The Signal
Researchers report quantum-key packets routed through two hops
A laboratory experiment sent quantum key distribution packets through two cascaded routers to three receivers. The researchers calculated nonzero secret key rates across one-hop and two-hop routes.

Quantum key distribution packets reached receivers after passing through as many as two successive optical-fiber routers, according to a research preprint describing a laboratory experiment. The laboratory system produced calculated secret key rates of 35–270 bps across one-hop and two-hop routes. The preprint was submitted to arXiv on November 10.
Routes through the lab
The experiment joined one sender to three receivers through two cascaded routers. One output from the first router led to a receiver; the other led to the second router, which had outputs to the remaining receivers. The components were in a laboratory, with fiber spools of 20 km, 10 km and 5 km placed before the respective receivers.
The researchers used packets with a classical header and a quantum payload, separated in both wavelength and time. The header carried the destination information used to choose a path, while the payload carried the quantum information. Earlier demonstrated routers read and regenerated headers at each routing node, the paper says.
Reading the header
The new router instead uses a partially reflecting fiber Bragg grating to tap 10% of the classical header. A photodiode reads the reflected signal, while the remaining 90% of the header and the quantum payload continue toward an optical switch. A programmable control board uses the detected pulse sequence to set the switch for the packet’s destination.
The experimental header contained a four-bit preamble, a two-bit destination address and a one-bit parity check, with a total duration of 3.5 microseconds. A fiber delay of 5 microseconds preceded the optical switch. The authors report that the packets were routed without regenerating their headers.
Key rates and errors
For the key-distribution test, the team used time-bin qubits and the BB84 protocol with a two-decoy-state method. The signal and decoy lasers were matched in wavelength, spectral width and pulse width to make the states indistinguishable. Receivers collected sifted key bits and processed the data to estimate final secret key rates.
In tests that alternated packet destinations, the researchers detected an average of 4–5 kbits of sifted key for each one-second packet at the intended receivers. The average quantum bit error rate across the three receivers was 3.4%. The researchers also measured a 0.14% difference in that error rate with the classical-channel laser on and off. They attributed that difference to classical light leaking into the quantum channel.
To calculate secret key rates, the team divided the resulting secret key length by the corresponding acquisition period. The authors said the causes of comparatively low rates at the second receiver were unclear. They suggested instability in the setup, including fiber-spool birefringence and ambient temperature, could have played a part.
Loss and future links
The team measured an average quantum-signal loss of 2.8 dB through a router, including losses from its optical components and connections. The paper also notes that the untapped header must retain enough optical power for the code to be distinguished at later nodes.
For applications involving heralded or entangled photons with broader bandwidths, the authors say dispersion-compensating fibers would be needed in the delay lines. They identify long-distance, multi-node networks distributing entangled photons as a possible direction for future work.