Enhancing Secure Key Rates via Redundant Information-Reconciliation Leakage Elimination in Practical Quantum Key Distribution Chips

In the world of secure communication, quantum key distribution offers a promise that sounds almost too good to be true: a method of sharing secret codes that is guaranteed by the laws of physics to be unbreakable. Imagine two people, Alice and Bob, trying to agree on a secret password while a third party, Eve, tries to listen in. In a quantum system, if Eve attempts to peek at the signals Alice sends, she inevitably disturbs them, leaving a trace that Alice and Bob can detect. This allows them to know if their line is safe. However, real-world systems are not perfect. The devices used to send these signals often emit more than one particle at a time, creating a vulnerability where a clever eavesdropper could steal information without being noticed. To fix this, scientists use a technique called the "decoy state," which involves sending out different types of signals to trick the eavesdropper and reveal their presence. Even with these safeguards, the process of turning raw data into a final secret code involves a step called information reconciliation, where Alice and Bob publicly discuss their data to fix errors. This public discussion is necessary, but it also leaks information to the eavesdropper, who can use those public clues to guess parts of the secret key. For years, the standard way of calculating how much secret key remains has been to assume the worst possible scenario: that every single piece of information shared during this error-correction step is new and dangerous information that the eavesdropper did not already know.

A team of researchers has now shown that this standard assumption is far too cautious, effectively throwing away a significant amount of usable secret key. By re-examining how the error-correction process works, they discovered that much of the information Alice and Bob share publicly is actually redundant. Because the eavesdropper is already assumed to know the contents of certain "flawed" signals (the multi-photon ones) before the error correction even begins, any public clues derived solely from those flawed signals reveal nothing new. It is as if the eavesdropper already has the answer key for a specific set of questions; when Alice and Bob publicly discuss those questions, they are not giving away any new secrets. The researchers developed a new mathematical model that filters out this redundant information, counting only the public data that actually involves the secure signals. They tested this idea not just on a computer, but on a real, functioning quantum system built into a tiny silicon chip, operating at speeds of 2.5 billion pulses per second. When they applied their new model to the data collected from this chip, the results were striking. The system was able to generate a secure key rate that was up to 78.82 percent higher than what the old, conservative method predicted.

The experiment took place in Hong Kong, where the researchers used a 2.5 gigahertz silicon-photonic chip to send polarized light pulses through a real-world fiber optic network. This network stretched 55.41 kilometers, looping through the city's existing infrastructure, connecting the university to data centers in a way that mimics a busy, practical communication line. The chip itself was small, fitting on a piece of silicon measuring just 2.1 by 7.8 millimeters, yet it was powerful enough to handle the complex task of encoding information into the polarization of light. The team sent these signals through the city's fibers, which included connectors, splices, and aging cables, creating a realistic environment full of noise and signal loss. At the receiving end, they measured the light and recorded every detection event. They then ran the same set of experimental data through two different calculations: the traditional method that assumes every public clue is a leak, and their new method that ignores the clues that the eavesdropper already knew. The difference was substantial. At the specific operating point of their field test, the new method allowed them to extract a secret key at a rate of 1.55 times 10 to the negative fifth per pulse, compared to the 8.64 times 10 to the negative sixth per pulse yielded by the old method. This translated to a secure data rate of 38.75 kilobits per second, nearly double the 21.6 kilobits per second of the conventional approach.

The significance of this finding extends beyond just getting more data; it changes how much distance the system can tolerate. In the world of quantum communication, signal loss is the enemy. The more fiber the light travels through, the more it fades, and eventually, the system stops working because there is too much noise to distinguish a real signal from a mistake. The researchers found that by using their tighter leakage model, the system could tolerate an additional 2.35 decibels of loss before the secure key rate dropped to zero. This might sound like a small number, but in the context of fiber optics, it represents a significant extension of the network's reach. The team also ran simulations to see how this would hold up over longer distances, finding that the new model could extend the maximum distance for a positive key rate by about 5 kilometers compared to the conventional method. Crucially, this improvement did not require any changes to the hardware. The researchers did not need to build a better chip, a more sensitive detector, or a stronger laser. They simply changed the way they counted the information, realizing that the old rules were penalizing the system for information that was never actually lost.

This approach offers a practical path forward for upgrading existing quantum networks without the massive cost of replacing physical infrastructure. The researchers demonstrated that by carefully distinguishing between information that is truly new to an eavesdropper and information that is already known, they could unlock a hidden reserve of security. The model works by identifying the specific moments in the error-correction process where the public discussion involves only the "tagged" bits—those known to be flawed—and ignoring them in the security calculation. It keeps the conservative safety net for the rest of the data, ensuring the system remains secure, but it stops double-counting the risk. The results were verified both in a controlled laboratory setting where they could precisely adjust the signal loss, and in the field test over the Hong Kong metropolitan network. In both cases, the new model consistently outperformed the old one, proving that the theoretical gain was real and measurable. The work suggests that many current quantum communication systems are operating below their true potential, simply because they are using an outdated way of measuring their own security. By adopting this more precise accounting, operators of quantum networks can significantly boost their performance, making secure communication more viable for high-demand city networks and even for future links between satellites and the ground.