Support Vector Machine–Based Model for 2.5–5.2 GHz CMOS Power Amplifier

Abstract

A power amplifier (PA) is the core module of the wireless communication system. The change of its specification directly affects the system’s performance and may even lead to system failure. Furthermore, change in the PA specification is closely related to changes in temperature. To study the influence of PA specification change on the system, we used a support vector machine (SVM) to model the temperature characteristics of PA. For SVM modeling, the question of how much experimental data should be used for modeling to meet the requirements is a constant problem. To address this issue, we investigate the effect of different amounts of training data on the modeling of SVM models. The results show that only 75% of the experimental data needs to be used in the modeling process to satisfy the modeling requirements of the SVM model. The number of measurement points required in the PA specification degradation experiment can be reduced by 25%. The results of this paper serve as a guide for planning the number of experimental measurement points and reducing the measurement cost and measurement time.

BPNN-Based Behavioral Modeling of the S-Parameter Variation Characteristics of PAs with Frequency at Different Temperatures

“…The literature [20] used the SVM to model the S-parameters of a 2.5-5.2 GHz CMOS PA, and the results show that the SVM can be used to model the S-parameters. Therefore, this paper has modeled the S-parameters of a 0.3-1.1 GHz CMOS PA, using a BPNN and SVM, respectively, and tried to compare these two models in terms of modeling speed and accuracy, and the results are shown in Table 1.…”

“…In 2020, Gaoming Xu et al from Ningbo University, China, proposed a behavioral model consisting of Chebyshev polynomials (CP) and a long short-term memory (LSTM) network, i.e., CP-LSTM, for the PA's behavioral modeling [19]. In 2022, we modeled the behavior of the PA's temperature characteristics, using a support vector machine (SVM) and extreme learning machine (ELM) [20,21], respectively.…”

Temperature behavior modeling based on resilient BPNN for a GaAs pHEMT high gain MMIC PA

“…Subsequently, the four two‐dimensional interpolation models based on a few key measurement points were applied to analysis the performance difference of a gallium nitride (GaN) class‐AB PA when temperature changed 21 . In 2022, the support vector machine (SVM) was first utilized to model the temperature behavior of a CMOS class‐A PA according to measurement data and the difference in the number of training data was discussed 22 . In the above investigations, ANNs have been applied to characterize the relationship between temperature and performance of PA based on some measured data.…”

“…Different from other works in References 17, 20–23, a GaAs pHEMT PA is selected as the experiment object, which makes circuit more susceptible to temperature changes due to the low thermal conductivity. It is more urgent to model the temperature behavior of this PA.…”

PSO‐GA‐SVR model for S‐parameters of radio‐frequency power amplifier under different temperatures

“…They proved that SVR has better modeling accuracy and appropriate complexity than the canonical piecewise linear model and ANN 12 . Zhou et al have experimented with the number of data used for modeling power amplifiers using SVR and concluded that only 75% of experimental data is required 13 . Perez‐Wences et al first combined nonlinear autoregressive exogenous architecture with SVR and proposed NARX‐SVR for modeling PA. By retaining the dynamic nonlinear accuracy of PA, NARX‐SVR reduces computational complexity to some extent and improves modeling accuracy 14 .…”