Frame Segmentation in Ultrahigh-Speed Spectrum Analyzer Based on Time-Lens

Abstract

With the development of high-frequency carriers and high-capacity communications, ultrahigh-speed spectrum analyzer based on all-optical methods has been demonstrated in many fields. Limited by current data processing algorithms, ultrahigh-speed spectrum analyzers based on time-lens, such as frequency-domain light intensity spectrum analyzer (f-LISA) and parametric spectro-temporal analyzer (PASTA), are not yet to automatically process acquired data, which hinders their practicality. In this paper, a frame segmentation algorithm based on time-domain waveform analysis is proposed to achieve the automatic data processing of these analyzers, which can directly generate the spectro-temporal image of the signal under test. In a single-soliton test based on the f-LISA platform, the algorithm achieves an accuracy of 93.28% when the soliton changes and 90.04% when the soliton is stable. And in this test, the single-frame processing time of the algorithm is 419us. We believe the proposed algorithm will facilitate the practical application of spectrum analyzers like f-LISA and PASTA and provide a reference for other real-time measurement schemes.

Keywords

• Correlation coefficient
• decision tree
• pulse detection
• spectrum analysis
• ultrafast optics