Concise security bounds for practical decoy-state quantum key distribution

Due to its ability to tolerate high channel loss, decoy-state quantum key distribution (QKD) has been one of the main focuses within the QKD community. Notably, several experimental groups have demonstrated that it is secure and feasible under real-world conditions. Crucially, however, the security and feasibility claims made by most of these experiments were obtained under the assumption that the eavesdropper is restricted to particular types of attacks or that the finite-key effects are neglected. Unfortunately, such assumptions are not possible to guarantee in practice. In this work, we provide concise and tight finite-key security bounds for practical decoy-state QKD that are valid against general attacks.

Figure 1

Secret key rate vs fiber length (dedicated fiber). Numerically optimized secret key rates (in logarithmic scale) are obtained for a fixed postprocessing block size $n_{\mathsf{X}}={10}^s$ with $s=4,5,...,9$ (from left to right). The dashed curve corresponds to the asymptotic secret key rate, i.e., in the limit of infinitely large keys; however, here we still assume that the number of intensity levels is 3. The number of laser pulses sent by Alice can be approximated with the secret key rate and the block size, i.e., $N\le n_{\mathsf{X}}/R$.

Figure 2

Secret key rate vs fiber length (DWDM). We consider a (4+1) DWDM channel architecture [35] that puts four classical channels and one quantum channel into an optical fiber. In the simulation, we take that each classical channel has a power of $-34$ dBm at the receiver [38]. Numerically optimized secret key rates (in logarithmic scale) are obtained for a fixed postprocessing block size $n_{\mathsf{X}}={10}^s$ with $s=4,5,...,9$ (from left to right). The dashed curve corresponds to the asymptotic secret key rate, i.e., in the limit of infinitely long keys.