Modeling and predistortion of envelope tracking power amplifiers using a memory binomial model

Abstract

Nowadays envelope tracking (ET) is considered one of the most appealing techniques for the efficiency enhancement of RF power amplifiers (PAs), but it also introduces a number of additional challenges for the system simulation and implementation. In this context, this paper aims to provide a new behavioral model capable of an improved performance when used for the modeling and predistortion of RF PAs deployed in ET transceivers. The proposed solution consists in a 2D behavioral model having as a dual-input the PA complex baseband envelope and the modulated supply waveform, peculiar of the ET case. The model definition is based on binomial series, hence the name of memory binomial model (MBM). The MBM is here applied to measured data-sets acquired from an ET measurement set-up. When used as a PA model the MBM showed an NMSE (Normalized Mean Squared Error) as low as -40 dB and an ACEPR (Adjacent Channel Error Power Ratio) below -51 dB. The simulated predistortion results showed that the MBM can improve the compensation of distortion in the adjacent channel of 5.8 dB and 5.7 dB compared to a memory polynomial predistorter (MPPD). The predistortion performance in the time domain showed an NMSE improvement of 2.5 dB against the MPPD.

Memory models for behavioral modeling and digital predistortion of envelope tracking power amplifiers

“…First, a 2D polynomial structure with memory is proposed, named Memory Binomial Model after its structure inspired by binomial power series [20]. Secondly a different approach is proposed, where the Cann model for RF PAs is extended including the new independent variable V ENV and defining new functions for modeling of dynamic AM/AM and AM/PM distortion effects [21].…”

“…The comparison among the Generalized Memory Polynomial [28], the Envelope Memory Polynomial [29] and the static polynomial showed that the addition of memory compensation capabilities was not able to improve the PA ACPR, while a small improvement was observed in terms of NMSE and EVM compared to the static polynomial DPD. In this journal we present linearization results obtained using a 2D-DPD architecture based on the same concept [20]. The proposed predistortion scheme is showed in Fig.…”

“…MPM performance evaluated in terms of ACEPR versus the maximum nonlinear order P and the memory depth M. The model is applied to an ET WCDMA data-set measured using the ET measurement set-up described in this paper. Table 1 Comparison of behavioral models for ET PAs presented in [21,20]. Fig.…”

“…This approach was used in[20]. Quadrature Modulators (QM) are used to represent ideal up-conversion and down-conversion.…”

Modeling envelope-tracking RF amplifiers

“…The reference [1] offers an exhaustive list referencing some of the most interesting works. Some recent ET publications focus on the compensation of the memory effects [1][2][3][4]. Unfortunately, it's not always easy to implement such compensations, because of the difficulties in characterizing the memory effects.…”