Identify Based On Wireless Transmitters' Physical Layer Multi-inherent Imperfection

Wireless network is a very important information infrastructure,and it has been widely used in industries and daily life.Due to the openness of the wireless channel,wireless network faces physical layer information leakage during transmission and other security threats.Traditional cryptography security mechanism can make sure network layer and upper layer communication security,by encrypting information(keys)among communication entities.In recent years,Internet of things(Io T)and the Fifth Generation of Communications(5G)have been constantly emerging,and heterogeneous network structure has been changing the connection way.The mobility and distribution of network not only increase the difficulty of key distribution and management,but also resault in wireless channel's physical layer attacks.Therefore,security based on keys has been unable to meet the needs,and new protection mechanisms are demand.Physical layer security,as a complement to cryptography,mainly uses physical layer characteristics,like wireless channels or RF front-to-end defects,to make sure the security of wireless communication.Because the surrounding environment and production process are random,physical layer characteristics are difficult to forge or tamper.In this paper,physical layer defect or fingerprint characteristics is used to distinguish the signal emission source and identity of wireless devices,and prevents detect forgery attacks to improve network security.The main studies are as follows.Due to the feature between devices is small and affected by environment randomly,devices' endogenous features is hard to collect and analyses.In this thesis,multi-inherent features are considered.With the help of ARMA model,the time-frequency and phase-frequency features are considered,and the stability of the features' time sequence is also analyzed.Using the fitting parameters of ARMA,the temporary identity is built to source signal discrimination,and the false signal source's mission success rate is at least 95%.The long-term and nonlinear of physical layer characteristics is also considered to avert device damnification.The M-P neuron model and LSTM neural network is also used to analyze and model nonlinear physical layer characteristics.Based on the LSTM cell's gates weights matrix,the global long term identity and local short term identity are built to cater long time communication process.For a fast authentication and identity system,the authentication scheme is also needed.For different communication scenes,two identity schemes are proposed.One is for the global long term identity and local short term identity,which based on the identity sequence correlation coefficient;another is for the temporary identify,which using the signal correlation cause by channel coherence.Experimental result shows that the sequence correlation authentication scheme's discrimination can reach to 99%.In order to verify the identity model and the authentication scheme,a platform has been built that bases on IEEE 802.11 protocol with USRP B210.The final experimental shows that using the identity can be used in improving wireless network communication security.