arXiv:2610.32923
One-sided-DIQKD using any pure bipartite entangled state secure against all attacks
Bohnishikha Ghosh, Chithra Raj, Shubhayan Sarkar
Abstract
Quantum Key Distribution (QKD) allows spatially separated parties to establish a secret key among themselves based on quantum principles. While QKD systems are being deployed worldwide, their reliance on trusted measurements makes them vulnerable. Moreover, existing protocols have largely focussed on the qubit cases, and rely on the maximally entangled state, a costly resource to generate. Here, we consider the one-sided-device-independent (1SDI) QKD protocols, where one of the measurement devices is trusted while the other is untrusted. The resource in this scenario is quantum steering, an asymmetric form of quantum nonlocality. This is inspired by a real-world scenario, where the trusted party can be considered to be a bank, or any large organisation that has sufficient resources to protect its systems and validate them, while the users are vulnerable to attacks. We then demonstrate that a secure key can be obtained using any pure bipartite entangled state in this setup. Considering realistic detectors and sources based on current technology, we demonstrate that even states that are almost separable can be used to extract a secure key. Moreover, for every pure entangled bipartite state, we find the critical detection efficiencies and visibility to obtain a secure key.
AI Summary
The paper asks whether every pure bipartite entangled state can generate secret key in one-sided-device-independent QKD, with an untrusted source and Alice’s measurement device, but trusted measurements at Bob. It considers a two-setting protocol using the Schmidt basis and its Fourier-conjugate basis, extracting key from Bob’s Schmidt-basis outcomes. For an ideal state $|\psi\rangle=\sum_i\alpha_i|ii\rangle$, existing one-sided self-testing results certify that Bob is decoupled from Eve. Together with perfect Schmidt-basis correlations, this gives an asymptotic key-rate bound $r_\infty\geq S(\rho_B)=-\sum_i\alpha_i^2\log_2\alpha_i^2$, positive for every pure entangled state. For depolarizing noise of visibility $\nu$ and Alice’s detection efficiency $\eta$, no-click events are retained explicitly. An entropic-uncertainty relation yields $r_\infty\geq\eta[N_Z+\log_2d-C_Z-C_X]-N_Z$, where $N_Z$ is Bob’s Schmidt-basis outcome entropy and $C_Z,C_X$ are conditional entropies for detected events. The corresponding sufficient efficiency threshold is $\eta_{\rm crit}=N_Z/[N_Z+\log_2d-C_Z-C_X]$, when attainable. For noiseless maximally entangled states this is $1/2$. A parameterized state family illustrates decreasing certified rates and increasing efficiency requirements as entanglement or visibility decreases. Coherent-attack security is connected to existing entropy-accumulation methods under their required sequential conditions; the calculations presented are asymptotic. Bob’s losses are discarded under trusted-device fair sampling. The paper identifies tighter entropy bounds as an open direction and notes that its noisy-case method does not extend directly to fully device-independent QKD because it requires characterized rank-one projective measurements.
Error found: The reported critical efficiencies are sufficient certification thresholds obtained from a particular key-rate lower bound, not established optimal or necessary efficiencies for generating secret key. The main-text noiseless efficiency corollary has the wrong sign for the Fourier-distribution entropy, yielding approximately 0.2141 instead of 0.6278 for Schmidt probabilities (0.9, 0.1). The protocol specifies Alice-to-Bob-only reconciliation, but its Bob-key rate with reconciliation cost H(B|A) requires reconciliation information from Bob to Alice.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 6+3 pages, comments are welcome:)