Automatic Piecewise Extreme Learning Machine-Based Model for S-Parameters of RF Power Amplifier

Area of Science:

• Electrical Engineering
• Computational Intelligence

Background:

• Accurate modeling of radio-frequency (RF) power amplifiers (PAs) is crucial for modern communication systems.
• Existing modeling techniques like Long-Short Term Memory (LSTM), Support Vector Regression (SVR), and conventional Extreme Learning Machine (ELM) face challenges in speed and accuracy.

Purpose of the Study:

• To develop an efficient and accurate automatic piecewise (Auto-PW) extreme learning machine (ELM) method for S-parameters modeling of RF PAs.
• To enhance the modeling speed and accuracy of RF PA S-parameters.

Main Methods:

• Proposed an automatic piecewise (Auto-PW) extreme learning machine (ELM) approach.
• Implemented a region-splitting strategy based on concave-convex characteristic changing points.
• Developed piecewise ELM models for distinct regions.

Main Results:

• The Auto-PW ELM method demonstrated superior performance on S-parameters measured from a 2.2-6.5 GHz CMOS PA.
• Achieved modeling speeds two orders of magnitude faster than SVR and LSTM.
• Obtained modeling accuracy more than one order of magnitude higher than conventional ELM.

Conclusions:

• The Auto-PW ELM method offers a significant advancement in RF PA S-parameters modeling.
• This approach provides a faster and more accurate alternative to existing modeling techniques.