Improving performance of decoy-state free-space QKD using information on fluctuating transmittance in turbulent channel
Wenyuan Wang,Feihu Xu,H. Lo
2017-01-01
Abstract:Quantum key distribution (QKD) allows two parties to share a random secret key with unconditional security. There has been increasing interest in implementing QKD through freespace channels, and over the past decade multiple demonstrations for free-space QKD have been made, such as [1–4], through groundground and even ground-satellite channels. A major characteristic of a free-space channel is its time-dependent transmittance fluctuations due to atmospheric turbulence. This fluctuation can be modeled as a probability distribution of transmission coefficient (PDTC) p(η), a function of the transmittance η (the level of turbulence is characterized by the scintillation index, σ, as a parameter of the PDTC) In previous literature discussing free-space QKD, such as [1, 2], the time variance of the channel is ignored, i.e. the secure key rate is calculated based on the time-average of channel transmittance η0 only. Having knowledge of the PDTC, however, Vallone et al. proposed a method named adaptive real-time selection (ARTS) [5] that acquires information about real-time transmittance fluctuation due to turbulence, and makes use of this information to perform post-selection and improve the key rate. The proposed method is to use a classical probe signal (a strong laser beam) alongside the quantum channel. Bob can gain information of the periods of high transmittance by observing the classical signal. Combined with a threshold, he can post-select only those signals received during high transmittance periods, thus increasing the average transmittance and the secure key rate among post-selected signals. However, as one increases the threshold, the number of signals also decreases. Therefore, an optimization of post-selection threshold is necessary in order to acquire the maximal key rate. This approach can significantly increase the secure key rate under high turbulence and high loss. FIG. 1: Comparison of ARTS vs non-ARTS decoystate BB84 under different levels of turbulence, using σ = 0.3, 0.6, 0.9 (0.3 for weak turbulence, while 0.9 for relatively strong turbulence) Rate is calculated using experimental parameters from [2]. Here we have already chosen an optimal threshold. We see that the improvement in rate from using ARTS increases with the level of turbulence, and has a significant improvement over static model for high-loss region. For σ = 0.9, for the same loss=29dB, with ARTS we can achieve as much as 278% increase over the case without ARTS. Also, for a minimum rate of R = 10−7, with ARTS we can support a 7.5dB increase in maximum tolerated loss.