CN103401511A - Power amplifier pre-distortion method used for real-time system - Google Patents

Abstract

The invention discloses a power amplifier pre-distortion method used for a real-time system. By means of the power amplifier pre-distortion method, good linearization can be conducted on a non-linear power amplifier with memory effects. On the basis of an existing Hammerstein algorithm and in order to solve the problems that the existing algorithm is large in calculation amount and long in calculation time, the existing algorithm is simplified through the combination of the ideas of a lookup table. The algorithm which is simplified greatly lowers the time needed by calculation on the basis that the performances of the existing algorithm can be maintained basically, and therefore the algorithm can be applied to systems which have high requirements for real-time performance.

Description

A kind of power amplifier predistortion method for Real-time System

Technical field

The present invention relates to the power amplifier predistortion technology of power amplifier, particularly the base band digital predistortion technology of memory nonlinear power amplifier arranged.

Background technology

Power amplifier is as one of requisite device in communication system, and its effect is that modulation signal is amplified to certain power, then at front end, by antenna, effectively launches.Power amplifier always shows nonlinear characteristic, and power amplifier just can reach maximum power efficiency always in inelastic region or during nearly saturation region.The communication technology in the past mostly adopts the modulation technique of permanent envelope, to the linearity of power amplifier, is not very high.Therefore, power amplifier can work in inelastic region or nearly saturation region, to reach higher power efficiency.But along with the continuous evolution and development of the communication technology, because number of wireless users constantly increases, the critical shortage of frequency spectrum resource, increasing non-constant enveloped modulation technology such as quadrature amplitude modulation (QAM), OFDM (OFDM) etc. is used.Although their application can reach higher spectrum efficiency, high peak-to-average force ratio (PAPR) signal that it produces is non-linear very responsive to transmission channel, particularly power amplifier but.During the nonlinear power amplifier of these non-constant enveloped modulation signals processes, can produce simultaneously inband distortion and out-of-band distortion.And, when the broader bandwidth of input signal, through power amplifier, also can produce memory effect.Therefore, under the prerequisite that there is no the other technologies application, the systematic function preferably under the high power efficiency of power amplifier and high linearity just becomes implacable contradiction, need to do to consider with compromise between them.

Because power amplifier cost in wireless communication system is higher, it is obviously inadvisable that the efficiency of sacrificing in a large number power amplifier increases the method for the linearity and systematic function.Therefore, can adopt a series of linearization technique to make its linearisation to non-linear the compensating of power amplifier.Pre-distortion method is wherein a kind of, and it refers to add an opposite predistortion circuit of nonlinear characteristic with it before nonlinear power amplifier, compensates the nonlinear distortion of power amplifier.Namely first with input signal by a nonlinear device (predistorter), and then enter power amplifier and amplify and export, thereby make whole system show linear character.And in pre-distortion method, pre-distortion algorithms is core wherein, by the calculating of pre-distortion algorithms, just can obtain input signal required to the pre-distorted signals of the non-linear work of power amplifier with compensation before entering power amplifier, the precision of this algorithm and efficiency are directly connected to the quality of predistortion performance.The present invention is based on the Hammerstein pre-distortion algorithms that provides in the paper " Experimental results of power amplifiers linearization using adaptive baseband digital predistortion " of Cottais.E., for long problem of the rooting process computation time in this algorithm, with the method for look-up table, the rooting process is simplified, greatly shortened the computing time of this algorithm, this algorithm can be applied in real-time system.

Summary of the invention

Main purpose of the present invention is large for existing Hammerstein algorithm amount of calculation and calculating long problem consuming time, and the thought that combines look-up table is simplified algorithm, and a kind of power amplifier predistortion method for Real-time System is provided.

