Predistortion device

Abstract

The disclosure relates to a predistortion device (100) for a non-linear PA device (101 ), comprising: an input terminal (102) for receiving an input signal (106); a predistortion filter (103), connected between the input terminal (102) and the non-linear PA device (101 ); a first delay element (105), coupled to the input terminal (102), and configured to delay the input signal (106) by a time delay D to provide a delayed input signal (104); and an adaptive filter unit (120), comprising an Adaptive Filter (107) having adjustable filter weights, and configured to filter the delayed input signal(104), and an adjusting unit (109), wherein the adjusting unit is configured to process an Adaptive Algorithm, based on the delayed input signal (104), to adjust the filter weights of the Adaptive Filter (107), and to provide both the Adaptive Filter (107) and the predistortion filter (103) with the same adjusted filter weights, wherein both, the Adaptive Filter (107) and the adjusting unit (109)are coupled to the first delay element (105) in order to receive the same delayed input signal (104).

Description

Technical field [0001] The present disclosure relates to predistortion devices for non-linear power amplifier (PA) devices. The invention also relates to a non-linear PA digital predistortion device used in a radio transmitter of a communication system. Background technique [0002] The PA linearization problem occurs in the following communication system equipment, in which the PA of the transmitter works in the saturation region, which provides high efficiency but suffers from nonlinear distortion. [0003] The problem of PA linearization is usually solved by applying predistortion to the PA input signal so that the cascaded output signal of the predistorter and the PA has no or almost no distortion. Since predistortion is usually implemented in the digital domain, that is, based on the use of digital signal processing principles, the technology is called digital predistortion and the predistortion device is called a digital predistorter (DPD). DPD produces a nonlinear transfer...

AI Technical Summary

Problems solved by technology

[0011] There are two main problems as indicated below which limit what is hereinafter referred to as traditional DPD cf. Figure 7 The efficiency of the DPD 700

[0012] First problem: compared to the case when both the adaptive filter 701 and the adaptive algorithm 713 will use the same input signal x(k) 702 see equation (5) below, the conventional direct learning DPD 700 The step size μ in the gradient search based adaptive algorithm used in max must have a small value, see equation (3) below

Smaller step size values ​​increase the duration of the transient response of the gradient search based adaptive algorithm, since the response increases if the step size decreases

[0013] Second question: Traditional direct learning DPD 700, see Figure 7 , the Recursive Least Square (RLS) adaptive algorithm, which is a more efficient algorithm compared with the gradient search-based adaptive algorithm, cannot be used because the RLS algorithm does not have parameters (such as step size), so the RLS algorithm becomes unstable in this architecture

Images