Quantum grid codes could extend direct secure messaging beyond 1,000 kilometres

Sending the message itself

Quantum key distribution establishes shared random keys that can subsequently protect a message. Quantum secure direct communication, or QSDC, instead encodes the message into quantum states themselves, with checks intended to detect eavesdropping.

The challenge is distance. Optical fibre loses photons, while noise damages the states carrying information. Meng-Dong Zhu and colleagues propose addressing both problems using Gottesman–Kitaev–Preskill, or GKP, qubits and error-correcting repeater stations. Their study is an arXiv preprint, with no journal publication listed in the supplied metadata. It reports calculations and numerical simulations, not an experimental long-distance link.

A grid that makes small errors repairable

A GKP code stores a quantum bit in an oscillator, such as a mode of light. Its structure can be pictured as a repeating grid in phase space: a mathematical map of two complementary properties called quadratures, analogous to position and momentum.

Small disturbances shift the state away from its grid locations. Error correction estimates those shifts and moves the state back without reading out the encoded logical information.

Perfect GKP states would require infinite energy. Realistic approximations have broadened grid peaks. The paper describes their quality using a squeezing level: higher squeezing means narrower peaks and better tolerance of displacement errors. Its main simulations assume approximately 23 decibels of squeezing.

At each repeater, auxiliary GKP states interact with the message-carrying state through controlled operations called SUM gates. Homodyne detection measures an auxiliary state's quadrature, producing an error estimate. A corrective displacement then adjusts the data state. Optical amplification helps compensate for loss, although it does not eliminate the resulting noise.

These are third-generation quantum repeaters: they rely on quantum error correction rather than entanglement purification, which the authors explain can erase information already encoded in the states.

An outward journey and a return

The protocol begins with Bob preparing randomly chosen GKP states and sending them through the repeater chain to Alice. Alice measures a random sample. Comparing compatible preparation and measurement choices over an authenticated classical channel allows them to estimate the outward journey's error rate.

If that check passes, Alice corrects the remaining states and encodes message bits using either no displacement or a displacement that shifts both quadratures. She sends the states back through the repeaters. Bob measures them using his original preparation choices, checks additional test states, and then decodes the message if the return-stage check passes.

The simulations track displacement-noise variance and calculate how logical bit and phase errors accumulate. From these errors, the authors derive a lower bound on secrecy capacity: the message information available to Bob after accounting for an upper bound on an eavesdropper's information.

What the simulations achieve

With approximately 23 dB squeezing and a combined fibre-coupling and homodyne-detection efficiency of 98%, the model gives a maximum distance of 4.70 kilometres without repeaters. Ten repeaters extend that to 19.04 kilometres; 120 reach 103.45 kilometres.

The headline result requires 2,750 repeater nodes and reaches 1,010.37 kilometres. That is a simulated endpoint separation, with the protocol also requiring a return transmission—not an experimentally demonstrated connection.

The researchers additionally optimize a resource-cost measure that balances repeater storage-mode time steps, distance and secrecy capacity. In their examined range, the lowest-cost point uses one repeater over 3.46 kilometres, with a capacity of 0.70 bit per state. This short-distance optimum is distinct from the maximum-distance result. These capacities are not measured data rates in bits per second, and the cost measure is not a financial deployment estimate.

The limits behind the distance

Security is analysed against collective attacks: an eavesdropper interacts separately with each transmitted state but can store auxiliary systems and measure them jointly later. The analysis does not establish protection against every possible attack.

The noise model also makes simplifying assumptions. It neglects the broad envelope of finite-energy GKP states, assumes independent Gaussian noise in the two quadratures and independent logical errors across repeaters, and sets overall signal gains to unity while incorporating attenuation and detector inefficiency into displacement noise.

Most importantly, demonstrated optical GKP-like states remain below the squeezing regime required here. Imperfect gates, feedback, storage and mode matching require further treatment, as does temporary storage during Alice's checks and encoding. Proposed microwave implementations would also need efficient, low-noise conversion to optical signals.

The authors suggest that detecting eavesdropping could help protect communication personnel. Their calculations do not establish location privacy or physical safety. The supported result is narrower: under specified assumptions, GKP error correction offers a theoretical route to much longer direct quantum communication.