Existing Hammerstein algorithm is as follows:

According to Fig. 1, the output s (nT) of memoryless nonlinear block and system output y (nT) can be expressed as respectively:

$s\left(\mathrm{nT}\right)={\mathrm{\Σ}}_{j=0}^N{b}_{2j+1}e\left(\mathrm{nT}\right){\left|e\left(\mathrm{nT}\right)\right|}^{2j}---\left(1\right)$

$y\left(\mathrm{nT}\right)={\mathrm{\Σ}}_{i=0}^{P-1}{h}_{i}s\left[(n-i)T\right]---\left(2\right)$

Formula (2) substitution formula (1) can be obtained:

$y\left(\mathrm{nT}\right)={\mathrm{\Σ}}_{i=0}^{P-1}{\mathrm{\Σ}}_{j=0}^N{a}_i{b}_{2j+1}e\left[(n-i)T\right]{\left|e\right[(n-i)T\left]\right|}^{2j}=G_{0}x\left(\mathrm{nT}\right)---\left(3\right)$

Wherein G0 represents the perfect Gain.

Further, we can be divided into output y (nT) static s (nT) and dynamic d (nT) two parts, and they are expressed as follows respectively:

y(nT)＝s(nT)+d(nT) （4）

$s\left(\mathrm{nT}\right)=y\left(\mathrm{nT}\right){|}_{i=0}={\mathrm{\Σ}}_{j=0}^N{a}_0{b}_{2j+1}e\left(\mathrm{nT}\right){\left|e\left(\mathrm{nT}\right)\right|}^{2j}---\left(5\right)$

$d\left(\mathrm{nT}\right)={\mathrm{\Σ}}_{i=1}^{P-1}{\mathrm{\Σ}}_{j=0}^N{a}_i{b}_{2j+1}e\left[(n-i)T\right]{\left|e\right[(n-i)T\left]\right|}^{2j}---\left(6\right)$

Static part s (nT) only depends on the input of current time nT, the input before dynamic part d (nT) depends on except current.

According to formula (3), formula (4), formula (5), we can obtain:

$s\left(\mathrm{nT}\right)={\mathrm{\Σ}}_{j=0}^N{a}_0{b}_{2j+1}e\left(\mathrm{nT}\right){\left|e\left(\mathrm{nT}\right)\right|}^{2j}=G_{0}x\left(\mathrm{nT}\right)-d\left(\mathrm{nT}\right)---\left(7\right)$

With pre-distorted signals e (nT) replace with e (nT)=| e (nT) | e ^{JArg (e (nT))}, formula (7) can be write as:

$e^{\mathrm{jArg}\left(e\left(\mathrm{nT}\right)\right)}{\mathrm{\Σ}}_{j=0}^N{a}_0{b}_{2j+1}{\left|e\left(\mathrm{nT}\right)\right|}^{2j+1}=G_{0}x\left(\mathrm{nT}\right)-d\left(\mathrm{nT}\right)---\left(8\right)$

At nT constantly, system input x (nT) and the d (nT) that is determined by input before are known, so corresponding each e of the equation left side (nT) can calculate.Whole process can be divided following steps:

The first step: the amplitude that calculates e (nT).Can ask minimum positive real root to formula (9), result of calculation is | e (nT) |.

$F_a\left(n\right)=\left|{\mathrm{\Σ}}_{j=0}^N{a}_0{b}_{2j+1}{\left|e\left(\mathrm{nT}\right)\right|}^{2j+1}\right|=|G_{0}x\left(\mathrm{nT}\right)-d\left(\mathrm{nT}\right)|---\left(9\right)$

Second step: the phase place that calculates e (nT) according to formula (10).

$\mathrm{Arg}\left(e\left(\mathrm{nT}\right)\right)=\mathrm{Arg}\left\{\frac{G_{0}x\left(\mathrm{nT}\right)-d\left(\mathrm{nT}\right)}{{\mathrm{\Σ}}_{j=0}^N{a}_0{b}_{2j+1}{\left|e\left(\mathrm{nT}\right)\right|}^{2j+1}}\right\}---\left(10\right)$

Purpose of the present invention is achieved through the following technical solutions:

A kind of power amplifier predistortion method for Real-time System, its realization comprises the steps:

Step (1): when input signal enters the Hammerstein pre-distortion system, according to input signal, first set up a look-up table, the input and output side of this look-up table is being stored respectively equation in algorithm known $F_a\left(n\right)=\left|{\mathrm{\Σ}}_{j=0}^N{a}_0{b}_{2j+1}{\left|e\left(\mathrm{nT}\right)\right|}^{2j+1}\right|=|G_{0}x\left(\mathrm{nT}\right)-d\left(\mathrm{nT}\right)|$ Value and the root of equation right-hand member | e (nT) |.

