Research On Active Physical Layer Key Generation Method For Quasi-static Environment

With the development of wireless communication and the popularization of wireless devices,a large amount of civil and military data is transmitted on wireless networks,so it is of significance to ensure the data security in wireless networks.Due to the limitations of wireless network topology,node computing power,and node power life,the traditional encryption method based on the key distribution center(KDC)is no longer suitable for low power consumption,decentralized wireless networks.The physical layer key generation(PLKG)uses wireless characteristics to generate keys between two communication ends.This technology no longer depends on the KDC,and has the characteristics of low overhead and high security.In practical application environment,the channel key generation method also faces some challenges,mainly including the low security,weakening of channel reciprocity caused by radio frequency fingerprints(RFF)and the poor randomness of keys under quasistatic channels.In order to solve the above problems,this thesis has carried out the following work:(1)This thesis analyzes the security of the existing PLKG algorithm based on backhaul,points out the security risks of the existing algorithm,and proposes a novel attack method,multiple-divide attack.Attackers use multiplication and division operation and inverse matrix to eliminate eavesdropping channel coefficients,so as to steal information.In this thesis,theoretical analysis and simulation are carried out in several scenarios,and the results show that the attack method has wide applicability and attack effectiveness.(2)In this thesis,an active channel key generation method based on antenna scheduling,antenna scheduling key generation(ASKG)is proposed.Two unrelated frequency bands are used for pilot transmission and backhaul to ensure that ASKG can resist multiple-divide attack,and the randomness brought by antenna scheduling makes the algorithm resist nearby attacks.ASKG improves reciprocity by backhaul and calculating channel eigenvalues,and generates consistent information between key negotiators.In the scenario of abundant antenna resources,key negotiators can improve the randomness of keys under quasi-static channels by randomly selecting antennas.Simulation results show that under multiple-divide attack and nearby attack,the information eavesdropped by eavesdroppers is irrelevant to the information obtained by key negotiators,and the cross-correlation coefficients remain near 0;Under different RFF strengths,the cross-correlation coefficients between key negotiators under ASKG algorithm are kept around 0.99;Under quasi-static channel,the channel information obtained by the key negotiator's two consecutive antenna scheduling is irrelevant,and the cross-correlation coefficient is kept around 0.15.(3)In this thesis,an active channel key generation method based on artificial random source,artificial randomness key generation(ARKG)is proposed.Pilot transmission and backhaul are carried out in two unrelated frequency bands,which ensures the security of ARKG under multiple-divide attack.Artificial random source is the information advantage of key negotiators over attackers,which ensures the security of the algorithm under nearby attack.At the same time,the introduction of artificial random source can also improve the randomness of key under quasi-static channel.ARKG eliminates the adverse effects of RFF on channel reciprocity by means of backhaul,matrix diagonalization function and matrix diagonal value extraction function,and generates consistent information between key negotiators.Simulation results show that under multiple-divide attack and nearby attack,the information eavesdropped by eavesdroppers is irrelevant to the information obtained by key negotiators,and the crosscorrelation coefficients remain near 0;Under different RFF strengths,the cross-correlation coefficients between key negotiators under ARKG algorithm are kept around 0.99;The channel information obtained by the key negotiator in the slow moving scene is highly random,and the autocorrelation coefficients remain near 0.