Photonic principal component analysis using an on-chip microring weight bank

Area of Science:

• Photonics
• Signal Processing
• Applied Physics

Background:

• Dimensionality reduction is crucial for wideband analog systems.
• Existing methods may lack efficiency or require signal waveform knowledge.
• Photonic approaches offer potential for high-performance solutions.

Purpose of the Study:

• To develop a photonic principal component analysis (PCA) method for dimensionality reduction.
• To implement PCA using an on-chip microring resonator (MRR) weight bank.
• To enable blind PCA extraction using only statistical output information.

Main Methods:

• Utilizing an on-chip microring resonator (MRR) weight bank for weighted addition of wavelength-division multiplexed (WDM) inputs.
• Configuring the MRR weight bank with high accuracy and precision.
• Developing a novel PCA algorithm for blind waveform extraction based on statistical output data.

Main Results:

• Experimental demonstration of two-channel photonic PCA with high correspondence to software-based PCA.
• Numerical simulations confirm generalization to higher dimensions (up to eight channels) with good convergence.
• Successful generation of multi-channel correlated input signals in a controllable manner.

Conclusions:

• The proposed photonic PCA technique offers a high-performance, blind dimensionality reduction solution.
• The on-chip MRR approach provides accurate and precise weighted operations.
• This method has potential applications in microwave communications, ultrafast control, and on-chip information processing.