Pilot-Free VCSEL Temperature Monitoring via Statistical Complexity

Artikel i vetenskaplig tidskrift, 2026

Vertical-cavity surface-emitting lasers (VCSELs) are the dominant light sources in short-reach optical interconnects, where cost, efficiency, and scalability are critical. However, their modulation bandwidth, output power, and signal integrity degrade markedly as ambient temperature rises and self-heating increases, making accurate device-level temperature awareness indispensable. Existing approaches rely on embedded sensors or forward-voltage monitoring, which require calibration, additional hardware, or pilot overhead, and are therefore not well suited for in-service operation. This work introduces a pilot-free and sensor-free method for inferring VCSEL operating temperature directly from payload signals. We establish, through an electro–thermal rate-equation model, that temperature rise manifests as a systematic reduction in the entropy of the optical waveform. Leveraging this property, we develop a regression-based estimator that achieves sub-5°C accuracy in simulation. The results demonstrate that entropy-based payload analysis provides a principled and low-cost proxy for junction temperature, with potential for integration into high-speed link management.