Optimal Sizing of Two-Stage Cascaded Sparse Memory Polynomial Model for High Power Amplifiers Linearization

Wang, Siqi; Hussein, Mazen Abi; Venard, Olivier; Baudoin, Geneviève

doi:10.1109/tmtt.2018.2838126

Cited by 17 publications

(16 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…where the coefficientsγ i are estimated using (5). The algorithm ends when a given stall condition is reached.…”

Section: B Compressed-sensing-based Methodsmentioning

confidence: 99%

“…These methods determine the optimal full models with minimal nonlinearity orders and memory depths for an acceptable linearization performance. Sparse models are considered in [5] for the case of 2-stage cascade MP models. Determining a sparse model structure is equivalent to pruning a complete model whose nonlinearity order and memory depth are large.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Optimization of a pruned 2-D Digital Predistortion Model Structure for Power Amplifiers Linearization

Wang

Roger

Lelandais-Perrault

2019

2019 17th IEEE International New Circuits and Systems Conference (NEWCAS)

Self Cite

View full text Add to dashboard Cite

Two-dimensional digital predistortion (2-D DPD) is one of the most commonly used approaches to linearize the concurrent dual-band radio frequency (RF) power amplifiers (PA). This paper explores the use of a hill-climbing optimization heuristic to determine the optimal structure for 2-D DPD. To improve convergence and reduce computation time, we propose a new parameterization for the 2-D DPD model. The proposed search criterion is based on the generalized information criterion, which represents the trade-off between the DPD linearization accuracy in both bands and the model complexity. A comparison against a compressed-sensing-based method is also made. The effectiveness of the proposed method is validated with two 20 MHz long-term-evolution (LTE) signals on two carriers with 100 MHz frequency separation.

show abstract

“…where the coefficientsγ i are estimated using (5). The algorithm ends when a given stall condition is reached.…”

Section: B Compressed-sensing-based Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Optimization of a pruned 2-D Digital Predistortion Model Structure for Power Amplifiers Linearization

Wang

Roger

Lelandais-Perrault

2019

2019 17th IEEE International New Circuits and Systems Conference (NEWCAS)

Self Cite

View full text Add to dashboard Cite

show abstract

“…Based on this approach, it is developed a MIMO system model using a transmission antenna selection technique (AST) algorithm of maximum ratio combining (MRC) for antenna selection, which is established to maximize the received SNR [39]. In general, nonlinear behavior effects and the parameter estimation can be obtained by different basis functions, such as conventional polynomials [40], orthogonal polynomials [41], or cubic-splines [30], [42]. According to (19), a gain factor can be expanded as in (20) in terms of a one-dimensional cubic-spline basis function to the gain…”

Section: A Cubic-spline Analytical Formulationmentioning

confidence: 99%

RF-PA Modeling of PAPR: A Precomputed Approach to Reinforce Spectral Efficiency

2020

View full text Add to dashboard Cite

This paper introduces the measurement/modeling of a nonlinear RF power amplifier (PA) model extraction of spectral overlaps and peak-to-average power ratio (PAPR). The modeling consists of a weighted cubic-spline basis approach that preserves a relationship by its generic asymptotic properties under adequate PAPR regime to estimate signaling conditions. To jointly provide a data-driven model a fieldprogrammable gate array (FPGA) testbed is proposed, in which a precomputed approach with deterministic signals is used to perform: (i) parameter estimation; (ii) adequate PAPR levels; (iii) reinforcement of sparsity data with extrapolation fitting; and (iv) FPGA implementation. Moreover, a technique for parameter identification is introduced to provide insights of a digital predistortion (DPD) PA model extraction with cubicspline to efficiently improve the linearization performance. A theoretical analysis that states on the benefits from coefficients precomputation to preserve multiple-input multiple-output (MIMO) relationship with antenna selection technique estimation, is also proposed. Also, a Cholesky FPGA algorithm implementation to matrix inversion is validated, which aims to show the good numerical and computational complexity for up to 64 × 64 MIMO arrays. Experimental results prove a good accuracy and close agreement between the modeling and estimation yielding a reliable model with a little overfitting.

show abstract

“…Various DPD models have been developed, such as memory polynomial (MP) [5] based on Volterra series, generalized memory polynomial (GMP) [6], and dynamic-deviation-reduction (DDR) [7]. Block-oriented non linear (BONL) systems [8] have also been studied.…”

Section: Introductionmentioning

confidence: 99%

Augmented Iterative Learning Control for Neural-Network-Based Joint Crest Factor Reduction and Digital Predistortion of Power Amplifiers

Wang

Roger

Sarrazin

et al. 2020

IEEE Trans. Microwave Theory Techn.

Self Cite

View full text Add to dashboard Cite

Digital predistorsion (DPD) is a commonly used approach to compensate for the power amplifiers (PA) nonlinearities and memory effects as well as to improve its power efficiency. To alleviate the restriction brought by high peak-to-average power ratio (PAPR) of input signals, crest factor reduction (CFR) is needed for higher efficiency. In modern communication systems, power of transmitted signals gets lower, which makes complexities of CFR and DPD become non-negligible. This paper proposes a new approach to realize a joint CFR and DPD model using neural networks (NN). The modeling accuracy is guaranteed by a new proposed augmented iterative learning control (AILC) algorithm for the NN training signals. Compared with conventional ILC, the proposed AILC is shown more robust according to simulation and experimental results. The proposed AILC-based NN-CFRDPD is experimentally evaluated on different test benches.

show abstract

Optimal Sizing of Two-Stage Cascaded Sparse Memory Polynomial Model for High Power Amplifiers Linearization

Cited by 17 publications

References 21 publications

Optimization of a pruned 2-D Digital Predistortion Model Structure for Power Amplifiers Linearization

Optimization of a pruned 2-D Digital Predistortion Model Structure for Power Amplifiers Linearization

RF-PA Modeling of PAPR: A Precomputed Approach to Reinforce Spectral Efficiency

Augmented Iterative Learning Control for Neural-Network-Based Joint Crest Factor Reduction and Digital Predistortion of Power Amplifiers

Contact Info

Product

Resources

About