Step (2): need not, to this equation rooting, only need this table of inquiry just can find out required equattion root, i.e. the amplitude of pre-distorted signals when input signal enters pre-distortion system | e (nT) |.

Step (3): according to the pre-distorted signals that inquires in look-up table | e (nT) |, according to formula $\mathrm{Arg}\left(e\left(\mathrm{nT}\right)\right)=\mathrm{Arg}\left\{\frac{G_{0}x\left(\mathrm{nT}\right)-d\left(\mathrm{nT}\right)}{{\mathrm{\Σ}}_{j=0}^N{a}_0{b}_{2j+1}{\left|e\left(\mathrm{nT}\right)\right|}^{2j+1}}\right\}$ Calculate the phase place of pre-distorted signals.

Step (4): according to formula e (nT)=| e (nT) | e ^{JArg (e (nT))}Just can obtain pre-distorted signals, the pre-distorted signals after input signal is adjusted enters power amplifier, and that the memory nonlinear power amplifier is shown as is linear.

In said method, step (1) is described when input signal enters the Hammerstein pre-distortion system, substitute rooting process in original algorithm by a look-up table of setting up according to input signal, rooting process the most consuming time in existing algorithm is simplified, also can greatly reduce the computing time of pre-distorted signals, and the Hammerstein pre-distortion algorithms after improvement can be applied in the higher system of requirement of real-time.

Compared with prior art, the present invention has the following advantages and beneficial effect:

Computing time faster: by a look-up table of setting up according to input signal, substitute rooting process in original algorithm, rooting process the most consuming time in existing algorithm is simplified, the method on the basis of substantially having kept original algorithm performance, has been reduced the computing time of pre-distorted signals greatly.

Description of drawings

Fig. 1 is the schematic diagram of digital pre-distortion.

Fig. 2 is Hammerstein predistortion principle figure.

Fig. 3 is the look-up table schematic diagram.

Embodiment

The present invention is described further below in conjunction with drawings and Examples, but the scope of protection of present invention is not limited to the scope of execution mode statement.

Input signal used is the WCDMA signal through the 16QAM modulation.As Fig. 1, dotted line has represented the linear response of the desired power amplifier that obtains, and solid line has represented that actual power amplifier is non-linear.Input signal when we postulated point A represents that Mean Input Power is P1, P3 is the average power of output signal so.And our desired average output power is P4 when in fact Mean Input Power is P1, and therefore from the angle of power output, the signal after our needed predistortion should be a B, and its corresponding Mean Input Power is P2.But after a B ultrasonic went out the saturation region of power amplifier Nonlinear Characteristic Curve, digital pre-distortion just can't be revised non-linear fully.And along with A movement from left to right, non-linear due to power amplifier, some B also can increase than P2/P1 gradually with the input power of an A.

Hammerstein predistortion principle figure as Fig. 2.The memory power amplifier carrys out modeling with the Wiener model, namely uses memoryless nonlinear block in a Linear Memory module (FIR) back cascade to represent to have the memory nonlinear power amplifier.Then we just can carry out predistortion to power amplifier with the Hammerstein model with Wiener Cascade reversed in order.Predistortion module and power amplifier can show reciprocal characteristic, whole system are shown as linear.

Concrete enforcement comprises the steps:

Step (1):, according to the WCDMA signal of input, first build a look-up table that replaces existing algorithm rooting process.F is classified in the input of this look-up table as _a(n)=| G ₀X (nT)-d (nT) | value, output is classified as | e (nT) |., according to Fig. 1 schematic diagram, can calculate | e (nT) | excursion be [0, min (1.9max (| x (nT) |), V _{In_sat})], V wherein _{In_sat}Input voltage when reaching capacity.Will | e (nT) | be stored in the output row LUT of look-up table behind 1000 unit of uniform distribution in its excursion _outIn, then according to formula $F_a\left(n\right)=\left|{\mathrm{\Σ}}_{j=0}^N{a}_0{b}_{2j+1}{\left|e\left(\mathrm{nT}\right)\right|}^{2j+1}\right|=|G_{0}x\left(\mathrm{nT}\right)-d\left(\mathrm{nT}\right)|$ Calculate corresponding F _aValue is stored in the input row LUT of look-up table _inIn, institute's look-up table of building such as Fig. 3.

