Nonlinear behavioral modeling of power amplifiers using radial-basis function neural networks

Isaksson, Magnus; Wisell, David; Rönnow, Daniel

doi:10.1109/mwsym.2005.1517128

Cited by 20 publications

(10 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The proposed model was applied to data sets that have been used extensively for model comparison and development by the authors and co-workers [9,11,13,[17][18][19]. This has the advantage that results exist for many different models to which the performance of the proposed model can be compared.…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Derivation of a behavioral RF power amplifier model with low normalized mean-square error

Wisell

Isaksson

2007

2007 European Microwave Integrated Circuit Conference

View full text Add to dashboard Cite

In this paper a behavioral power amplifier model with substantially lower normalized mean-square error than the well-known and widely used parallel Hammerstein or tapped delay line model is presented. This is significant since the parallel Hammerstein model currently must be considered to be the baseline for behavioral power amplifier modeling efforts. The proposed model does also have at least as low total error as any other model known to the authors for measurements on class-AB LDMOS power amplifiers. Further, the proposed model exhibits an equal or lower out-of-band error than the parallel Hammerstein model. The parameters of the model can be extracted using straightforward system identification techniques. Together this makes the proposed model a strong candidate to replace the parallel Hammerstein model as the model of choice for behavioral power amplifier modeling for many applications.

show abstract

Section: Resultsmentioning

confidence: 99%

“…In practice, considerable simpler models, such as the parallel Hammerstein (PH) model has at least as good performance for important goodness measures such as the adjacent channel error power ratio (ACEPR) [9]. The PH model was first proposed for amplifier modeling in [10], and is now widely used both for modeling and DPD of PAs, see e.g.…”

Section: Introductionmentioning

confidence: 99%

Derivation of a behavioral RF power amplifier model with low normalized mean-square error

Wisell

Isaksson

2007

2007 European Microwave Integrated Circuit Conference

View full text Add to dashboard Cite

show abstract

“…For instance, the input history of I in (t) and Q in (t) has been used to train TDNN to learn I out (t) and Q out (t) [16]. As well, an approximate technique to directly learn the output envelope (||y(t)||) and phase (ffy(t)) using only the input envelope (||x(t)||) has been shown using time-delay radial basis function networks (RBF) [17]. Here, we apply the RNN to allow a more complete and compact representation of the inputoutput envelope due to the presence of feedback.…”

Section: Rnn Envelope Modeling Of Power Amplifiersmentioning

confidence: 99%

“…Time-domain solvers can obtain wideband behavior of EM structures more efficiently than conventional frequency-domain solvers. On the other hand, nonlinear circuit modeling, such as power amplifier behavior modeling continues to be an important topic [14][15][16][17]. Envelope simulation of PA is essential for simulating modulated signals for wireless system design.…”

Section: Introductionmentioning

confidence: 99%

Automated time domain modeling of linear and nonlinear microwave circuits using recurrent neural networks

Sharma

Zhang

2008

Int J RF and Microwave Comp Aid Eng

View full text Add to dashboard Cite

In this article, a recurrent neural network (RNN) method is employed for dynamic time-domain modeling of both linear and nonlinear microwave circuits. An automated RNN modeling technique is proposed to efficiently determine the training waveform distribution and internal RNN structure during the offline training process. This technique extends a recent automatic model generation (AMG) algorithm from frequency-domain model generation to dynamic time-domain model generation. Two types of applications of the algorithm are presented, transient electromagnetic (EM) behavior modeling of microwave structures, and time-domain envelope modeling of power amplifiers (PA). For transient EM modeling, we consider EM structures with varying material and geometrical parameters. AMG automatically varies the EM structural parameters during training and drives time-domain EM simulators to generate necessary amount of data for RNN to learn. AMG aims to model the transient behavior with minimum RNN order while satisfying accuracy requirements. In modeling PA behavior, an envelope formulation is used to specifically learn the AM/AM and AM/PM distortions due to third-generation (3G) digital modulation input. The RNN PA model is able to model these time domain distortions after training and can accurately model the amplifier behavior in both time (AM/AM, AM/PM) and frequency (spectral re-growth).

show abstract

“…The time domain metrics mainly include the normalized mean squared error (NMSE), the memory effects ratio (MER), and the memory effects modeling ratio (MEMR) [9] [10]. In frequency domain, the adjacent channel error power ratio (ACEPR), and the weighted error-tosignal power ratio (WESPR) are among the metrics commonly used to evaluate the performance of behavioral models [11] [12]. Prior work reported that, among the various performance evaluation criteria, the NMSE and the WESPR metrics are the strongest candidates for capturing the in-band and out-of-band errors, respectively [9] [12].…”

Section: Introductionmentioning

confidence: 99%

Accurate estimation of memory polynomial model dimension for dynamic nonlinear RF power amplifiers

Hanzala

Sharawi

Hammi

2015

2015 IEEE 10th Conference on Industrial Electronics and Applications (ICIEA)

View full text Add to dashboard Cite

In this paper, the dimension estimation of memory polynomial based power amplifiers' behavioral models is investigated. Comprehensive study of state of the art performance assessment metrics is carried out. To ensure the generality of the conclusions, the study is experimentally validated using three power amplifiers prototypes designed using various transistors technologies and amplifier topologies. The results show that the performance assessment metrics are equally reliable for the estimation of the device under test nonlinearity order. However, they are inconsistent when used to estimate its memory depth. This limitation is circumvented by applying memoryless post-compensation technique during the memory depth estimation. The proposed approach demonstrates that the dimension estimation can be made objective and independent of the performance assessment metric.

show abstract

Nonlinear behavioral modeling of power amplifiers using radial-basis function neural networks

Cited by 20 publications

References 10 publications

Derivation of a behavioral RF power amplifier model with low normalized mean-square error

Derivation of a behavioral RF power amplifier model with low normalized mean-square error

Automated time domain modeling of linear and nonlinear microwave circuits using recurrent neural networks

Accurate estimation of memory polynomial model dimension for dynamic nonlinear RF power amplifiers

Contact Info

Product

Resources

About