Extreme learning machine for the behavioral modeling of RF power amplifiers

Zhang, Chengyu; Zhu, Yi; Cheng, Qian‐Fu; Fu, Haipeng; Ma, Jan; Zhang, Qi‐Jun

doi:10.1109/mwsym.2017.8058626

Cited by 12 publications

(14 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Two-dimensional interpolation is a standard and mature mathematical modeling method, which is also suitable for nonlinear modeling problems and convenient to use at the system level. There is a strong nonlinear relationship between the temperature/humidity and the PA's performance and the relationship between the input and output [34], [51]. Therefore, to get the temperature and humidity characteristics of a PA in the whole temperature and humidity range, four different two-dimensional interpolation methods are used to model and predict the temperature and humidity characteristics of the PA.…”

Section: ) Proposed Measurement-based Modeling Processmentioning

confidence: 99%

Characteristics Modeling of GaN Class-AB Dual-Band PA Under Different Temperature and Humidity Conditions

Zhou

Jha

2021

IEEE Access

View full text Add to dashboard Cite

As the technology scales down, besides the CAD tools and design guidelines, understanding circuit performance degradation as a consequence of transistor degradation becomes essential for designers. Due to several adverse environmental conditions, the study of performance degradation in the power amplifier (PA) is very demanding research. In this paper, an RF PA is experimentally studied to observe various characteristic degradations in a broad range of operating temperature and humidity conditions. Based on a few key measurement points, a measurement-based modeling method is proposed to help designers make intelligent decisions to minimize the performance degradation effects. This method uses four two-dimensional interpolation models, and the results show that the prediction results of the four planar interpolation models are in good agreement with the measurement results. The results presented in this article help select design models for the RF power amplifier that can analyze the performance degradation of the transistor parameters in advance. Cubic and spline interpolation has the highest model accuracy among the four two-dimensional models, while the nearest interpolation offers the shortest training time.

show abstract

Section: ) Proposed Measurement-based Modeling Processmentioning

confidence: 99%

Characteristics Modeling of GaN Class-AB Dual-Band PA Under Different Temperature and Humidity Conditions

Zhou

Jha

2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Conventional optimization routines require numerous EM simulations, which entails significant computational costs. To alleviate this difficulty, a number of techniques have been proposed, including adjoint sensitivities [14][15][16][17][18][19], surrogate-based methods involving multi-fidelity simulation models [13,[20][21][22], response surface approximations [23], several variations of space mapping (SM) [24] (e.g., aggressive space mapping [25], implicit SM [26]), feature-based optimization [27], but also machine learning methods [28,29], and surrogate-assisted versions of nature-inspired algorithms [30,31].…”

Section: Introductionmentioning

confidence: 99%

Optimization-Based Antenna Miniaturization Using Adaptively Adjusted Penalty Factors

Mahrokh

Koziel

2021

Electronics

View full text Add to dashboard Cite

The continuing trend for miniaturization of electronic devices necessitates size reduction of the comprising components and circuitry. Specifically, integrated circuit-antenna modules therein require compact radiators in applications such as 5G communications, implantable and on-body devices, or internet of things (IoT). The conflict between the demands for compact size and electrical and field performance can be mitigated by means of constrained numerical optimization. Evaluation of performance-related constraints requires expensive electromagnetic (EM) analysis of the system at hand; therefore, their explicit handling is inconvenient. A workaround is the penalty function approach where the primary objective (typically, antenna size) is complemented by additional terms quantifying possible constraint violations. The penalty coefficients that determine contributions of these terms are normally adjusted manually, which hinders precise control over antenna performance figures and often leads to inferior results in terms of achieved miniaturization rates. This paper proposes a novel algorithm featuring an automated adjustment of the penalty factors throughout the optimization process. Our methodology is validated using three broadband antenna structures. The obtained results demonstrate that the presented adaptive adjustment permits a precise control over the constraint violations while leading to better miniaturization rates as compared to manual penalty term setup.

show abstract

“…Recently, the extreme learning machine (ELM), a singlehidden layer feed-forward neural network with the efficient calculating ability [18], has been employed to model and design microwave components [19], [20]. In [19], a modified optimization algorithm is proposed to set the optimal initial weights and thresholds of ELM training, and accurate results are obtained with fewer training samples.…”

Section: Introductionmentioning

confidence: 99%

Parametric Modeling of Microwave Components Based on Semi-Supervised Learning

et al. 2019

View full text Add to dashboard Cite

In the artificial neural network (ANN), the most time-consuming part for parametric modeling of microwave components is the collection of training datasets from full-wave electromagnetic simulations. However, the reported models for parametric modeling of EM behaviors are based on supervised learning, in which the labeled sampling data from full-wave EM simulations ought to be sufficient for ANN training. Thus, the number of full-wave simulations is the main factor that influences the effectiveness of collecting training and testing samples. Based on the dynamic adjustment kernel extreme learning machine, this paper proposes a semi-supervised learning model lying between supervised learning and unsupervised learning to largely reduce the number of required training samples. The proposed model contains two training processes, the initial training, and the self-training. In the initial training process, a small number of training samples from full-wave simulations are used to make the model rapidly converge. Then, in the self-training process, the model produces unlabeled training datasets to train itself till the testing accuracy is satisfied. Two numerical examples of a microstrip-to-microstrip vertical transition and a dual-band four-pole filter are employed to verify the effectiveness of the semi-supervised learning model. INDEX TERMS Dynamic adjustment kernel extreme learning machine (DA-KELM), self-learning, semi-supervised learning, unlabeled dataset.

show abstract

Extreme learning machine for the behavioral modeling of RF power amplifiers

Cited by 12 publications

References 11 publications

Characteristics Modeling of GaN Class-AB Dual-Band PA Under Different Temperature and Humidity Conditions

Characteristics Modeling of GaN Class-AB Dual-Band PA Under Different Temperature and Humidity Conditions

Optimization-Based Antenna Miniaturization Using Adaptively Adjusted Penalty Factors

Parametric Modeling of Microwave Components Based on Semi-Supervised Learning

Contact Info

Product

Resources

About