OPTICAL NETWORKING & SENSING
PROJECTS
PEOPLE
PUBLICATIONS
PATENTS
AI-Aided Transmission Systems with Photonic Signal Processing
With exponential traffic growth demand from 5G and cloud applications, optical networks powering the Internet backbone will require orders-of-magnitude improvement in capacity and latency while minimizing the carbon footprint of transmission systems. Advanced signal processing approaches aided by artificial intelligence are at the core of our innovation in this area.
Some examples are photonic neural network-based nonlinearity compensation, QoT prediction for advanced modulation formats, and novel transmission fiber engineering that optimizes spectrum utilization and maximizes transmission reach while reducing energy costs. Our recent world-leading achievements include the first silicon photonic neural network chip that compensates fiber nonlinearity for >10,000-km submarine cable, the highest C-band field fiber capacity at 41.5-Tb/s using probabilistic shaping and AI-aided GSNR prediction, and in-service transmission cable health monitoring with > 1,000-km range.