Low-complexity FPGA implementation of Volterra predistorters for power amplifiers

Abstract

In this paper we present a FPGA design of a digital predistorter (DP) for power amplifiers (PAs) regarding memory effects. As model description the baseband Volterra series are utilized. A reduction of Volterra coefficients can be achieved using their symmetry properties. An advantage of our DP approach is a direct offline model identification, without need to analytically or iteratively calculate a PA model inverse. The DP implementation is very flexible and saves FPGA ressources. Our simulation and measurement results show a good linearization performance applying simple Volterra model structures. DP with and without memory are compared

Smart RF signal processing for advanced transmitters: From theory to practice

“…Fig. 3 shows the setup of the Smart RF prototype implementation, which follows the system models in [8,11]. As mentioned above, the experimental platform has two Tx-/Rx paths, each having an I/Q-MOD/DMOD.…”

“…The TRF370333 quadrature modulator from Texas Instruments 3 and a Siemens class-AB 85-W PA are used in the Tx path. As in [5,8,11], a digital IF down-conversion architecture is used in both feedback paths to avoid additional I/Q-imbalance. The corresponding IF frequency is 40 MHz.…”

“…pure digital BB compensation, e.g. predistortion [4,8], and digitally assisted hybrid compensation, e.g. feed-forward [9] or mixer IP2 calibration [10]).…”

“…In our previous works [8,11,12], efficient mechanisms were proposed for digital compensation of I/Q-imbalance, DCoffset and PA nonlinearity. In this paper, we describe the corresponding prototype implementation, which demonstrates the effectiveness of the proposed compensation mechanisms in realtime.…”

“…While inband distortion increases the Error Vector Magnitude (EVM) of the Tx signal, spectral regrowth causes adjacent channel interference. Moreover, in broadband systems, the nonlinearity has memory effects, giving more challenge to the modeling and compensation [8,12]. Fig.…”