CN102118334B - Method and device for processing digital predistortion - Google Patents

Abstract

The invention discloses a method and device for processing digital predistortion. The method comprises the steps of: respectively filtering a radio-frequency signal of each frequency band; collecting the nonlinear intermodulation distortion of each frequency band; synchronizing the nonlinear intermodulation distortion of each frequency band with the input signal of the frequency band; and updating digital predistortion factors by utilizing the synchronized signals. In the invention, through respectively filtering the radio frequency of each frequency band and carrying out common digital predistortion processing on digits, a high efficient digital predistortion processing effect of the digital predistortion can be achieved under the condition of low-complicated hardware.

Description

Digital pre-distortion processing method and device

Technical Field

The present invention relates to signal processing technologies, and in particular, to a digital predistortion processing method and apparatus.

Background

The frequency band according to China Mobile planning is divided into four frequency bands of F/A/D/E, which are expressed as follows:

table: F/A/D/E quad band planning

In the future, a requirement of a broadband RRU (Radio Remote Unit) is to meet a common antenna of the four frequency bands and a common hardware platform of the a + F frequency bands, so that hardware cost and hardware design complexity are greatly saved. However, the industry only applies 30MA + F at present, and no formal product exists for continuous 50MHz of the D frequency band/E frequency band.

The application of the wideband RRU will inevitably become the mainstream by adopting a mixed mode, that is, a TD (Time Division) and LTE (Long Term Evolution) common platform. The use of the E band 50M and subsequent D bands introduces a need for greater bandwidth. The hybrid mode requires a TD and TD-LTE (TD-SCDMA Long Term Evolution, TD-SCDMA Long Term Evolution; TD-SCDMA time Division Synchronized Code Division Multiple Access) common platform, and for 50M band applications, it is not only 2 LTE with 20M and 1 TD sharing band with 10M. Or all LTE signals, e.g., 2 LTE of 20M and 1 LTE of 10M, share the 50MHz band.

The broadband is a clear development direction of carrier frequency technology in the industry, and relates to the whole industry chain of devices (intermediate frequency chips), power amplifiers (broadband power amplifier tubes), complete machines and the like. The reason is that operator wireless cellular network multi-band deployment and LTE technology introduce a large boost to the single device supported bandwidth, which is briefly described below.

Mainstream network operators (VDF/TIM/Orange/AT & T) RRU broadband requirements:

european and american mainstream operators have put strong demands on the broadband of the RRU radio frequency module many years ago. The goal is for a single radio frequency path to achieve a bandwidth of 200MHz or higher. A plurality of narrow-band radio frequency modules are not needed to be spliced, so that power and carrier resources between frequency bands are not favorably shared, the reliability of multiple components is poor, and the size/weight is large; or broadband processing of continuous signals above 50 MHz.

The broadband RRU technology comprises two technical schemes:

(1) the technical scheme for expanding the single-frequency band working bandwidth is that the current supportable 30MHz working bandwidth is expanded to 40-50 MHz, for example, a continuous 50 MHx-60 MHz is adopted in a D/E frequency band;

(2) the multi-band adopts a single span power amplification mode. That is, F + a shares a power amplifier, which determines that FA merges in the digital part.

Meanwhile, the average output power spectral density is generally supported to be 10W/10MHz in the industry, and TD-SCDMA and TDD-LTE are basically consistent, so that 50W is required for the signal power of a signal output with a 50MHz bandwidth, and a DPD (Digital PreDistortion) technology is necessarily adopted for outputting such a large signal power, so that how to utilize the DPD technology in a wideband RRU is also a difficult problem in the industry.

The defects of the prior art are as follows:

the feedback problem is difficult to solve in the prior art, for example, the feedback of a + F generally needs to adopt a ZIF (Zero-IF, Zero intermediate frequency) mode or adopt a 500MHZ non-real-time ADC (Analog-digital converter) to perform signal acquisition. However, the problem of I/Q imbalance is generally difficult to solve by adopting ZIF, and signal acquisition is carried out by adopting a 500MHZ non-real-time ADC, so that on one hand, the ADC cannot be shared with a normal receiving channel, and on the other hand, the price is relatively high, so that the price of the whole RRU product is low.

Disclosure of Invention

The technical problem to be solved by the present invention is to provide a digital predistortion processing method and apparatus, which are used to simplify RRU design and simultaneously ensure DPD performance in a wideband RRU.

