A protocol for detecting rare signals in noisy environments using quantum error correction is proposed. The core principle is to separate signal from noise via unique higher-order correlations that arise from nonlinear processing during syndrome extraction. Error correction serves two functions: it suppresses physical noise, extending the coherence time of the logical qubit for signal accumulation; it discards part of the signal—the detected logical phase is on the order of the cube of the original signal strength. For rare signals occurring at random moments in the presence of local Markovian noise, a clear sensitivity advantage over conventional sensing strategies is demonstrated. This result paves the way for highly sensitive quantum sensors for sporadic events.
Detecting a faint signal against a noise background is like trying to hear a distant bell in a storm. Ordinary amplification only makes the din more deafening. Quantum error correction solves the problem in an unexpected way. This trick, originally devised to protect quantum computers from glitches, works as a clever filter: it not only cuts out noise but also deliberately distorts the signal itself. As a result, all that remains of the actual ringing is a faint rhythmic echo whose strength depends on the cube of the original loudness. In other words, if the real signal is weakened by a factor of 10, after filtering it will become 1000 times quieter.
The paradox is that this is precisely what allows detecting rare events. By suppressing noise almost to zero, the detector gains the ability to accumulate data for a very long time. By repeatedly reading the weakened echo, it separates it from random bursts — much like subtracting identical images reveals differences. Sensitivity to phenomena such as the passage of a dark matter particle or a burst from the early Universe increases by orders of magnitude. This approach once again proves: in the quantum world, loss can turn into gain.
🎯 This method grew out of the fight against errors in quantum computing. Its transfer to the sensor domain was so non-obvious that the idea was initially rejected as contradicting the logic of amplification.