2026/01/07 by Zichao Gao, Siyu Lu, Mingming Zhang +18
Computer Science · Engineering · #Neural Networks and Reservoir Computing #Optical Network Technologies #Photonic and Optical Devices
paper · pdf · doi:10.1038/s41467-026-75732-z
The rapid growth of artificial-intelligence computing demands high-bandwidth and energy-efficient data-center interconnects. Although self-homodyne coherent (SHC) transmission reduces digital signal-processing complexity, its practical deployment is limited by ultrafast state-of-polarization (SOP) fluctuations that destabilize coherent reception. Here we show an ultrafast adaptive polarization controller (APC) based on thin-film lithium tantalate, enabled by the co-design of a photonic chip, a finite-boundary gradient-descent (FBGD) algorithm, and a low-latency real-time control system. The FBGD algorithm incorporates the finite phase boundary directly into the control strategy, enabling reset-free continuous SOP tracking. The implemented APC is validated through both standalone polarization-tracking measurements and a dual-polarization 16-QAM SHC 400-Gbps transmission system. Stable reset-free operation under continuous polarization disturbances is maintained up to 1 Mrad·s⁻¹. These results establish a previously inaccessible operating regime for integrated polarization control and provide a practical route toward ultrafast and reset-free polarization control in coherent optical interconnects and other polarization-sensitive photonic systems.