Adaptive entropy loading via direct activation for an optical frequency comb-based few-mode transmission system

Area of Science:

• Optical Communications
• Information Theory
• Neuromorphic Engineering

Background:

• Optical frequency combs offer efficient multi-carrier light sources but face performance variations across channels and modes.
• Power fluctuations, intermodal crosstalk, and environmental instability degrade system performance in few-mode transmission.

Purpose of the Study:

• To develop an adaptive entropy loading mechanism for mitigating performance discrepancies in optical frequency comb-based few-mode transmission systems.
• To enhance channel utilization and system adaptability to dynamic network conditions.

Main Methods:

• Implemented a neuronal direct activation structure for multi-dimensional entropy loading.
• Utilized a model-free, neuromorphic-inspired framework with tanh activation and max-pooling for instantaneous entropy reallocation.
• Demonstrated iterative-free dynamic entropy adjustment.

Main Results:

• Achieved effective mitigation of non-uniform channel performance.
• Reported a 4.82% capacity enhancement, translating to a 31.7 Gbit/s data rate increase.
• Verified instantaneous adaptability to varying network conditions.

Conclusions:

• The proposed multi-dimensional entropy loading mechanism significantly improves performance in optical frequency comb systems.
• This neuromorphic-inspired approach offers a scalable and robust solution for next-generation data center interconnects.
• Extends entropy loading principles to multi-dimensional coherent systems.