Reconfigurable parallel photonic matrix-vector multiplication processor based on multi-dimensional multiplexing

Area of Science:

• Photonics
• Optical Computing
• Signal Processing

Background:

• Matrix-vector multiplication (MVM) is crucial for data processing and neural networks.
• Existing photonic MVM processors face limitations in computing power and flexibility.
• There is a growing demand for high-performance, reconfigurable computing architectures.

Purpose of the Study:

• To propose a novel reconfigurable parallel MVM (RP-MVM) processor.
• To enhance the parallelism and flexibility of photonic MVM architectures.
• To demonstrate the effectiveness of the RP-MVM scheme for complex computations.

Main Methods:

• Incorporation of wavelength division multiplexing (WDM) and digital subcarrier multiplexing (DSM) into a photonic MVM processor.
• Development of a reconfigurable architecture allowing dynamic adjustment of input data channels.
• Simulation and analysis of the RP-MVM scheme's performance.

Main Results:

• The RP-MVM scheme achieves parallelism N times greater than traditional WDM-MVM architectures.
• Demonstrated parallel computing of eight MVMs with a speed of 128 GOPs.
• Achieved a root mean square error (RMSE) of 1E-3 for 6-bit resolution data and processed four grayscale images for edge extraction.

Conclusions:

• The proposed RP-MVM processor offers a significant advancement in photonic computing.
• The scheme provides a highly parallel and reconfigurable solution for large-scale MVM tasks.
• This architecture presents a viable alternative for future high-performance optical computing systems.