Automated Deep Neural Learning-Based Optimization for High Performance High Power Amplifier Designs

Kouhalvandi, Lida; Ceylan, Osman; Özoğuz, Serdar

doi:10.1109/tcsi.2020.3008947

Cited by 39 publications

(29 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…3 shows ML divides into three groups of supervised, unsupervised, and reinforcement learning. Each group of learning can be modeled by using the artificial neural network (ANN) [41], i.e., a shallow neural network with one or two hidden layers, or the deep neural network (DNN) [42], i.e., a network with more than two hidden layers. ANN/DNN networks can model the nonlinear behavior of circuits by using input and output data that extracted from designs and provide automated environments for predicting design parameters that meet the design specifications.…”

Section: Neural Network Techniquementioning

confidence: 99%

“…Compared with ANNs, DNNs are highly successful and accurate in modeling high-dimensional microwave designs [71][72][73][74][75]. For providing a bright view of using DNN in circuit designs, we clarify the DNN based optimization process presented in [42]. Figure 5 depicts the summarized optimization process for automatically designing high power amplifiers.…”

Section: Neural Network Techniquementioning

confidence: 99%

“…Fig. 5 An overview of the automated DNN-based optimization process in [42] for designing high power amplifiers enhancing efficiency, a design method with harmonic impedance tuning is presented in [91] for obtaining the optimal harmonic impedance [92]. The design methodology [93] is based on the probability distribution function of the signal.…”

Section: Regression Dnn Design Parametersmentioning

confidence: 99%

See 2 more Smart Citations

Optimization techniques for analog and RF circuit designs: an overview

Kouhalvandi

Ceylan²,

Özoğuz

2020

Analog Integr Circ Sig Process

Self Cite

View full text Add to dashboard Cite

Optimizations have gained much consideration from the researchers working in the domains of analog and radio frequency (RF), recently. Dealing with highly nonlinear behavior of active components and aiming to meet design specifications are common issues in all nonlinear circuit designs. As a consequence and accordingly, many studies have been conducted on diverse optimization methods and algorithms for tackling the design problems and meeting optimal solutions with high accuracy. The main purpose of this article is to provide a comprehensive and systematic literature review for the optimization approaches applied by the researchers for designing various analog and microwave circuits. We focus on considering the existing optimization methods from the newly published optimization methods in the last decade. Thus, this study can guide and enlighten complementary metal-oxide-semiconductor analog and RF microwave circuit designers to consider optimization methods commonly used in both areas and to expand conventional performance figures used in their area.

show abstract

Section: Neural Network Techniquementioning

confidence: 99%

Section: Neural Network Techniquementioning

confidence: 99%

Section: Regression Dnn Design Parametersmentioning

confidence: 99%

See 1 more Smart Citation

Optimization techniques for analog and RF circuit designs: an overview

Kouhalvandi

Ceylan²,

Özoğuz

2020

Analog Integr Circ Sig Process

Self Cite

View full text Add to dashboard Cite

show abstract

“…Nevertheless, DNN-based modeling is challenging due to the necessity of appropriate adjustment of the network architecture and its hyper-parameters, as well as to address potential issues such as overtraining [39], [40]. The latter can be avoided by automated architecture determination involving numerical optimization methods [41]. A recent example if a fullyconnected regression model (FCRM) [42], where all components of the DNN, including its architecture, are adjusted through Bayesian optimization (BO) [43].…”

Section: Introductionmentioning

confidence: 99%

Improved Modeling of Microwave Structures Using Performance-Driven Fully-Connected Regression Surrogate

et al. 2021

View full text Add to dashboard Cite

Fast replacement models (or surrogates) have been widely applied in the recent years to accelerate simulation-driven design procedures in microwave engineering. The fundamental reason is a considerable-and often prohibitive-CPU cost of massive full-wave electromagnetic (EM) analyses related to solving common tasks such as parametric optimization or uncertainty quantification. The most popular class of surrogates are data-driven models, which are fast to evaluate, versatile, and easy to handle. Notwithstanding, the curse of dimensionality as well as the utility demands (e.g., so that the model covers sufficiently broad ranges of the system operating conditions), limit the applicability of conventional methods. A performance-driven modeling paradigm allows for mitigating these issue by focusing the surrogate setup process in a constrained domain encapsulating designs being of high quality w.r.t. the assumed figures of interest. The nested kriging framework capitalizing on this idea, renders the constrained surrogate using kriging interpolation, and has been shown to surpass traditional approaches. In pursuit of further accuracy improvements, this work incorporates the performance-driven concept into the fully-connected regression model (FRCM). The latter has been recently introduced in the context of frequency selective surfaces, and combined deep neural networks with Bayesian optimization, the latter employed to determine the network architecture and hyper-parameters. Using two examples of miniaturized microstrip couplers, our methodology is demonstrated to outperform both conventional modeling techniques and nested kriging, with reliable models constructed over multi-dimensional parameters spaces using just a few hundreds of samples.INDEX TERMS Dara-driven modeling; surrogate modeling; performance-driven surrogates; nested kriging; deep regression model; Bayesian optimization.

show abstract

“…Unfortunately, this solution introduces several convergence issues, making the process slow and, most of the time, failing to obtain the optimum solution, especially for wideband designs where high order matching networks (MNs) are used. To tackle this issue, distinct [6][7][8], however, they all neglect the PA performance change with harmonic terminations.…”

Section: Introductionmentioning

confidence: 99%

Automatic Methodology for Wideband Power Amplifier Design

Belchior

Nunes

Cabral

et al. 2021

IEEE Microw. Wireless Compon. Lett.

View full text Add to dashboard Cite

This letter presents an automatic power amplifier (PA) design methodology that uses a multi-dimensional search algorithm to find the best compromise between fundamental and harmonic impedance terminations for a specified bandwidth. In conventional design methodologies, PA designers need to preselect the optimum impedances, which normally follow non-Foster trajectories, and are thus impossible to achieve with passive matching networks for a wide frequency range. Instead, the proposed automatic design method directly synthesizes the matching networks to achieve the desired output power, efficiency, and gain performance, without forcing any impedance profiles. For that, load-pull data is interpolated using artificial neural networks, and an algorithm based on the simplified real frequency technique is used to obtain the matching networks. Finally, the method is validated with a single-ended PA implementation.

show abstract

Automated Deep Neural Learning-Based Optimization for High Performance High Power Amplifier Designs

Cited by 39 publications

References 28 publications

Optimization techniques for analog and RF circuit designs: an overview

Optimization techniques for analog and RF circuit designs: an overview

Improved Modeling of Microwave Structures Using Performance-Driven Fully-Connected Regression Surrogate

Automatic Methodology for Wideband Power Amplifier Design

Contact Info

Product

Resources

About