Combined photonic time-delay reservoir computing and LSTM based on VCSEL for network traffic prediction

Area of Science:

• Computer Science
• Network Engineering
• Artificial Intelligence

Background:

• Accurate network traffic prediction is crucial for intelligent network management, security, and user experience.
• Network traffic data exhibits complex characteristics like non-stationarity, nonlinearity, and long-range dependence, challenging traditional prediction models.
• Existing prediction approaches are insufficient for the complexities of modern network traffic forecasting.

Purpose of the Study:

• To develop an advanced network traffic prediction model that overcomes the limitations of current methods.
• To enhance the accuracy and robustness of network traffic forecasting for improved network management.
• To leverage the strengths of reservoir computing and LSTM for complex time-series data analysis.

Main Methods:

• Integration of time-delay reservoir computing (RC) with a vertical cavity surface emitting laser (VCSEL) for high-dimensional feature mapping.
• Replacement of the RC output layer with a long short-term memory (LSTM) network to exploit temporal characteristics.
• Utilizing nonlinear mapping through virtual nodes for complex feature extraction and noise resistance.

Main Results:

• The proposed RC-LSTM model demonstrated superior prediction accuracy compared to standalone RC and LSTM.
• Significant reductions in normalized root mean square error (NRMSE) were observed: 19-24% on UK Academic Network data and 19-20% on US Research and Education Network data.
• The model maintains good prediction performance across different network environments, showing robustness to noise and outliers.

Conclusions:

• The integrated time-delay RC and LSTM model offers a powerful solution for accurate and robust network traffic prediction.
• This approach effectively captures complex temporal dynamics and nonlinear features in network traffic data.
• The findings support the model's potential for enhancing network security and optimizing user service experience through intelligent network management.