The embodiment of the invention provides a digital predistortion processing method, which comprises the following steps:

respectively filtering the radio frequency signals of each frequency band;

collecting nonlinear intermodulation distortion of each frequency band;

synchronizing the nonlinear intermodulation distortion of each frequency band with the input signal of the frequency band;

and updating the DPD coefficient by using the synchronized signal.

The invention also provides a digital predistortion processing device, which comprises:

the filtering module is used for respectively filtering the radio frequency signals of each frequency band;

the acquisition module is used for acquiring the nonlinear intermodulation distortion of each frequency band;

the synchronization module is used for synchronizing the nonlinear intermodulation distortion of each frequency band with the input signal of the frequency band;

and the updating module is used for updating the DPD coefficient by using the synchronized signal.

The invention has the following beneficial effects:

in the technical scheme provided by the embodiment of the invention, the DPD can obtain a high-efficiency DPD processing effect under a low-complexity hardware condition by respectively filtering the radio frequency of each frequency band and digitally processing the common DPD.

Drawings

FIG. 1 is a schematic flow chart illustrating an implementation of a digital pre-distortion processing method according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of an F + A signal output by a power amplifier according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of an F-band signal after an IF feedback anti-aliasing filter according to an embodiment of the present invention;

FIG. 4 is a schematic diagram of an A-band signal after an IF feedback anti-aliasing filter according to an embodiment of the present invention;

FIG. 5 is a schematic diagram illustrating amplitude comparison after synchronous amplitude adjustment of the intermediate frequency signal and the feedback signal according to an embodiment of the present invention;

FIG. 6 is a schematic structural diagram of a digital pre-distortion processing apparatus according to an embodiment of the present invention;

FIG. 7 is a block diagram of a digital pre-distortion processing apparatus according to an embodiment of the present invention;

FIG. 8 is a schematic diagram of a second embodiment of a digital pre-distortion processing apparatus;

fig. 9 is a schematic diagram of inter-modulation information of the F band and the a band in the embodiment of the present invention;

fig. 10 is a diagram illustrating a DPD effect of splitting or combining according to an embodiment of the present invention;

fig. 11 is a schematic diagram of frequency spectrums before and after pre-distortion in the embodiment of the invention.

Detailed Description

The following describes embodiments of the present invention with reference to the drawings.

The embodiment of the invention provides a design scheme of a DPD hardware channel aiming at cross-band signal feedback. In the scheme, the radio frequency signals are filtered respectively, then nonlinear intermodulation distortion of respective frequency bands is acquired, filtering is carried out from the respective frequency bands to the respective frequency bands for acquisition, and after synchronization, the DPD coefficient is updated.

Fig. 1 is a schematic flow chart of an implementation of the digital predistortion processing method, as shown in the figure, the method may include the following steps:

step 101, respectively filtering radio frequency signals of each frequency band;

step 102, collecting nonlinear intermodulation distortion of each frequency band;

103, synchronizing the nonlinear intermodulation distortion of each frequency band with the input signal of the frequency band;

and step 104, updating the DPD coefficient by using the synchronized signal.

However, through the nature and the effects of the implementation of the embodiments disclosed in the specification, those skilled in the art should also know how to implement the embodiments in other combinations, and the processing in the F/a band is only used to teach those skilled in the art how to implement the present invention specifically, but it is not intended that the technical solution provided by the embodiments of the present invention can only be used for the processing in the F/a band, and it can be determined that equivalent technical means are used in corresponding band combinations to achieve the same effect in the implementation process according to practical needs.

In step 101, when the radio frequency signals of each frequency band are filtered, the method may include:

respectively acquiring radio frequency signals of each frequency band through switch setting;

respectively obtaining signals of each frequency band, and entering the signals into a mixer;

and (4) passing the mixed signal through an intermediate frequency feedback anti-aliasing filter.

Taking the F/a frequency band processing as an example, the feedback channels F/a respectively perform radio frequency analog filtering, only obtain F-band radio frequency signals or a-band radio frequency signals each time through switch setting, and then the respectively obtained F-band/a-band signals enter a shared mixer, an intermediate frequency amplifier and an intermediate frequency anti-aliasing intermediate frequency filter. Wherein the intermediate frequency amplifier may be considered as part of the gain adjuster.

When the radio frequency signals of each frequency band are respectively acquired, taking the processing of the F/a frequency band as an example, the respective signals of the F frequency band and the a frequency band can adopt a time-sharing acquisition mode, that is, firstly, the system acquires the signals of the F frequency band, and then acquires the signals of the a frequency band.

