# Can QEC Codes Make Underwater QKD Viable?

A 4.5 dB signal-to-noise ratio gain at an eleven percent [error threshold](https://quantumintel.tech/glossary/error-threshold) — that is the headline result from new modelling of four-qubit Calderbank-Shor-Steane (CSS) quantum error correction codes applied to underwater quantum key distribution. The work, authored by Juliette Florin, Nicolas Le Josse, Arnaud Coatanhay, and Gilles Burel and posted to arXiv (2609.18666), demonstrates that a specific variant of the CSS code — the discard version — outperforms the standard CSS implementation, which achieves only 3 dB at the same security threshold. That 1.5 dB gap is not cosmetic: in severely attenuating underwater optical channels, it is the difference between a link that closes and one that does not.

The simulations target vertical BB84 QKD protocols and explicitly model photon loss, geometric spreading, and solar noise in Jerlov water type III conditions — among the most turbid, optically challenging water classifications. At this performance point, the researchers identify operational depth windows where secure key generation was previously considered infeasible. The trade-off is throughput: adding QEC complexity reduces achievable data rates, and the paper shows that reducing photon arrival probability by factors of ten produces significant drops in key generation rate across all tested configurations.

For operators of subsea cable networks and offshore energy infrastructure, this is consequential. QKD's security guarantee — eavesdropping is physically detectable — is compelling for high-value links, but photon loss in water has historically capped its practical range. This work quantifies a concrete path to extending that range using established QEC techniques, without exotic hardware.

---

## What the CSS Discard Code Actually Does

Standard Calderbank-Shor-Steane codes correct errors by encoding quantum information redundantly across multiple physical qubits, allowing the receiver to detect and fix bit-flip and phase-flip errors without measuring the encoded state directly. The four-qubit variant studied here is compact by QEC standards — most surface code implementations for fault-tolerant computing target far larger qubit counts — but the underwater QKD context favors lightweight codes because photon-based channels cannot simply add more qubits the way a superconducting processor can.

The discard variant introduces an additional step: measurement outcomes that fall below a confidence threshold are discarded rather than corrected. This sacrifices some raw key material but improves the signal-to-noise ratio of the retained bits. At the eleven percent security threshold modelled in the study, this strategy yields 4.5 dB versus the 3 dB from the standard code — a meaningful gain in a regime where every decibel of link budget matters.

The analytical results were validated through simulation, covering both qubit error rates and secure key generation rates. The researchers explicitly acknowledge the throughput penalty, framing deployment as scenario-dependent: CSS codes add value when signals are weak, but their benefit diminishes as received power increases.

---

## Why Underwater QKD Is a Hard Problem

[Quantum key distribution](https://quantumintel.tech/glossary/algorithmic-qubits) relies on single photons or entangled photon pairs to distribute encryption keys whose security is guaranteed by quantum mechanics rather than computational hardness. Any intercept-and-resend attack by an eavesdropper disturbs the quantum states in a detectable way — that is the physics.

The problem underwater is brutal photon loss. Water absorbs and scatters optical signals, and even in relatively clear ocean conditions the attenuation per meter is orders of magnitude higher than in optical fiber or free-space atmospheric links. Solar background noise compounds the problem for near-surface or vertical links. The result: conventional QKD implementations hit range limits rapidly in aquatic environments.

Existing fiber QKD deployments — operated commercially by companies including [ID Quantique](https://quantumintel.tech/companies/id-quantique) and [QuantumCTek](https://quantumintel.tech/companies/quantumctek) on terrestrial networks — benefit from low-loss silica fiber and well-characterized noise environments. Underwater is a fundamentally different engineering challenge, and the academic literature on it remains thin relative to free-space and fiber QKD.

This paper is a simulation and modelling study, not a hardware demonstration. The Jerlov type III water scenario is a defined optical classification, not a specific geographic location, and the results describe achievable performance bounds rather than measured field data. Independent experimental validation will be necessary before any infrastructure operator could size a real system around these numbers.

---

## Industry Implications

The immediate relevance is to subsea critical infrastructure protection. Offshore energy platforms, intercontinental cable landing stations, and naval communication nodes all represent high-value targets where quantum-secured key exchange has strategic appeal — if the physics can be made to work at useful ranges and data rates.

