Research On Transmission Distance Promotion And Security Issues Of Continuous-Variable Quantum Key Distribution

Quantume key distribution (QKD) is a crucial approach of information security, based on the principles of quantum physics. It offers unconditional security in the encrypting and transmitting of information theoretically, establishing it as a quantum information technology with the best practicability. There are two approaches to implement QKD, namely discrete variable (DV) and continuous variable (CV). DV QKD protocols, based on the operations of single photons, although having made significant achievements, is difficult to accomplish due to the single photon source and single photon detectors. Meanwhile, CV QKD protocols encode information of the quadratures of the optical field. Coherent state CV QKD protocols use semiconductor laser and optical modulators for state preparation and balanced homodyne detector constructed by photodiodes for quadrature measurement, offering a unique advantage in reliability and infrastructure cost over DV QKD protocols.While CV QKD strides it way into practical application, it DOES have unresolved problems. Since the physical devices are imperfect and do not fully fit the delicate theoretical models applied in security analysis, there are still hidder dangers in practical CV QKD. On the other side, it requires longer transmission distance and higher secret key rate for real secure communications. Motivated by this, transmission distance promotion techniques and security issues of CV QKD are addressed in this thesis, namely:1. Improving the maximal transmission distance CV QKD with noisy coherent states using a noiseless linear amplifier (NLA). To promote the secret key rate and the transmission distance of CV QKD with source noise, it is proposed to probabilistically amplify the signal state with an NLA before homodyne detection. Utilizing the covariance matrix transformation of the noiseless linear amplifying operation, the expression of secret key rate is derived under collective entangling cloner attacks, assuming that the source noise is trusted. Numerical simulations show that NLA can compensate the detrimental effect with an enhancement of the secret key rate and noise resistance, despite that the enhancement is not as much as that of an ideal coherent state.2. Security of two-way CV QKD with source noise is analyzed. It is analyzed whether the source noise is trusted or not respectively. When the source noise is untrusted, it is treated as part of the excess noise in the quantum channel. Once trusted, the source noise is modeled via a two-mode squeezed (EPR) state through a beam splitter. Numerical simulations of both cases show that deteriorates the security of two-way CV QKD significantly. In comparison, the trusted model of source noise offers a better secret key rate performance over the untrusted model. The reason lies in that under the trusted model, the eavesdropper is not overestimated, making it more appropriate. What’s more, it is shown numerically that two-way CV QKD is still advantageous over its one-way counterparts in the presence of source noise.3. A novel model of source noise based on mixed entangled state is established. The motivation of the proposed model lies in that the impact of the source noise should be degrading the accuracy of the key information measurement, rather than modifying the key information directly. The proposed model generalizes the EPR-based entanglement to two-mode squeezed thermal state, and simulates the behavior of source noise with intrinsic property of non-maximal entanglement rather than extrinsic modification. What distinguishes it from the beam splitter model is that the signal, not the information, is corrupted by the source noise. Moreover, the equivalence between the proposed scheme and the practical setup with Gaussian source noise is proved theoretically and verified numerically.4. The effects of the coarse grainess and the finite range of the detectors in CV QKD are analyzed. Previous works on detectors are mostly focused on the quantum efficiency and the electronical noise, while the coarse grainess and finite range are left aside. In this thesis, both features are taken into consideration in the security analysis and modeled. The results show that while coarse graining introduce a fixed bias to the outputs of parameter estimation, the impact of the finite range is more severe and more uncontrollable, which may lead to underestimation of the secret key rate and even misjudgment of the availability. To eliminate the destructive effect due to the finite range, tuning the modulation variance is a method available.