Analysis of Differential Phase Shift Quantum Key Distribution

We review the implementation of two QKD protocols (BB84 and B92) keeping in mind that their implementations do not easily satisfy the requirement of use of single photons. We argue that current models do not take into account issues raised by the Uncertainty Principle related to time-location and transmission characteristics of single photons. This indicates that security proofs of current implementations even after the fixes for the recent successful hacks are made will be hard to obtain.

💡 Research Summary

The paper provides a critical review of practical implementations of two well‑known quantum key distribution (QKD) protocols—BB84 and B92—when they are realized using differential phase‑shift (DPS) encoding. It begins by recalling the promise of QKD: exploiting quantum properties of photons to achieve unconditional security, a claim that rests on the assumption that each logical qubit is carried by a true single photon. In practice, commercial systems (e.g., those from MagiQ and idQuantique) employ very weak laser pulses with an average photon number far below one. Because the photon number in each pulse follows a Poisson distribution, many pulses contain two or more photons. This deviation from the single‑photon ideal opens the door to Photon‑Number‑Splitting (PNS) attacks, where an eavesdropper siphons off one photon while allowing the rest to continue to the legitimate receiver, thereby gaining information without introducing detectable disturbances.

The authors describe the DPS implementation in detail. Alice prepares a strong reference pulse and a weak signal pulse separated by a fixed time interval Δt. By applying a 0° or 180° phase shift to the signal pulse, she encodes logical 0 or 1. Bob’s interferometer recombines the two pulses; constructive interference (detected as a strong pulse) occurs only when his phase setting matches Alice’s, while destructive interference yields no detection. The paper enumerates the four possible combinations of Alice’s and Bob’s phase choices, illustrating each case with schematic diagrams and temporal waveforms. Successful key bits are retained only when Bob’s basis aligns with Alice’s, following the standard sifting procedure.

Implementation challenges are then examined. First, the lack of a reliable on‑demand single‑photon source forces designers to rely on attenuated coherent states, which inevitably produce multi‑photon events. Second, polarization drift in optical fibers makes phase encoding preferable, yet phase information becomes easier to extract when several photons travel together, weakening security. Third, background noise and fiber loss compel the use of higher signal powers, again increasing the multi‑photon probability. Fourth, the transmission distance is limited to roughly 100 km; beyond that, accumulated phase noise and reflections degrade the encoded information.

The paper connects these technical shortcomings to real‑world attacks. It cites the bright‑illumination blinding attack (Lydersen et al., 2010) and a full‑field perfect eavesdropper demonstration (Gerhardt et al., 2011), both of which successfully compromised commercial BB84/B92 systems. These attacks exploit the very assumptions—single‑photon operation, ideal detector behavior, and perfect basis alignment—that underpin most security proofs. Consequently, the authors argue that existing security proofs are not directly applicable to current hardware.

In the concluding section, the authors suggest that the Kak three‑stage protocol (Kak06) may offer a more robust alternative. Because Alice and Bob independently apply random polarization rotations to each qubit, even if multiple photons are present, an eavesdropper cannot reliably infer the secret key. The protocol also benefits from classical authentication mechanisms (trusted certificates) to thwart man‑in‑the‑middle attacks. The paper emphasizes that future QKD research must bridge the gap between theoretical security models and practical engineering constraints, focusing on genuine single‑photon sources, ultra‑low‑noise channels, and protocol designs that remain secure under realistic multi‑photon conditions.