[Paper Review] Time-frequency bases for BB84 protocol

Susanne F. Yelin, Bing C. Wang|ArXiv.org|Sep 12, 2003

Bluetooth and Wireless Communication Technologies3 citations

TL;DR

This paper proposes a time-frequency basis for the BB84 quantum key distribution protocol using transform-limited pulses, where qubits are encoded in time or frequency bins instead of polarization or phase. The scheme enables high-dimensional quNdit systems and achieves a 5:1 ratio of eavesdropper information gain to introduced error—superior to the traditional 2:1 ratio—enhancing security in long-distance fiber networks.

ABSTRACT

The bases traditionally used for quantum key distribution (QKD) are a 0 or pi/2 polarization or alternatively a 0 or pi/2 phase measured by interferometry. We introduce a new set of bases, i.e. pulses sent in either a frequency or time basis if the pulses are assumed to be transform limited. In addition it is discussed how this scheme can be easily generalized from a binary to an N-dimensional system, i.e., to ``quNdits.'' Optimal pulse distribution and the chances for eavesdropping are discussed.

Motivation & Objective

• To develop a new QKD scheme that overcomes limitations of polarization and phase-based encoding in optical fiber.
• To leverage the time-frequency duality of single photons to enable high-dimensional quNdit systems.
• To analyze the security of the scheme against eavesdropping, particularly in terms of information gain and introduced error.
• To demonstrate experimental feasibility using standard fiber optic components.
• To quantify the trade-off between fidelity, eavesdropping success, and error introduction in the time-frequency basis scheme.

Proposed method

• Encoding qubits in time bins (e.g., t₀, t₁) or frequency bins (e.g., ν₀, ν₁) using transform-limited Gaussian pulses.
• Using the uncertainty relation ΔT·Δν ≥ 1 to ensure distinguishability in the correct basis and indistinguishability in the wrong basis.
• Modeling detection fidelity using the error function Φ(x), where x = δt/(2√2 Δt), to quantify correct measurement probability.
• Deriving the eavesdropper’s success probability P and introduced error E using Gaussian integrals and cumulative distribution functions.
• Generalizing the scheme to N-dimensional quNdit systems by using N time or frequency bins, maintaining pulse area normalization.
• Evaluating the ratio R = E/P as a figure of merit to compare eavesdropping efficiency across different schemes.

Experimental results

Research questions

• RQ1 How can time and frequency bases be used to encode qubits in a way that maintains quantum indistinguishability across bases?
• RQ2 What is the optimal pulse width and separation to achieve high detection fidelity while minimizing eavesdropping success?
• RQ3 How does the time-frequency basis scheme compare to traditional polarization-based BB84 in terms of eavesdropping resistance?
• RQ4 Can the scheme be generalized to quNdit systems with N > 2, and what are the implications for security?
• RQ5 How does increasing the number of bases affect the eavesdropper’s ability to gain information while introducing detectable errors?

Key findings

• For a fidelity of 99%, the eavesdropper’s information gain (P) and introduced error (E) yield a ratio R = E/P ≈ 5, significantly better than the traditional 2:1 ratio.
• At 90% fidelity, the ratio R drops to slightly above 2, indicating that eavesdropping becomes less efficient as fidelity decreases.
• The eavesdropper’s success probability P is approximated by (y/√π)(1 + e^(-x²/z²)/z), where y is the spectral width and z = ΔT/Δt.
• The introduced error E is approximated by (y/√π)(e^(-4x²)(1 + Φ(x))/2 + e^(-x²/z²)/z), quantifying detectability.
• The ratio R increases linearly with the number of bases N, meaning higher-dimensional systems improve eavesdropping efficiency, partially offsetting the advantage of quNdit encoding.
• The scheme is experimentally realizable using standard fiber optic components, including Mach-Zehnder interferometers and optical filters, enabling deployment in wavelength-division multiplexed networks.