Research On Cooperative Spectrum Sensing For Uncertain Networks Based On Dempster-Shafer Theory

Due to the static spectrum licensing and allocation strategies and greatly increased demand for spectrum resources,currently spectrum regulation has resulted in extreme scarcity of available spectrum,while plenty of radio frequencies are actually unused in time and space dimensions.Motivated by the need for flexible management and efficient utilization of spectrum resources,cognitive radio is brought up at the historic moment and reliable and efficient spectrum sensing is an important component of cognitive radio.Due to a lot of disadvantages of the single-user spectrum detection,cooperative spectrum sensing which makes benefit from the spatial diversity has been extensively studied.But cooperative spectrum sensing is vulnerable to the attacks initiated by malicious cognitive radio users.Therefore,effective secure mechanisms are essentially required in a hostile wireless environment which has malicious cognitive radio users.The main objective of this paper is to ensure the good cooperative spectrum sensing performance in an uncertain wireless network environment.So the paper has studied cooperative spectrum sensing for uncertain network environments based on Dempster-Shafer(D-S)theory.And cooperative spectrum sensing has been implemented and tested and analyzed based on the Universal Software Radio Peripheral(USRP)platform.The main work of the paper as follows:(1)It established a system model of cooperative sensing method based on D-S theory.It briefly analyzed the basic principles of D-S theory and three different attack patterns.(2)A cooperative sensing method of D-S theory based on signal-noise ratio(SNR)and trustworthiness degree is proposed.Current reliability and historical reputation are evaluated for sensing information of cognitive radio users and normalized as trustworthiness degree of each user.According to trustworthiness degree,it determined whether cognitive radio users are malicious user.Taking into account different SNR,basic probability assignments(BPA)of cognitive radio users are adjusted depending on trustworthiness degree and SNR after excluding malicious user.Finally,we got final decision according to the combination rule of D-S theory.We conducted simulation experiment in several different network environments,such as environment without malicious user and environment with different attack patterns and different number of malicious users.The simulation results can be seen that the method can achieve better sensing performance.(3)A cooperative sensing method based on projection approximation and improved D-S theory is proposed.We conducted pretreatment by the projection approximation on sensing information of cognitive radio users.So it can reduce the data that cognitive radio user submitted to the fusion centers and achieve reducing the bandwidth overhead.Sensing information of cognitive radio users are excluded and adjusted depending on trustworthiness degree and SNR.Finally,we got final decision according to the combination rule of improved D-S theory.The simulation results showed that the method can maintain good sensing performance on the basis of reducing bandwidth overhead of system.(4)Cooperative sensing method was realized and tested based on USRP platform.Cognitive radio users conducted local spectrum sensing by using energy detection.The sensing information obtained by energy detection is sent to the fusion center through Time Division Multiple Access(TDMA).The fusion center received sensing data and got a final decision based on the sensing data.