that the proposed TD method can effectively guarantee the detection probability in the channel uncertainty environment.Secondly,we focuse on the MSS problem in the time-varying Rayleigh fading channel.(i)Considering the scenario where the SU is equipped with a large number of low correlated antennas,and a robust blind sensing method based on the Ljung-Box(LB)test is proposed,which utilizes the autocorrelation function of the receive time-series to probe the primary signal.In the scenarios where the correlation between receive antennas is low,numerical results demonstrate that the proposed LB-based method is robust to the noise uncertainty(NU),and is capable of outperforming the existing covariance-based methods in the time-varying Rayleigh fading channel.(ii)Considering the scenario where the SU is equipped with a small number of correlated antennas,the weighted covariance-based detection(WCD)method for MSS can deliver a desirable performance in the spatially correlated time-varying Rayleigh fading channel.However,it suffers a high computational complexity.It is well known that one complex multiplication involves two real additions and four real multiplications.To circumvent this issue,the complex-valued MSS problem is equivalently converted into the real-valued MSS problem,and a real-valued WCD(RWCD)method is proposed.In particular,the asymptotic expression for the detection probability of the proposed RWCD method is also derived in the low SNR regime.Numerical results reveal that the RWCD method can yield almost the same performance as the WCD method,but with much lower complexity.(iii)Based on the RWCD method,a generalized RWCD(GRWCD)method is presented.Meanwhile,the distribution of the GRWCD statistic under the null hypothesis is derived,which enables us to calculate theoretical detection threshold for a predefined false alarm probability.Additionally,the distribution of the GRWCD statistic under the alternative hypothesis is also presented,which allows us to provide an analytical expression for the detection probability as well as theoretical receiver operating characteristic(ROC)curve.Simulation results are provided to verify the accuracy of the derived results and demonstrate that the proposed GRWCD method is capable of providing performance improvement over the RWCD method.Thirdly,we concentrate on the MSS problem for noncircular(NC)signal in the scenario that all antennas have the same noise variance.By taking the elements of both covariance and complementary covariance matrices into account for constructing the test statistic,a novel detection method called NC local average variance(NC-LAV)is developed to exploit the NC characteristic of the primary signals for MSS.The distribution of the NC-LAV statistic under the null hypothesis is first derived.Then,based on the derived distribution,we present an asymptotic threshold determination of the proposed NC-LAV method.Compared to the peer methods,the proposed NC-LAV method exhibits much lower complexity.Meanwhile,numerical results illustrate that the proposed method achieves better performance than the conventional methods.Lastly,the problem of MSS is addressed for NC signal in the scenario where perantenna noise variances are unequal.Although NC-LAV method can utilize the NC characteristic of the primary signals to improve the detection performance,howbeit the NCLAV method relies heavily on the assumption that the noise variances at all antennas are identical.In practice,such an assumption may not hold because the SU receivers are usually uncalibrated,thereby its performance is sensitive to unequal per-antenna noise variances.To do so,by taking both the standard covariance and complementary covariance information of the NC signal into account,a new robust MSS method called NC covariance(NCC)is proposed.Specifically,we show that the standard covariance and complementary covariance matrices of the received signals differ between the null and the alternate hypotheses,which can be used for detecting the presence of PUs.Meanwhile,we derive the theoretical detection threshold of the NCC method.Simulation results reveal that the proposed NCC method is capable of outperforming state-of-the-art methods.