The processing of the rf filtering is explained by the difference of the spectrum distribution before and after the filtering, fig. 2 is a schematic diagram of an F + a signal output by the power amplifier, and fig. 2 is a spectrum diagram coupled back from the power amplifier, including signal spectrums of the F band and the a band.

Fig. 3 is a schematic diagram of signals in the F frequency band after the intermediate frequency feedback anti-aliasing filter, signals compressed by the respective intermediate frequency feedback anti-aliasing filters in the F/a frequency band are shown in fig. 3, and the passband in the diagram is a signal spectrum in the F frequency band, so that it can be seen that the signals in the a frequency band are effectively compressed, and the signals in the F frequency band and intermodulation information are completely retained. Fig. 4 is a schematic diagram of the signal in the frequency band a after the intermediate frequency feedback anti-aliasing filter, where the passband is the signal spectrum in the frequency band a, and it can be seen that the signal in the frequency band F is effectively compressed, and the signal in the frequency band a and the intermodulation information are completely retained.

In step 102, when acquiring the nonlinear intermodulation distortion of each frequency band, the method may include:

the ADC samples the filtered signal;

down-converting the sampled signals of each frequency band to zero frequency;

and filtering and extracting the down-converted signal.

In the implementation, taking the F/a frequency band processing as an example, signals in both the F frequency band and the a frequency band are first down-converted to fundamental frequency (zero frequency), and then image signals acquired by real number are filtered out by a low-pass filter, so as to retain useful signals. Because the signal obtained by ADC sampling only has F frequency band/A frequency band signal, the signal of the single frequency band is filtered and extracted after down-conversion to zero frequency, and the suppression of the image is finished (due to real number sampling, in the concrete implementation, because F/A is respectively sampled, the requirement of ADC sampling signal rate can be reduced.

In step 103, when synchronizing the nonlinear intermodulation distortion of each frequency band with the input signal of the frequency band, taking the F/a frequency band process as an example, the signal of the F/a single frequency band and the input original F/a frequency band signal are respectively synchronized in a correlated manner, so as to precisely adjust the synchronization of the F/a signals and simultaneously complete the gain adjustment of the respective frequency band signals.

In a specific implementation, let the correlation length of the transmitted and received data be DEFAULT _ XCORR _ L (the length of both the transmitted tx _ d and the received signal fb _ d) with an exemplary value of 4096.

[max_value，position]＝xcorr(fb_d，tx_d)

And obtaining the position where the maximum correlation value is located through correlation operation, and obtaining the delay _ m of the transmission signal in the loop:

delay_m＝positon-DEFAULT_XCORR_L

the transmit and feedback signal signals may then be synchronized by delaying the transmit signal by the magnitude of delay _ m. Fig. 5 is a diagram showing comparison of the amplitude of the intermediate frequency signal and the feedback signal after synchronous amplitude adjustment, and fig. 5 is a diagram showing the signal amplitude after synchronous calibration.

The amplitude calibration algorithm using root mean square is:

$y=y\·\frac{\mathrm{RMS}\left(z\right)}{\mathrm{RMS}\left(y\right)}$

wherein RMS (z) is as follows:

$\mathrm{RMS}\left(z\right)=\sqrt{\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}(\mathrm{zi}+\mathrm{zj})\·(\mathrm{zi}-\mathrm{zj})}{M}}$

the feedback signal and the transmit signal are calibrated using rms amplitude such that y and z are substantially equal in amplitude before filtering is applied, so that the estimated polynomial processing function F (·) does not change the amplitude of the signal. Figure 5 shows the amplitude comparison in the time domain after the feedback signal and the transmit signal have been amplitude calibrated for synchronization.

In step 104, updating the DPD coefficients with the synchronized signals may include two ways, which are:

1. in a first mode

Synchronously completing amplitude calibration after feeding back the signals of each frequency band respectively to form feedback signals of F + A;

performing information superposition on each frequency band signal to form an F + A signal, and then performing CFR (Crest factor reduction) processing;

carrying out DPD coefficient updating processing on the F + A signal fed back by the power amplifier and the F + A signal after CFR;

and updating the DPD channel by using the updated coefficient.

