Research Of The Random Resource On The Security Of Quantum Information

Random numbers(RNs) refer to a set of data which have unpredictable property. The applications of RNs in cryptography are very common, for example in the process of key generation and distribution of classic encryption algorithms, and the preparation and measurement of quantum states in quantum key distribution protocol-BB84. The randomness of RNs significantly affects the safety of these encryption algorithms and protocols.Generation mechanism of RNs is called a random number generator(RNG). RNs generated by the classical algorithms rely on the computational complexity of the algorithms. Since the algorithm becomes deterministic once initialized, its RNs do not have the complete upredictability, thus these RNs are called pseudo-random numbers. On the other hand, the RNs generated based on random physical processes can meet the unpredictable principle, so these RNs are called true-random numbers. This thesis focuses on the mechanism of producing true-random numbers. Besides, according to whether the proportion of ’0’ and ’1’ in the RN sequence is the same or not, the RNs can be clasified into balanced random numbers(bRNs) and unbalanced random numbers(uRNs). uRNs are applied in the biased-BB84protocol, optimal basis selection in long-distance transmission BB84protocol, etc. At present, researchers have proposed many physical RNGs, which can only generate bRNs. In this thesis we propose an unbalanced quantum RNG(UQRNG), which can generate true-random numbers with arbitrary proportion of "0" and "1"This dissertation mainly completed the following work:We first introduced the research status of the current physical RNGs, including the generation mechanism, generation rate, system robustness and so on. Then according to the practical demand, we propose an UQRNG based on quantum mechanics. The proposed RNG can generate RNs with any desired proportion. And we give its experimental implementation, data generation rate possible optimization methods.Then we studied the randomness of the RNs generated from experimental system. For the generated sequences that the proportion of ’0’ and ’1’ is the same, we evaluated their randomness through using the National Institute of Standards and Technology(NIST) test suits and Three-Standard-Deviations test standards. As for the sequences that the proportion of ’0’ and ’1’ is different, since there exist no universal testing standards for uRNs at present, we assessed their randomness of these sequences through comparision on the bias, autocorrelation, and minimum entropy of the sequence with some reference unbanced sequences that obey "category uniform distribution". These reference sequences are converted from balanced numbers, which are generated by a commercial physical random number chip-Quantis through using some algorithm. These balanced RNs from Quantis are transmitted into personal computer(PC) through FPGA and USB interface of a circuit board.