Nonlinear Identification of Continuous-Time Radio Frequency Power Amplifier Model

Mourad Djamai (1) , Smail Bachir (2) , Claude Duvanaud (2) , Guillaume Mercère (1)

1 LAII [Poitiers] - Laboratoire d'automatique et d'informatique industrielle [EA 1219]
2 SIC - SIGNAL-IMAGE-COMMUNICATION

In this paper, we present a three-step identification procedure for radio frequency Power Amplifier (PA) in the presence of nonlinear distortion which affect the modulated signal in the Radiocommunication transmission system. The proposed procedure uses a grey box model where PA dynamics are modelled with a MIMO continuous filter and the nonlinear characteristics are described as general polynomial functions, approximated by means of Taylor series. Using the baseband input and output data, model parameters are obtained by an iterative identification algorithm based on Output Error method. Initialization and excitation problems are resolved by an association of a new technique using initial values extraction with a multi-level binary sequence input exciting all PA dynamics. Finally, the proposed estimation method is tested and validated on experimental data.

Domaines: Electronique; Automatique / Robotique

Mourad Djamai, Smail Bachir, Claude Duvanaud, Guillaume Mercère. Nonlinear Identification of Continuous-Time Radio Frequency Power Amplifier Model. European Control Conference, Jul 2007, Kos, Greece. pp.ECC. ⟨hal-00782925⟩