In the implementation, taking the processing of the F/a frequency band as an example, after the signals of the F frequency band and the a frequency band are all acquired, the signals of the respective frequency bands are subjected to synchronous calibration, and then mixed to the respective frequency bands, so as to form feedback digital signals of F + a, which are compared with the input F + a signals to complete the adjustment of the gain, and simultaneously, synchronous calibration is performed again, and then the calculation of the DPD coefficient update of the F + a is completed. Specifically, after the signals of the respective frequency bands of the F/a are respectively subjected to respective synchronous gain adjustment, the F/a signals are superposed to generate a signal of F + a, the signal of F + a is subjected to updating processing of a DPD coefficient with the input signal of F + a to generate a DPD coefficient of F + a, and the DPD coefficient of F + a is used for performing predistortion processing on the signal of F + a to complete digital predistortion processing of the broadband F + a.

2. Mode two

Updating DPD coefficients by using the signals synchronized by each frequency band and the input signals of the frequency band respectively;

respectively carrying out CFR processing on the input signals of each frequency band;

and carrying out DPD updating on the signals after each frequency band CFR and the coefficients after the frequency band updating.

In this manner, the F band and the a band perform respective DPD coefficient updating.

In implementation, different Nyquist zones are used for each frequency band feedback.

Taking the processing of the F/A frequency band as an example, the feedback of the F frequency band adopts a first Nyquist zone, and the feedback of the A frequency band adopts a second Nyquist zone, so that the transmission and the reception can share one LO, and the number of devices is saved.

Based on the same inventive concept, the embodiment of the present invention further provides a digital predistortion processing apparatus, and since the principle of solving the problem of these devices is similar to that of the digital predistortion processing method, the implementation of these devices may refer to the implementation of the method, and repeated details are not repeated.

Fig. 6 is a schematic structural diagram of a digital predistortion processing apparatus, as shown in the figure, the apparatus may include:

a filtering module 601, configured to filter the radio frequency signals of each frequency band respectively;

an acquisition module 602, configured to acquire nonlinear intermodulation distortion of each frequency band;

a synchronization module 603, configured to synchronize the nonlinear intermodulation distortion of each frequency band with an input signal of the frequency band;

an updating module 604, configured to update the DPD coefficient by using the synchronized signal.

In implementation, the implementation of the filtering module, the acquisition module, and the synchronization module may be the same, and the updating module may use the synchronized signal to update the DPD coefficient in two ways, so that schematic structural diagrams of two embodiments are provided, which are: fig. 7 is a schematic diagram of a digital predistortion processing apparatus, fig. 8 is a schematic diagram of a digital predistortion processing apparatus, and embodiments of the apparatus will be described below by taking two diagrams as examples.

In the implementation, the filtering module is used for filtering the radio frequency signals of each frequency band respectively; the concrete structure can be as follows:

the switch is used for controlling the passing of the radio frequency signals of each frequency band;

a mixer connected to the switch for mixing the passed signals;

and the intermediate frequency feedback anti-aliasing filter is connected with the mixer and used for filtering the mixed signal.

In the implementation, radio frequency signals of each frequency band are respectively obtained through switch setting; then, respectively obtaining signals of each frequency band, and entering the signals into a mixer; and finally, the mixed signals pass through an intermediate frequency feedback anti-aliasing filter.

In the implementation, the acquisition module is used for acquiring the nonlinear intermodulation distortion of each frequency band; the concrete structure can be as follows:

the ADC is used for sampling the filtered signal;

the down converter is used for carrying out down conversion on the sampled signals of each frequency band to zero frequency;

and the filter is used for filtering and extracting the down-converted signal.

In an implementation, the updating module 604 is configured to update the DPD coefficients by using the synchronized signals; the specific structure can be the following two types:

1. as shown in fig. 7:

the power amplifier feedback device is used for synchronously completing amplitude calibration after respectively feeding back the signals of each frequency band to form feedback signals of F + A;

the superposition and CFR processor is used for carrying out information superposition on the signals of all frequency bands to form F + A signals and then carrying out CFR processing;

the coefficient updating module is used for carrying out DPD coefficient updating processing on the F + A signal fed back by the power amplifier and the F + A signal after the CFR;

and the channel updating module is used for updating the DPD channel by using the updated coefficient.

2. As shown in fig. 8:

a coefficient updating module, which is used for updating DPD coefficients by using the signals synchronized by each frequency band and the input signals of the frequency band;

the CFR processor is used for respectively carrying out CFR on the input signals of all frequency bands;

and the frequency band updating module is used for carrying out DPD updating on the signals after each frequency band CFR and the coefficients after the frequency band updating.

