Research On Highly-efficient And Secure Cooperative Spectrum Sensing Techniques

The advent of the internet of everything era will bring geometric growth of data traffic to wireless communication services,making the spectrum resource becomes increasingly scarce.Cognitive radio(CR)technology,by using dynamic spectrum access mechanism,is regarded as one of the promising technologies to resolve the spectrum scarcity and promote the development of the next cognitive networks.To realize the CR,spectrum sensing is the indispensable step to reliably detect the existence of available spectrum.Cooperative spectrum sensing(CSS)by utilizing multiuser spatial diversity to improve the overall sensing performance.However,due to the introduction of cooperation and the openness of CR networks,CSS faces specific security threats.Among them,spectrum sensing data falsification(SSDF)attack intentionally misleads the fusion center(FC)to make wrong spectrum decision result,which causes a waste of spectrum resource and interference to the primary user.In view of above problems,considering the limited communication overhead and SSDF attacks in CSS.This dissertation studies multibit quantization CSS,secure CSS under probabilistic and massive SSDF attack,high energy efficiency CSS under SSDF attack,and develops attack model theoretical analysis,security defense algorithm research and simulation verification.The main contributions are as follows:(1)Aiming at the limited communication resources in CSS,a multibit quantization soft data fusion CSS algorithm based on local energy detection is proposed.In the proposed algorithm,the central quantization threshold is firstly calculated according to the global false alarm probability.Furthermore,the optimal fusion rule with quantized data is also derived,and then a quantizer design algorithm by using the Bhattacharyya distance that describes the error probability bound as performance criteria is proposed.The mathematical model of the quantizer is constructed,and the optimum quantization thresholds are obtained by using particle swarm optimization algorithm.Simulation results show that the proposed multibit quantization CSS algorithm not only achieves detection performance approach to that of unquantized method,but also significantly reduces the size of transmitted data.(2)To defend against the probabilistic SSDF attack in CSS.From the perspective of attack detection,a malicious user detection algorithm based on Bayesian inference with sliding window reputation model is proposed,and the detected malicious users are forbidden from participating into the final data fusion.The proposed algorithm utilizes Bayesian inference to obtain the reputation model,and then fuses the information comes from multiple time windows to form the cumulative weighted reputation value.The reputation update scheme and the periodical evaluation mechanism are also presented.Simulation results show that the proposed algorithm achieves superior performance for identifying independent or collaborative probabilistic SSDF attackers without cognitive users' any prior information,and with low implementation complexity.(3)To defend against the massive SSDF attack in CSS.From the perspective of data fusion,an entropy-based weighted decision fusion secure CSS algorithm is proposed,where each cognitive user's reliability is evaluated based on the inconsistency property of received data at the FC within two consecutive sensing slots.By weighting the reported decisions of cognitive users,the proposed algorithm not only enhances the CSS performance,but also ensures the security of CSS under massive SSDF attack.(4)Considering the tradeoff optimization problem between CSS and energy efficiency under SSDF attack.Firstly,the data transmission time of each cognitive user is allocated proportional to their trust degree,then the energy efficiency constraint optimization mathematical model is establised.Secondly,to obtain the CR network's maximum energy efficiency,the corresponding spectrum duration and cognitive transmit power,an alternate iteration algorithm based on exhaustive search and fractional programming is proposed to iteratively solve the above parameters.Finally,in order to resist SSDF attack,a reliable user selection strategy by utilizing online learning algorithm is proposed.Simulation results show that,compared with the scheme that transmission time is allocated averagely,the proposed proportional allocation scheme achieves higher energy efficiency of the CR network,and realizes secure CSS under malicious attack.