Practical Security Analysis And Performance Improvement On Measurement-Device-Independent Quantum Key Distribution

With the development of supercomputers and quantum computers,software encryption can be broken faster and faster,making decryption possible.In early 1994,Peter Shor demonstrated that the RSA encryption algorithm,designed based on computational complexity,can be decrypted by quantum computers in polynomial time.In 2019,the quantum advantages demonstrated by Google’s experiment further highlights the threat posed by the development of quantum technology to current communication security.Quantum Key Distribution(QKD)provides a new solution for secure communication.Based on the basic principles of quantum mechanics,QKD can be used to safely share a string of random numbers that can be used as keys among legitimate users in different locations.Since the first QKD protocol(the BB84 protocol)was formally proposed,researchers have been exploring practical applications in this field,with security and practicability being two important aspects.Measurementdevice-independent(MDI)QKD protocol is naturally immune to all side-channel attacks against measurement devices.It achieves a good balance between security and practicability,and has become a research hotspot in the field of QKD.Although the MDI-QKD protocol addresses all security vulnerabilities in the measurement side,its stringent requirements on the source sides are extremely challenge.As research deepens,these security assumptions are difficult to meet in real-world systems.For instance,real-world MDI-QKD systems are vulnerable to state preparation flaws,potential side-channel vulnerabilities,and active attacks by eavesdroppers at the source sides.Addressing these security vulnerabilities has become an urgent research problem.Currently,the research on the source-side security of MDI-QKD system is still in the preliminary exploration stage,and there is a need to improve the actual security and performance of the system.To address these issues,our paper focuses on the following research:1.To address the state preparation flaws that are commonly existed in practical MDI-QKD system,we propose a four-intensity decoy state loss-tolerant MDI-QKD protocol.In this protocol,an additional intensity(signal state)modulated in Z basis is introduced specifically for key generation in state preparation stage,while other intensities(decoy states and vacuum state)modulated both in Z and X bases are used for parameter estimation.In addition,collective constraint and joint estimation are adopted in parameter estimation stage to reduce the influence of statistical fluctuation.Simulation results show that the new protocol can hugely improve the secure key rate and transmission distance.Furthermore,in view of the qubit assumption of the aforementioned loss-tolerant protocol,we propose an MDI-QKD protocol that can use any insecure sources based on the reference technique.This protocol further considers the side channels information leakage of the transmitted signal in noncoding dimensions,such as mode dependence effect and pulse correlation,and removes the qubit assumption of loss-tolerant protocol.This research has a guiding significance for the practical security analysis and design of high-speed MDI-QKD system.2.To protect against potential information leakage from transmitters,we propose an improved finite-key security analysis of MDI-QKD against Trojan-horse attack.This scheme utilizes the Cauchy-Schwartz inequality to incorporate potential information leakage from bit/base and intensity coding settings into security analysis,and the central inequality of independent variables for finite-key analysis.Simulation results show that the both decoy-state MDI-QKD and loss-tolerant MDI-QKD can achieve long-distance key distribution by increasing the isolation degree of the light sources,which has important guidance on the protection for transmitters of MDI-QKD systems.Finally,we strictly prove the advantage of CauchySchwartz inequality over trace distance criterion in parameter estimation.3.Finite-size effect can introduce statistical fluctuation that has previously been studied with poor results or low security.To address this issue,we propose an improved statistical fluctuation analysis for MDI-QKD protocol.This protocol employs the inverse and symmetric forms of the Chernoff bound,making it more suitable for real-world systems.Simulation results show that it has same security level compared to the Chernoff-Hoeffding analysis method and achieves performance similar to that of Gaussian analysis method.Lastly,to solve the parameter optimization problem of MDI-QKD,we propose a QKD optimal parameter prediction model based on random forest.It can predict the modulation parameters quickly and accurately according to the current system conditions,greatly improving the optimization speed and providing a solution for the real-time parameter optimization largescale QKD networks and low-power mobile QKD terminals in the furture.