Experimental evaluation demonstrates order-of-magnitude accuracy gains for piecewise ELM modeling in a CMOS power amplifier, highlighting improved RF device characterization across temperatures.
This study aimed to enhance the modeling accuracy of the conventional extreme learning machine (ELM) for radio frequency (RF) power amplifier (PA) small-signal scattering coefficient (S-parameter) characterization across multiple ambient temperatures. The study proposes an automatic piecewise ELM (PW-ELM) for the small-signal S-parameter magnitude characterization of RF PAs at multiple discrete temperature points. The proposed method is validated using S-parameter measurements of a 2.2–6.5 GHz CMOS PA acquired at three critical temperature corners: −35 °C, 25 °C, and 125 °C. Experimental results on this challenging CMOS PA demonstrate that the proposed method achieves an order of magnitude higher accuracy compared to backpropagation neural networks (BPNNs), support vector regression (SVR), and conventional ELM, suggesting promising potential for extension to other devices and technologies.
No takes yet. Share an insight, caveat, or question.
Zhang et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: