Physical Layer Joint Authentication Technology Based On Channel And Radio Frequency Characteristics

Traditional authentication technology based on cryptography has been difficult to ensure the security of the communication system,while the physical layer authentication technology can use wireless channels and radio frequency fingerprints for authentication,which greatly improves security and reliability.At present,there are many schemes that use single feature parameters for physical layer authentication.Due to the influence of noise,environment,incomplete estimation,etc.,the extracted parameters have certain volatility,and it is inevitable that there are limitations.For higher detection performance,our paper proposes a joint authentication scheme based on Channel State Information(CSI)and Carrier Frequency Offset(CFO),which combines wireless channel characteristics and hardware radio frequency characteristics,which fully reflects the channel and hardware characteristics,improve the certification performance.The main content of this article is divided into three parts:1.Propose a joint authentication scheme based on hypothesis testing.The likelihood ratio test is used to construct test statistics for CSI and CFO respectively,and the probability distribution of the test statistics is derived.The detection threshold for authentication is given by the false alarm probability.Jointly judge the certification results of CSI and CFO,When one of the parameters is judged to be illegal,it is judged as an illegal user,thereby designing a more stringent detector.The simulation results show that compared with single-variable authentication,the proposed joint authentication scheme has better detection performance for illegal users2.It is difficult to determine the optimal detection threshold of the hypothesis testing scheme,so machine learning is introduced and a machine learning joint authentication scheme based on feature fusion is proposed.Feature fusion methods can be divided into data fusion and detection result fusion,and feature combination authentication schemes and ensemble learning authentication schemes are respectively proposed.The feature combination authentication scheme is first to splice the data of CSI and CFO,and then use the classic K-Nearest Neighbor(KNN)algorithm for classification.In order to have better classification performance and robustness,further use Convolutional Neural Network(CNN)algorithm.Aiming at the problems of data redundancy and conflicts that may occur in data fusion methods,an ensemble learning authentication scheme that uses ensemble algorithms to fuse the detection results of CSI and CFO is proposed.Combine Adaptive Boosting(Ada Boost)algorithm with KNN algorithm and CNN algorithm,and propose Ada Boost-KNN algorithm and Ada Boost-CNN algorithm.The combination of conventional Ada Boost algorithm and CNN will reduce the number of effective training samples.This paper uses transfer learning to solve this problem.Without sacrificing time complexity,the Ada Boost-CNN algorithm can effectively combine the characteristics of CSI and CFO to improve authentication performance.The simulation results show that the performance of the proposed joint authentication scheme is better than that of the single-parameter authentication scheme3.This paper uses the software radio platform USRP and the programming software Lab View Communication to build a set of OFDM-based signal transceiver system to collect CSI data and CFO data in the real environment.Use measured data to verify the algorithm performance and time complexity of the proposed joint authentication scheme.The results show that the existing single-parameter authentication scheme is the worst,the ensemble learning authentication scheme proposed in this paper has the best performance,and the feature combination authentication scheme is the second.Explain that the proposed joint certification scheme is feasible and practical...