Machine learning corrects wavefront error in 2.4km free-space optical link, reducing phase error variance by 2/3

Atmospheric turbulence poses significant challenges to free-space optical communications, distorting optical signals and limiting reliable data transmission. Nathan K. Long of the University of New South Wales, along with Benjamin P. Dix-Matthews and Alex Frost of the University of Western Australia, demonstrated that standard assumptions about signal and reference beam distortion in such links are inaccurate. The research team experimentally reveals measurable differences in wavefront errors between these beams over a 2.4-kilometer atmospheric link and develops machine learning algorithms to correct for these relative distortions. This innovative approach achieves significant reductions in phase errors and, importantly, suggests the potential for orders of magnitude increases in secure key rates for new continuously variable quantum key distribution techniques, paving the way for more robust and secure optical communication systems.

Correcting kilometer-scale optical turbulence with machine learning

Researchers demonstrate machine learning techniques to correct wavefront errors and achieve improved communication fidelity in a 2.4km free-space optical link. This research addresses the critical challenge of atmospheric turbulence, which distorts optical signals and limits the performance of free-space optical communication systems. The team developed a system that learns to predict and compensate for these distortions in real time, improving signal quality and extending communication range. This approach utilizes wavefront sensors to measure the distortions introduced by the atmosphere and uses machine learning algorithms to calculate and apply the necessary corrections to the transmitted signal. The results show that signal quality can be significantly improved, paving the way for more reliable and efficient free-space optical communication systems, especially for applications that require high bandwidth and secure communication, such as satellite communications and data transfer between buildings.

For coherent optical communication across turbulent atmospheric channels, reference beacons are commonly used along with information-encoded signals. The team presents experimental evidence of relative wavefront errors between polarization-multiplexed reference beacons and their post-transmission signals over a 2.4km atmospheric link. They developed a machine learning-based wavefront correction algorithm that uses phase retrieval techniques to compensate for these observation errors, and were able to reduce observation errors by up to two-thirds.

Wavefront error reduced by machine learning

This study presents experimental results demonstrating the relative wavefront error between an optical reference beam and a weaker signal beam in a free-space optical communication system. A central question addressed is whether these errors degrade signal quality and whether they can be mitigated using machine learning-based wavefront correction. This research focuses on implications for continuously variable quantum key distribution systems. The team established a 2.4km free-space optical link and multiplexed the signals using multiple orthogonal Hermite-Gaussian modes. They observed that the reference and signal beams experienced different wavefront distortions due to atmospheric turbulence. This is a phenomenon called relative wavefront error.

A machine learning model was trained to correct for these relative wavefront errors. Experiments have confirmed that significant relative wavefront errors exist between the reference and signal beams. The machine learning-based wavefront correction algorithm was successful in reducing the total dispersion of the wavefront, achieving up to a factor of 2 reduction. This reduction in wavefront error leads to a reduction in effective excess noise in quantum key distribution systems, which can lead to a significant increase in secure key rates. Increasing the number of Hermitian-Gaussian modes to increase capacity also increased the transmission, resulting in an increase in effective excess noise, but wavefront correction helped offset this increase.

This study suggests that implementing wavefront correction algorithms can significantly improve the performance of continuously variable quantum key distribution systems over turbulent free-space optical channels. Machine learning-based correction provides a way to reduce the negative effects of atmospheric turbulence on optical communications. This study highlights the trade-off between increased channel capacity and increased noise, and demonstrates that wavefront correction can help balance this trade-off. The researchers were able to reduce total wavefront dispersion by up to a factor of 2 and reduce effective excess noise by up to 17%, while maintaining a high coherence efficiency of approximately 0.98 even with corrections.

Wavefront correction expands quantum communication range

This study presents important discoveries regarding coherent optical communication through turbulent atmospheric channels. Scientists have experimentally observed the relative wavefront error between a reference beacon and an information-carrying signal, a phenomenon previously thought to be negligible. The research team confirmed the existence of these relative wavefront errors across a 2.4km atmospheric link and developed a machine learning-based wavefront correction algorithm to mitigate their effects, successfully reducing the phase error variance by up to two-thirds. Successful implementation of these algorithms has important implications for continuous variable quantum key distribution.

Our analysis shows that adopting a similar wavefront correction technique can increase the secure key rate by an order of magnitude, potentially significantly increasing the efficiency and security of long-range quantum communications. Although the source of these relative wavefront errors remains to be determined, the research team emphasizes that identifying the source is not essential to the effectiveness of the correction scheme. The authors acknowledge that the effect of wavefront correction on coherent efficiency may be more pronounced between longer links or under stronger turbulence conditions. Future research will focus on investigating these effects and further improving algorithms to optimize performance under different atmospheric conditions, ultimately contributing to more robust and secure free-space optical communication systems.