Research On Theoretical Limit And Quantization Methods In Secret Key Generation Based On Characteristics Of Wireless Channel

Secure transmission of private message over an open channel is a critical issue in wirelesscommunication due to the broadcast nature of wireless channel. Cryptography is one of theefficient tools that protect the secret information from being eavesdropped. The widely usedencryption algorithms are based on unproven mathematical problems or computation complexity,which can only be computationally secure. The wireless channel is of reciprocity, temporalvariability and spatial variability. Secret key generation based on the property of the wirelesschannel, combined with the “one time pad” method can realize the unconditional security. Thus,undoubtedly, the research of the protocol design, theoretical limits of secret key rate andquantization methods are of great theoretical and practical importance.1. With the characteristics of the radio propagation, this paper introduces the basic componentsand process flow of the secret key generation systems, the time flow of the probing system isalso designed. By exploiting the reciprocity of the phase information, a secret key generationprotocol is proposed based on multiple carrier waves. On the basis of performance derivation,the simulation results show that the proposed method can provide high secret key rate, which isin direct proportion to the number of orthogonal carrier waves under abundant system resource.2. With the case of an eavesdropper with access to side information and the imperfectreciprocal channel between the legitimate users, the channel correlation coefficient betweenlegitimate users is introduced to the existing model of the wiretap channel, and the correspondingsecret key rate is deduced. Considering the fact that the legitimate users always utilize theenvelope and phase information to generate secret keys in actual secret key agreement schemes,secret key capacity of the envelope and phase are deduced in detail respectively, and two usefulconclusions are also obtained: firstly, correlation exists between the secret keys generated by theenvelope and phase of received signals, secondly, the secret key generation based on theenvelope and phase of received signals cannot utilize the abundant information of wirelesschannel.3. According to the characteristics of the key generation, the classical quantization methodssuch as equiprobable quantization, uniform quantization and MMSE (Minimum Mean SquareError) quantization and their application on secret key generation are researched. By regardingthe BER (Bit Error Rate) as the performance indicator, the final length of the secret key is alsoconsidered as another indicator. In this case, the performance of these algorithms is comparedand analyzed. The simulation results show that the performance of the uniform quantization andMMSE quantization are close to each other, and the performance of equiprobable quantization is superior to the other two algorithms.4. Based on the fact that the quantization parameters between legitimate users have strongcorrelation, the idea of information exchange is introduced to the course of quantization. Firstly,a great emphasis is placed on the double-threshold quantization algorithm which interact thelocation information of the samples. The concept of optimized quantization factor is put forward,and the optimized factors are obtained by optimization method. The simulation results show thatthe performance of the algorithm is improved obviously by taking the optimized factors. Next,The CQA (channel quantization adaption) quantization algorithm which relies on the exchangeof quantization errors is studied thoroughly. The final length of secret key generated bylegitimate users is deduced, and the optimal number of quantization bits is also obtained. Finally,in view of the case that the legitimate users wish to further reduce the bit error rate afterquantization, a post-quantization treatment algorithm based on parity-check is presented, and itseffectiveness is validated by typical information reconciliation algorithms. The simulation resultsshow that the proposed algorithm can further improve the performance of the secret keygeneration system, especially in the case of high original bit error rate.