Research On Spectrum Sensing Strategies In Cognitive Radio Networks

With the advance of wireless communications and the increase in the number of wireless devices,the growing scarcity of spectrum resources has emerged as a critical issue.As an intelligent spectrum sharing technology,cognitive radio(CR)is regarded as one of the most promising technologies for future wireless communications.With the help of cognitive radio technology,secondary users(SUs)can sense idle authorized spectrum bands for opportunistic use,or SUs can transmit data with minimal harmful interference to primary users(PUs).Therefore,spectrum sensing is a key function of cognitive radio to require SUs to be able to quickly and accurately detect whether there is an idle spectrum band at a certain time or place for their opportunity to use.This dissertation focuses on optimal spectrum sensing strategies in cognitive radio network(CRN),and studies the optimal sensing parameters by constructing the objective function of the optimization problem which considering the tradeoff of different performance parameters.Firstly,the tight error bound of Gaussian approximation(GA)for chi-square distribution is given,and the square root Gaussian approximation(SRGA)and cubic root Gaussian approximation(CRGA)for chi-square distribution are applied to performance analysis of spectrum sensing.Secondly,considering the tradeoff between false alarm probability and miss-detection probability,the detection error rate function is constructed and the closed-form analytical expression of optimal detection threshold is derived.Then,considering the tradeoff between sensing performance and successful packet transmission probability,the spectrum opportunity utilization function is constructed to determine the optimal sensing time.Finally,the detection error rate function and spectrum opportunity utilization function of multi-user cooperative sensing are constructed,and the optimal detection threshold and optimal sensing time of cooperative sensing are determined respectively.The main contributions of this dissertation are as follows.1.For the accurate approximations of chi-square distribution,the tight asymptotic error bound of Gaussian approximation for chi-square distribution is derived,and square root Gaussian approximation and cubic root Gaussian approximation for chi-square distribution are applied to performance analysis of spectrum sensing which outperform traditional Gaussian approximation method.Firstly,through numerical and theoretical analysis of the Gaussian approximation error function of chi-square distribution,the tight asymptotic error bound of Gaussian approximation is derived.The deduced tight error bound can be used to quickly determine freedom parameters that meet certain error requirement.Secondly,two more accurate approximation methods,square root Gaussian approximation and cubic root Gaussian approximation for chi-square distribution,are presented,and they are applied to the performance analysis of spectrum sensing.The closed-form expression of the receiver operating characteristic(ROC)is derived and the relationship between number of samples required for energy detector and signal-to-noise ratio(SMR)is determined.The numerical results show that the performance of the above two approximation methods are significantly better than that of the traditional Gaussian approximation method,and cubic root Gaussian approximation for chi-square distribution has minimal error of all three approximation methods.2.For the optimal detection threshold for spectrum sensing,by considering the tradeoff between false alarm probability and miss-detection probability,the detection error rate function is constructed and the closed-form analytical expression of optimal detection threshold is derived,which outperforms traditional Newman-Pearson(NP)criterion.Firstly,according to the effect of detection threshold on false alarm probability and miss-detection probability,it is found that the appropriate threshold setting can achieve the tradeoff between the above two performance parameters.Secondly,for traditional NP criterion,detection threshold is set by a given target false alarm probability,but this cannot guarantee lower miss-detection probability.Therefore,minimizing the detection error rate function to determine the detection threshold can make the two performance indicators to achieve effective tradeoff.Finally,the closed-form analytical expression of the optimal detection threshold is derived based on the minimum detection error rate,and the effect of different system parameters on the optimal detection threshold is analyzed.The numerical results show that the optimal detection threshold can reduce the SNR requirement,and the detection performance is better than the traditional NP criterion.3.For optimal sensing time for spectrum sensing,by considering the tradeoff between sensing performance and successful packet transmission probability,spectrum opportunity utilization function is constructed and the optimal sensing time for the non-synchronous PU-SU system is determined,which improved the utilization efficiency of the idle spectrum.Firstly,the effect of sensing time on sensing performance of CRN is analyzed.Longer sensing time can improve the spectrum sensing performance and increase the opportunity to access the idle spectrum band,but the data transmission time will be reduced,which will lead to a reduction in the amount of data transmitted,or a data transmission failure due to the collision to the PU.Secondly,in the case of the synchronous PU-SU system,the throughput function is constructed by using the tradeoff between sensing performance and throughput,and the optimal sensing time is determined by maximizing throughput function.Finally,for the non-synchronous PU-SU system,the opportunity spectrum utilization function is constructed based on the tradeoff between sensing performance and successful packet transmission probability and the optimal sensing time is determined by maximizing the spectrum opportunity utilization function.The numerical results show that the optimal sensing time can reduce the SNR requirement and improve the utilization efficiency of idle spectrum.Also,the synchronization information among PU and SU is beneficial to reduce the sensing time.4.For multi-user cooperative sensing parameter optimization,the analytical solution for the optimal detection threshold for soft decision cooperative sensing is derived by minimizing the detection error rate function and the optimal sensing time is determined based on the tradeoff between sensing performance and throughput.Firstly,the relative increase in detection probability(RIDP)is used to represent the cooperative gain of hard decision,and several common soft decision combination schemes are presented.Secondly,by minimizing the detection error rate function,the analytical expression of the optimal threshold for soft decision cooperative sensing is reduced.Simulation results show that the optimal detection threshold increases with increase of number of SUs.This is because the multi-user collaboration can be equivalent to improve SNR of sensing channel.Finally,based on the tradeoff between sensing performance and throughput,the problem of determining optimal sensing time of cooperative sensing is discussed.The numerical results show that the using of sensing channel information of soft decision method can effectively reduce the sensing time and improve the utilization efficiency of idle spectrum.