Research On Spectrum Sensing And Power Recognition With Multiple Transmission Powers In Cognitive Radio Networks

The emerging new wireless communication techniques and explosive growth of mobile user equipment require future upcoming wireless networks to provide higher network capacity.At the same time,the huge increase of wireless services and the demands of high data transmission rate for mobile users have led to a dramatic increase in the demand of spectrum resources,which inevitably results in the shortage of the natural spectrum resources.Compared to the traditional static spectrum allocation strategy,cognitive radio allows the cognitive users to dynamically use the under-utilized licensed spectrum bands without affecting the communication of the primary users.Since cognitive radio has great potential in improving spectrum utilization and reducing network interference,it has become one of the current research hotspots.Spectrum sensing,deemed as the core technique of cognitive radio system,provides the guarantee for cognitive user to dynamically and opportunistically access the licensed spectrum band by monitoring the channel state of primary users.Specifically,with the development of wireless communication technology,the primary users no longer use a fixed transmission power,but may have multiple transmission powers according to the external environment.In this case,the traditional spectrum sensing schemes that only detect whether the primary user is active or not are inapplicable.This motivates us to investigate new spectrum sensing schemes that not only can detect the presence of the primary user,but also can recognize the transmission power of the primary user.In this paper,spectrum sensing and power recognition schemes are studied from the practical application perspective.The main research work and contributions are summarized as:1.A robust spectrum sensing scheme based on the differential eigenvalues is proposed to overcome the noise uncertainty.The difference between the maximum eigenvalue and the average of the remainder eigenvalues of sample covariance matrix is utilized as the test statistic.According to the random matrix theory,the distribution of the test statistic is derived.Then,the false alarm probability and the sensing threshold is obtained by using the Neyman-Pearson detection criterion.Numerical and simulation results reveal that the proposed scheme is invulnerable to the noise uncertainty,and can achieve better detection performance than the other existing methods.2.A maximum eigenvalue based spectrum sensing and power recognition scheme is proposed in hybrid spectrum sharing cognitive radio network,where the primary user has multiple transmission powers.Different with the traditional spectrum sensing scheme,the proposed scheme can not only detect the presence of the primary user,but also recognize the transmission power of the primary user.In this scenario,a multiple hypothesis testing model is utilized to identify the transmission power of primary users by leveraging the maximum a posteriori probability detection criterion.By approximating the distribution of the maximum eigenvalue of sample covariance matrix with Gaussian and Gamma distribution,closed-form sensing threshold as well as the decision regions are derived for identifying the transmission power of primary users.The detection performance and recognition performance are theoretically analyzed,and the theoretical analyses reveal that the proposed scheme can efficiently recognize the transmission power of the primary user,which allows cognitive users dynamically switch between the underlay and interweave model in the hybrid spectrum sharing cognitive radio network.3.In cognitive small cell networks,two spectrum sensing and power recognition schemes are proposed considering spatially correlated noise.Specifically,when the signaling features of transmitted signal are available,an optimal spectrum sensing and power classification(OSC)scheme based on the coherent detector is proposed.The total error rate is proved as a convex function with respect to the sensing threshold,such that the optimal sensing threshold is derived to minimize the total error rate.Once the primary is detected to be active,the transmission power of the primary user is identified by leveraging the maximum a posteriori probability detection criterion.Closed form decision regions for all the hypotheses are theoretically derived.For a realistic case where the signaling features of transmitted signal are unavailable,an alternative generic spectrum sensing and power classification(GSC)scheme based on the non-coherent detector is proposed.The optimal sensing threshold that minimizes the total error rate as well as closed form decision regions are derived.Extensive simulation results reveal that the detection probability is always higher than the classification probability of both the proposed schemes.Moreover,compared to GSC scheme,OSC scheme requires less samples to achieve a desired classification performance within a short sensing period,such that it can reduce the energy consumption during the sensing period.4.High order cumulants based spectrum sensing and power recognition schemes are investigated in hybrid spectrum sharing cognitive radio networks to deal with the non-Gaussian transmitted signal of primary users.For a given order and time lag,a single high-order cumulant based spectrum sensing and power recognition(SCSR)scheme is proposed.The proposed SCSR scheme extracts non-Gaussian information that embedded in the received signal to recognize the transmission power of the primary user,and is robust to noise uncertainty.To excavate the rich statistical non-Gaussian information of the primary signal,a hybrid multiple high-order cumulants based spectrum sensing and power recognition(HCSR)scheme is proposed.The proposed HCSR scheme exploits multiple cumulants with different order and time lag to conduct the test statistic.The sensing performance and recognition performance of proposed two schemes are theoretically analyzed.Simulation results are provided to show that both the proposed schemes are robust to the noise uncertainty.Moreover,HCSR scheme provides more accurate sensing and recognition performance at the cost of increasing the computational complexity.