Research On Key Technologies Of Physical Layer Security Transmission For 5G

With the popularization and application of the fifth generation(5G)mobile communications and the explosive growth of mobile smart terminals,wireless communications security has become more and more important.Owing to the hugeness of the transmission node,the openness and broadcast characteristic of wireless transmission,it is extremely vulnerable to be attacked by illegal users during data transmission.Therefore,the security issues in the fifth generation and beyond fifth generation(B5G)mobile communications has become the focus of current research.Traditional cryptography-based security highly relies on high deciphering complexity,and it is difficult to respond to complex environments in time.Physical layer security has the advantages of rapid response,high efficiency and reliability.Based on this,this thesis studies the key technologies of physical layer secure transmission for the 5G system.Firstly,this thesis researches the downlink transmission scheme in massive MIMO,and sorts out the basic principles of the classic precoding technology based on the zero-forcing criterion,and then studies the Kalman iterative precoding technology based on Kalman channel prediction.The numerical simulation results verified that the proposed precoding technology based on Kalman iteration has improved performance compared with the traditional precoding technology based on zero-forcing criteria;Meanwhile,the technology based on Kalman channel prediction can improve the precoding transmission performance in the presence of channel feedback delay scenarios.The thesis studies the physical layer authentication technology based on radio frequency fingerprints.In detail,the RBF neural network is used to model and analyze the radio frequency fingerprint recognition algorithm based on signal envelope characteristics and signal bispectral transformation,and the detailed procedures of the two schemes are given.The numerical simulation results verified that the radio frequency fingerprint identification algorithm based on signal bispectral transformation uses more information dimensions than the radio frequency fingerprint identification algorithm based on signal envelope characteristics,so its authentication performance is better.This thesis studies the physical layer authentication technology based on signal watermarking.The physical layer authentication scheme based on phase rotation embedding,additive embedding of watermarks based on the position of all constellations is sorted out,and proposes a physical layer authentication scheme based on the watermark embedded in part of the constellation position.And the two schemes are digitally simulated around 16QAM rectangular constellation mapping and circular constellation mapping.The numerical simulation results verified that in the case of the same watermark embedding power coefficient,the authentication scheme based on the phase rotation embedding the watermark shows better performance,because the algorithm based on additive embedding of the watermark has disturbed the amplitude and phase of the carrier information.When the bit error rate of information is equivalent,the scheme based on additive watermarking has an advantage over the scheme based on phase rotation in terms of authentication performance,but its watermark transmission power is higher.On the other hand,when the original watermark bits and the watermark power coefficient are fixed,in the scheme of embedding the watermark based on the partial position of the circular constellation diagram,the watermark position that can be embedded is more than that based on the partial position of the rectangular constellation diagram,so the probability of recovering the original watermark bits is higher,and the physical layer authentication performance is better.This thesis studies the physical layer authentication technology based on channel characteristics.Under the background of the physical layer authentication technology based on the correlation of channel frequency response,the theoretical analysis of the authentication threshold based on the constant false alarm criterion and bayes criterion under the test statistics of Manhattan distance metric and Euclidean distance metric is given.Then,a novel physical layer authentication scheme based on multi-antenna channel prediction is proposed,the numerical simulation is carried out under the CDL-C channel environment in the TR38.901 protocol.The numerical simulation results verified that in terms of authentication rate and false alarm rate indicators,the proposed authentication scheme has better performance in the fast time-varying channel environment.At last,this thesis summarizes the research results and clarifies the further research direction.