The throughput penalty identified in this work is the central engineering constraint any product team would face. The paper's finding that reducing photon arrival probability by factors of ten causes significant key rate reductions is a quantitative statement of the range-versus-rate trade-off that dominates all QKD link engineering. A 4.5 dB gain in SNR is useful, but it does not eliminate that trade-off — it shifts the feasibility boundary.

From a QEC perspective, this application is analytically distinct from the fault-tolerant computing context. Underwater QKD does not require [logical qubit](https://quantumintel.tech/glossary/logical-qubit) operations or the surface code overhead that dominates current fault-tolerant quantum computing research. The four-qubit CSS code studied here is purpose-fit for a photonic communication channel, not a gate-based processor. That is a useful reminder that QEC techniques developed for computing platforms can find application in quantum networking with quite different design constraints.

The authors — affiliated with French institutions based on the name set — have posted to arXiv; peer review status is not stated in the source material. The work should be treated as a strong modelling contribution pending independent review and, ultimately, experimental demonstration in controlled water tank or open-water conditions.

---

## Key Takeaways

- The discard variant of a four-qubit CSS quantum error correction code achieves a **4.5 dB SNR gain** in underwater BB84 QKD at an **eleven percent security threshold**, outperforming the standard CSS code's 3 dB gain at the same threshold.
- Simulations explicitly modelled **Jerlov water type III** conditions — turbid, high-attenuation — incorporating photon loss, geometric spreading, and solar noise.
- The QEC approach extends viable transmission ranges but **reduces data throughput**; the paper identifies this as a scenario-dependent deployment trade-off, not a solved problem.
- Reducing photon arrival probability by factors of ten produces **significant key rate reductions** across all configurations tested.
- This is a **modelling study**, not a hardware demonstration; experimental validation in real water channels remains the critical next step.
- Authors: Juliette Florin, Nicolas Le Josse, Arnaud Coatanhay, and Gilles Burel (arXiv: 2609.18666).

---

## Frequently Asked Questions

**What is a CSS quantum error correction code?**
A Calderbank-Shor-Steane (CSS) code is a class of quantum error correcting code that independently corrects bit-flip and phase-flip errors. It is constructed from two classical linear codes and is widely studied because it maps cleanly onto many quantum hardware architectures. The four-qubit variant in this study is compact relative to codes used in fault-tolerant computing, making it suitable for photonic communication channels where qubit overhead is constrained.

**How does QEC help underwater QKD specifically?**
Underwater optical channels suffer from high photon loss due to absorption and scattering, plus solar background noise in near-surface configurations. QEC codes add redundancy to the transmitted quantum information so that the receiver can recover the key even when a significant fraction of photons are lost or corrupted. The result is an extended viable transmission range — quantified here as up to 4.5 dB at an eleven percent error threshold — in conditions where uncorrected QKD would fail entirely.

**What is the eleven percent security threshold?**
In BB84 QKD, the quantum bit error rate (QBER) must remain below a threshold for the communicating parties to be confident that no eavesdropper has extracted useful information. The eleven percent figure cited in this paper represents the error rate boundary at which the researchers evaluated the SNR gains from the CSS codes. Operating above this threshold would compromise the security guarantee that makes QKD physically meaningful.

**Does this mean underwater QKD is now commercially deployable?**
Not yet. This is a simulation and modelling result validated analytically, not an experimental hardware demonstration. Operators of subsea infrastructure would need field trials in representative water conditions, system integration work, and independent validation before basing procurement decisions on these figures. The throughput penalty from QEC also requires careful link budget analysis for any specific application.

**How does this relate to terrestrial QKD networks?**
Commercial terrestrial QKD is already operational in fiber networks — companies such as ID Quantique and QuantumCTek have deployed metropolitan and long-haul fiber QKD links. Underwater QKD faces fundamentally different channel physics: much higher attenuation per meter and dynamic noise conditions. The CSS code approach studied here is specific to the underwater photonic channel problem and does not directly transfer to fiber network architecture, though the underlying QEC mathematics is shared.