Fig. 9 is a schematic diagram of the inter-modulation information of the F-band and the a-band, as shown in the figure, in the above implementation process, the inter-modulation information of the F-band and the a-band falls into the mutual band or the vicinity. Whatever the implementation scheme adopted, the respective intermodulation information of F + A needs to be effectively collected for the following reasons:

let the transfer function of the power amplifier to the F + a signal be: two signals F1 and F2 in the F band and two signals a1 and a2 in the a band have:

S_OUT＝f(S_f+S_a)＝f(S_f1+S_f2+S_a1+S_a2)＝C₀+C₁×(S_f1+S_f2+S_a1+F_a2)+C₂×(S_f1+S_f2+S_a1+S_a2)²+C₃×(S_f1+S_f2+S_a1+S_a2)³+.....

c0, C1, C2, C3 and C4. Due to the non-linearity of the device, it can be considered that C0, C1, C2, C3 and C4 are not equal to 0, and only the term with coefficient C3 is paired with B according to the above formula₁Or B₂The specific analysis is as follows:

${(S_f+S_a)}^3=S_f^3+3\&times;{\mathrm{<figure-callout\; id="2"\; label="S\; f"\; filenames="HDA0000034854060000012.png,HDA0000034854060000031.png"\; state="\{\{state\}\}">S</figure-callout>}}_f^{}\&times;S_a+3\&times;S_f\&times;S_a^2+S_a^3$

$={({\mathrm{<figure-callout\; id="1"\; label="S\; f"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">S</figure-callout>}}_f+S_{f2})}^3+3\&times;{({\mathrm{<figure-callout\; id="1"\; label="S\; f"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">S</figure-callout>}}_f+S_{f2})}^2\&times;(S_{a1}+S_{a2})+3\&times;({\mathrm{<figure-callout\; id="1"\; label="S\; f"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">S</figure-callout>}}_f+S_{f2})\&times;{(S_{a1}+S_{a2})}^2+{(S_{a1}+S_{a2})}^3$

$={({\mathrm{<figure-callout\; id="1"\; label="S\; f"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">S</figure-callout>}}_f+S_{f2})}^3+{(S_{a1}+S_{a2})}^3+3\&times;({\mathrm{<figure-callout\; id="1"\; label="S\; f"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">S</figure-callout>}}_f^1+{\mathrm{<figure-callout\; id="2"\; label="S\; f"\; filenames="HDA0000034854060000012.png,HDA0000034854060000031.png"\; state="\{\{state\}\}">S</figure-callout>}}_f^2+2*S_{f1}{S}_{f2})\&times;(S_{a1}+S_{a2})+$

$3\&times;(S_{a1}^2+S_{a2}^2+2*S_{a1}{S}_{a2})\&times;({\mathrm{<figure-callout\; id="1"\; label="S\; f"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">S</figure-callout>}}_f+S_{f2})$

the third order intermodulation distribution between F/A is:

${(S_f+S_a)}^3=S_f^3+3\&times;{\mathrm{<figure-callout\; id="2"\; label="S\; f"\; filenames="HDA0000034854060000012.png,HDA0000034854060000031.png"\; state="\{\{state\}\}">S</figure-callout>}}_f^{}\&times;S_a+3\&times;S_f\&times;S_a^2+S_a^3$

the intermodulation information falling in or near the F-band or a-band is as follows

The intermodulation information is distributed in or near the signal, the DPD effect is seriously influenced, and the intermodulation information can be effectively collected only when the power amplifier outputs the F + A signal. DPD updating is carried out on intermodulation collection nearby in the band, and the influence of PA (power amplifier) third-order intermodulation can be effectively eliminated by the calculated coefficient.

Fig. 10 is a diagram illustrating a DPD effect of splitting or combining, and fig. 11 is a diagram illustrating frequency spectrums before and after predistortion, as shown in the following:

s_{FA_IF}(n)＝(I_F+jQ_F)·exp(jω_f(n))+(I_A+jQ_A)·exp(jω_a(n))

＝(I_{F_IF}+j_{QF_IF})+(I_{A_IF}+jQ_{A_IF})＝(I_{F_IF}+I_{A_IF})+j(Q_{F_IF}+Q_{A_IF})

after the digital signal passes through a DAC and a low-pass filter, the digital signal is modulated into a radio frequency signal fcos omega through a quadrature modulator at a radio frequency_f+acosω_a：

s_{FA_RF}(n)＝(I_{F_IF}+I_{A_IF})·cosω_lo-(Q_{F_IF}+Q_{A_IF})·sinω_lo

＝(I_{F_IF}·cosω_lo-Q_{F_IF}·sinω_lo)+(I_{A_IF}·cosω_lo-Q_{A_IF}·sinω_lo)

＝fcosω_f+acosω_a

The intermodulation exhibited by this F + a signal after passing through the PA is as follows:

FA＝c1·(fcosω_f+acosω_a)+c3·(fcosω_f+acosω_a)³+c5·(fcosω_f+acosω_a)⁵+L

＝c1·S+c3·S³+c5·S⁵+L

$S^3={(f\cos {\mathrm{\&omega;}}_f+a\cos {\mathrm{\&omega;}}_a)}^3$

$=(f\cos {\mathrm{\&omega;}}_f+a\cos {\mathrm{\&omega;}}_a)\&CenterDot;({\mathrm{<figure-callout\; id="2"\; label="f"\; filenames="HDA0000034854060000012.png,HDA0000034854060000031.png"\; state="\{\{state\}\}">f</figure-callout>}}^2\&CenterDot;{\mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HDA0000034854060000012.png,HDA0000034854060000031.png"\; state="\{\{state\}\}">cos</figure-callout>}}^2{\mathrm{\&omega;}}_f+a^2\&CenterDot;{\mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HDA0000034854060000012.png,HDA0000034854060000031.png"\; state="\{\{state\}\}">cos</figure-callout>}}^2{\mathrm{\&omega;}}_a+2f\&CenterDot;a\&CenterDot;\cos {\mathrm{\&omega;}}_f\&CenterDot;\cos {\mathrm{\&omega;}}_a)$

$=\cos {\mathrm{\&omega;}}_f\&CenterDot;(\frac{3f^3}{4}+\frac{3{\mathrm{<figure-callout\; id="2"\; label="fa"\; filenames="HDA0000034854060000012.png,HDA0000034854060000031.png"\; state="\{\{state\}\}">fa</figure-callout>}}^2}{2})+\cos {\mathrm{\&omega;}}_a\&CenterDot;(\frac{3a^3}{4}+\frac{3f^{2}a}{2})+L$

$S^5=S^3(\frac{{\mathrm{<figure-callout\; id="2"\; label="f"\; filenames="HDA0000034854060000012.png,HDA0000034854060000031.png"\; state="\{\{state\}\}">f</figure-callout>}}^2}{2}(1+\mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HDA0000034854060000012.png,HDA0000034854060000031.png"\; state="\{\{state\}\}">cos</figure-callout>}2{\mathrm{\&omega;}}_f)+\frac{a^2}{2}(1+\mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HDA0000034854060000012.png,HDA0000034854060000031.png"\; state="\{\{state\}\}">cos</figure-callout>}2{\mathrm{\&omega;}}_a)+2\mathrm{fa}\cos {\mathrm{\&omega;}}_f\&CenterDot;\cos {\mathrm{\&omega;}}_a)$

$={\cos \mathrm{\&omega;}}_f\&CenterDot;(\frac{{5f}^5}{8}+\frac{{30f}^3{a}^2}{8}+\frac{{15\mathrm{fa}}^4}{8})+{\cos \mathrm{\&omega;}}_a\&CenterDot;(\frac{{5a}^2}{8}+\frac{{30f}^2{a}^3}{8}+\frac{{15f}^{4}a}{8})+K$

$\mathrm{FA}=\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">c</figure-callout>}1\&CenterDot;({f\cos \mathrm{\&omega;}}_f+a\cos {\mathrm{\&omega;}}_a)+c3\&CenterDot;{({f\cos \mathrm{\&omega;}}_f+{a\cos \mathrm{\&omega;}}_a)}^3+c5\&CenterDot;{({f\cos \mathrm{\&omega;}}_f+a{\cos \mathrm{\&omega;}}_a)}^5+L$

$=\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">c</figure-callout>}1\&CenterDot;S+c3\&CenterDot;S^5+c5\&CenterDot;S^5+L$

$=\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">c</figure-callout>}1\&CenterDot;(f\&CenterDot;{\cos \mathrm{\&omega;}}_f+a\&CenterDot;{\cos \mathrm{\&omega;}}_a)+c3\&CenterDot;[{\cos \mathrm{\&omega;}}_f\&CenterDot;(\frac{{3f}^3}{4}+\frac{{3\mathrm{<figure-callout\; id="2"\; label="fa"\; filenames="HDA0000034854060000012.png,HDA0000034854060000031.png"\; state="\{\{state\}\}">fa</figure-callout>}}^2}{2})+{\cos \mathrm{\&omega;}}_a\&CenterDot;(\frac{{3a}^2}{4}+\frac{{3f}^{2}a}{2})]+$