Step (2): need not, to this equation rooting, only need this table of inquiry just can find out required equattion root, i.e. the amplitude of pre-distorted signals when input signal enters pre-distortion system | e (nT) |.

Step (3): according to the pre-distorted signals that inquires in look-up table | e (nT) |, according to formula $\mathrm{Arg}\left(e\left(\mathrm{nT}\right)\right)=\mathrm{Arg}\left\{\frac{G_{0}x\left(\mathrm{nT}\right)-d\left(\mathrm{nT}\right)}{{\mathrm{\Σ}}_{j=0}^N{a}_0{b}_{2j+1}{\left|e\left(\mathrm{nT}\right)\right|}^{2j+1}}\right\}$ Calculate the phase place of pre-distorted signals.

Step (4): according to formula e (nT)=| e (nT) | e ^{JArg (e (nT))}Just can obtain pre-distorted signals, the pre-distorted signals after input signal is adjusted enters power amplifier, and that the memory nonlinear power amplifier is shown as is linear.

Employing is improved one's methods and is compared original Hammerstein algorithm, Adjacent Channel Power Ratio rising 0.3dB, little on the systematic function impact, but amount consuming time but shortened to about 0.5 second from original about 1.9 seconds, greatly strengthened the real-time performance of system, this algorithm can be applied among real-time system.

Above-described specific embodiment; purpose of the present invention, technical scheme and beneficial effect have been carried out further detailed description; institute is understood that; the foregoing is only specific embodiments of the invention; not in order to limit scope of the present invention; any those skilled in the art, the equivalent variations of making under the prerequisite that does not break away from design of the present invention and principle and modification, all should belong to the scope of protection of the invention.

Claims (2)

1. power amplifier predistortion method that is used for Real-time System, in the Hammerstein pre-distortion system, when applying existing pre-distortion algorithms pre-distorted signals being calculated, replace with look-up table the rooting process in this algorithm.

2. power amplifier predistortion method that is used for Real-time System is characterized in that comprising the following steps:

Step (1): when input signal enters the Hammerstein pre-distortion system, according to input signal, first set up a look-up table, the input and output side of this look-up table is being stored respectively equation in algorithm known $F_a\left(n\right)=\left|{\mathrm{\Σ}}_{j=0}^N{a}_0{b}_{2j+1}{\left|e\left(\mathrm{nT}\right)\right|}^{2j+1}\right|=|G_{0}x\left(\mathrm{nT}\right)-d\left(\mathrm{nT}\right)|$ Value and the root of equation right-hand member | e (nT) |.

Step (2): need not, to this equation rooting, only need this table of inquiry just can find out required equattion root, i.e. the amplitude of pre-distorted signals when input signal enters pre-distortion system | e (nT) |.

Step (3): according to the pre-distorted signals that inquires in look-up table | e (nT) |, according to formula $\mathrm{Arg}\left(e\left(\mathrm{nT}\right)\right)=\mathrm{Arg}\left\{\frac{G_{0}x\left(\mathrm{nT}\right)-d\left(\mathrm{nT}\right)}{{\mathrm{\Σ}}_{j=0}^N{a}_0{b}_{2j+1}{\left|e\left(\mathrm{nT}\right)\right|}^{2j+1}}\right\}$ Calculate the phase place of pre-distorted signals.

Step (4): according to formula e (nT)=| e (nT) | e ^{JArg (e (nT))}Obtain pre-distorted signals, the pre-distorted signals after input signal is adjusted enters power amplifier, and that the memory nonlinear power amplifier is shown as is linear.

Images