Finite-Key Analysis of Quantum Key Distribution with Characterized Devices Using Entropy Accumulation

The Entropy Accumulation Theorem (EAT) was introduced to significantly improve the finite-size rates for device-independent quantum information processing tasks such as device-independent quantum key distribution (QKD). A natural question would be whether it also improves the rates for device-dependent QKD. In this work, we provide an affirmative answer to this question. We present new tools for applying the EAT in the device-dependent setting. We present sufficient conditions for the Markov chain conditions to hold as well as general algorithms for constructing the needed min-tradeoff function. Utilizing Dupuis’ recent privacy amplification without smoothing result, we improve the key rate by optimizing the sandwiched Rényi entropy directly rather than considering the traditional smooth min-entropy. We exemplify these new tools by considering several examples including the BB84 protocol with the qubit-based version and with a realistic parametric downconversion source, the six-state four-state protocol and a high-dimensional analog of the BB84 protocol.

💡 Research Summary

This paper investigates how the Entropy Accumulation Theorem (EAT), originally devised to improve finite‑size key rates for device‑independent quantum key distribution (DI‑QKD), can be leveraged to obtain tighter finite‑key bounds for the far more practical device‑dependent QKD (DD‑QKD) scenario. The authors identify two major technical obstacles that have prevented a direct application of the original EAT to DD‑QKD: (i) ensuring that the protocol satisfies the Markov‑chain conditions required by the theorem, and (ii) constructing the min‑tradeoff function that lower‑bounds the conditional min‑entropy per round.

To address (i), they prove a sufficient condition based on the structure of the public announcements. If the announcements are generated by POVMs whose Kraus operators are block‑diagonal in a fixed basis, then the side‑information revealed to an eavesdropper obeys the required Markov property. This block‑diagonal structure is naturally fulfilled by many realistic QKD implementations, including those that announce measurement outcomes, error‑correction syndromes, or basis choices. Consequently, a broad class of entanglement‑based DD‑QKD protocols—those in which all quantum signals are exchanged before any classical communication—are shown to meet the EAT’s prerequisites without needing artificial random seeding of all side information.

For (ii), the paper introduces two complementary algorithms for constructing near‑optimal min‑tradeoff functions. The first algorithm mirrors the numerical approach used for asymptotic key‑rate optimization: it solves a convex program that maximizes the asymptotic key rate under the given device model, and then lifts the solution to a finite‑size bound. The second algorithm improves upon the first by incorporating second‑order terms from the original EAT via Fenchel duality. This dual formulation reduces the dimensionality of the optimization problem and provides a cleaner handling of classical post‑processing steps. Crucially, the authors exploit Dupuis’s recent “privacy amplification without smoothing” result, which allows the security analysis to be expressed directly in terms of the sandwiched Rényi entropy rather than the traditional smooth min‑entropy. By optimizing the sandwiched Rényi entropy, they obtain strictly tighter key‑rate formulas than those derived from smooth‑entropy bounds.

The theoretical framework is validated on four representative protocols: (1) the standard qubit‑based BB84, (2) an entanglement‑based six‑state “four‑state” protocol, (3) a high‑dimensional analogue of BB84 using two mutually unbiased bases (MUBs), and (4) an entanglement‑based BB84 implemented with a realistic spontaneous parametric down‑conversion (SPDC) source in a lossy, noisy channel. For each case, the authors perform numerical simulations that incorporate realistic device parameters (detector efficiencies, dark‑count rates, channel loss, etc.). The results demonstrate that the EAT‑based finite‑key rates consistently outperform those obtained via the post‑selection technique, especially in high‑dimensional settings where the post‑selection method suffers from an exponential penalty in the signal dimension. In the SPDC‑based BB84 example, the method successfully handles multi‑photon contributions and background noise, yielding positive key rates at distances comparable to current experimental demonstrations.

The discussion section acknowledges that the current work is limited to entanglement‑based protocols. Extending the approach to prepare‑and‑measure schemes would require integrating the source‑replacement technique while preserving the Markov property—a non‑trivial task because the generalized EAT used in prior work assumes a strictly sequential implementation of rounds, which is often unrealistic. The authors suggest that future research should aim at relaxing this sequentiality requirement and at developing systematic ways to embed source‑replacement into the original EAT framework.

In summary, the paper makes three key contributions: (1) a clear sufficient condition for the Markov‑chain requirement in device‑dependent QKD, (2) two practical algorithms for constructing min‑tradeoff functions, and (3) a security proof that directly optimizes sandwiched Rényi entropy, leading to superior finite‑key rates across a variety of protocols. These advances bring the theoretical performance of DD‑QKD much closer to its asymptotic limits and provide concrete tools for practitioners seeking to certify real‑world QKD systems with finite resources.