$c5\&CenterDot;[{\cos \mathrm{\&omega;}}_f\&CenterDot;(\frac{{5f}^5}{8}+\frac{{30f}^3{a}^2}{8}+\frac{{15\mathrm{fa}}^4}{8})+{\cos \mathrm{\&omega;}}_a\&CenterDot;(\frac{{5a}^5}{8}+\frac{{30f}^2{a}^3}{8}+\frac{{15f}^{4}a}{8})]+L$

$={\cos \mathrm{\&omega;}}_f\&CenterDot;[\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">c</figure-callout>}1\&CenterDot;f+c3\&CenterDot;(\frac{{3f}^3}{4}+\frac{{3\mathrm{<figure-callout\; id="2"\; label="fa"\; filenames="HDA0000034854060000012.png,HDA0000034854060000031.png"\; state="\{\{state\}\}">fa</figure-callout>}}^2}{2})+c5\&CenterDot;(\frac{{5f}^5}{8}+\frac{{30f}^3{a}^2}{8}+\frac{{15\mathrm{fa}}^4}{8})]+$

$\cos {\mathrm{\&omega;}}_a\&CenterDot;[\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">c</figure-callout>}1\&CenterDot;a+c3\&CenterDot;(\frac{{3a}^3}{4}+\frac{{3f}^{2}a}{2})+c5\&CenterDot;(\frac{{5a}^5}{8}+\frac{{30f}^2{a}^3}{8}+\frac{{15f}^{4}a}{8})]+L$

$=f\&CenterDot;{\cos \mathrm{\&omega;}}_f\&CenterDot;[\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">c</figure-callout>}1\&CenterDot;c3\&CenterDot;({\mathrm{<figure-callout\; id="2"\; label="f"\; filenames="HDA0000034854060000012.png,HDA0000034854060000031.png"\; state="\{\{state\}\}">f</figure-callout>}}^2+{2a}^2)+c5\&CenterDot;(f^4+{6f}^2{a}^2+{3a}^4)]+$

$a\&CenterDot;{\cos \mathrm{\&omega;}}_a\&CenterDot;[\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">c</figure-callout>}1+c3\&CenterDot;(a^2+{2f}^2)+c5\&CenterDot;(a^4+6f^2{a}^2+{3f}^4)]+L$

from the above formula, it can be seen that_fIn the frequency band, there are not only the fundamental frequency, third order and fifth order frequency band information of the F band, but also the signal of the a band, which is consistent with the conclusion that the above F/a transmits 2 tones. Also omega_aIn the frequency band, there are not only the fundamental frequency, third order and fifth order frequency band information of the A frequency band, but also the signal of the F frequency band, and the signals of F and A also fall into the corresponding frequency band.

Intermodulation information of all frequency points can not be calculated completely, only intermodulation components which are mutually inquired by the F/A and fall in the frequency bands of the F/A are considered, and the size of the intermodulation components is related to the signal bandwidth. Although frequency components of the F + A signal are numerous after power amplification, the F + A signal can be represented by a DPD coefficient and a model, so that the magnitude of third-order, fifth-order and other high-order components of F/A mutual influence can be reflected in the coefficient and the model, and the following formula shows a mathematical model of F + A intermodulation in a DPD channel;

z_FA(n)＝y_f(n)·[c1·+c3·(|y_f(n)|²+2|y_a(n)|²)+c5·(|y_f(n)|⁴+6|y_f(n)|²|y_a(n)|²+|y_a(n)|⁴)]

+y_a(n)·[c1·+c3·(|y_f(n)|²+2|y_a(n)|²)+c5·(|y_f(n)|⁴+6|y_f(n)|²|y_a(n)|²+|y_a(n)|⁴)]

if the memory of the system is considered, the derivation of the hardware LUT (Look-Up-Table) model is as follows:

$z_{\mathrm{FA}}\left(n\right)=$

$\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}\begin{array}{c}{y}_f(n-k)\&CenterDot;[\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">c</figure-callout>}1\&CenterDot;+c3\&CenterDot;({\left|y_f(n-k)\right|}^2+2{\left|y_a(n-k)\right|}^2)+c5\&CenterDot;({\left|y_f(n-k)\right|}^4+6{\left|y_f(n-k)\right|}^2{\left|y_a(n-k)\right|}^2+{\left|y_a(n-k)\right|}^4)]\\ y_a(n-k)\&CenterDot;[\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="HDA0000034854060000051.png"\; state="\{\{state\}\}">c</figure-callout>}1\&CenterDot;+c3\&CenterDot;({\left|y_f(n-k)\right|}^2+2{\left|y_a(n-k)\right|}^2)+c5\&CenterDot;({\left|y_f(n-k)\right|}^4+6{\left|y_f(n-k)\right|}^2{\left|y_a(n-k)\right|}^2+{\left|y_a(n-k)\right|}^4)]\end{array}$

$=\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}(y_f(n-k)\&CenterDot;{\mathrm{LUT}}_k|y_f(n-k)|+y_a(n-k)\&CenterDot;{\mathrm{LUT}}_k|y_a(n-k)\left|\right)$

$=\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}(y_f(n-k)\&CenterDot;{\mathrm{LUT}}_k|y_f(n-k)\left|\right)+\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}{y}_a(n-k)\&CenterDot;{\mathrm{LUT}}_k\left|y_a(n-k)\right|$

through the derivation of the series of formulas, the model of F + A only needs to add one LUT table and multiple K multipliers compared with the model of a single frequency band. Other hardware resources are the same as for the single band.

Of course, the effects of seventh order intermodulation and the effects of even order intermodulation can also be considered and a similar derivation can be done.

1. Setting input parameter M as { K, P }, signal length as N

2. Introducing reference signal z (n) and feedback signal y (n)

U＝[u₁，L u_kpL，u_KP]_N*KP，u_kp＝[u_kp(0)，L，u_kp(N-1)]^T，

u_k1(n)＝y(n-k)，

u_k3(n)＝y(n-k)·(|y(n-k)|²+2|y(n-k)|²)，

u_k5(n)＝y(n-k)·(|y(n-k)|⁴+6|y(n-k)|²|y(n-k)|²+|y(n-k)|⁴)

M

z＝Uc，KP×KP

z＝[z(0)，L，z(N-1)]^T，

$\hat{c}={\left(U^{H}U\right)}^{-1}{U}^H\&CenterDot;z$

After the predistortion coefficient is obtained through the formula, a lookup table can be designed to update the DPD coefficient of the input F + A signal.

As can be seen from the foregoing embodiments, in the technical solutions provided in the embodiments of the present invention, the respective feedback channels are respectively a radio frequency feedback filter and a frequency band selection switch. And after respective synchronous calibration, combining the signals into a feedback F + A signal again, and comparing the signal with a forward F + A signal to obtain a DPD coefficient.

Specifically, the feedback channels F/a respectively perform radio frequency analog filtering, only an F-band radio frequency signal or an a-band radio frequency signal is obtained each time through switch setting, and the respectively obtained F-band/a-band signals then enter a shared mixer, an intermediate frequency amplifier and an intermediate frequency anti-aliasing intermediate frequency filter.

Specifically, the signals obtained by ADC sampling only include F band/a band signals (because F/a is sampled respectively, the requirement for ADC sampling signal rate can be reduced), the signals of the individual band are down-converted to zero frequency and then filtered and extracted, image rejection (caused by real number sampling) is completed, the signals of the F/a individual band and the input original F/a band signals are respectively correlated and synchronized, synchronization of the F/a signals is accurately adjusted, and gain adjustment of the respective band signals is completed at the same time.

Specifically, after the signals of the respective frequency bands of the F/a are respectively subjected to respective synchronous gain adjustment, the F/a signals are superposed to generate a signal of F + a, the signal of F + a is subjected to updating processing of a DPD coefficient with the input signal of F + a to generate a DPD coefficient of F + a, and the DPD coefficient of F + a is used for performing predistortion processing on the signal of F + a to complete digital predistortion processing of the broadband F + a.

Specifically, the first Nyquist zone is adopted for F frequency band feedback, and the second Nyquist zone is adopted for A frequency band feedback, so that the transmission and the reception can share one LO, and the number of devices is saved.

Through the radio frequency respective filtering, the DPD can obtain high-efficiency DPD processing effect under the low-complexity hardware condition through the digital common DPD processing.

It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.

Claims (7)

1. A Digital Predistortion (DPD) processing method is characterized by comprising the following steps:

respectively filtering the radio frequency signals of each frequency band;

collecting nonlinear intermodulation distortion of each frequency band;

synchronizing the nonlinear intermodulation distortion of each frequency band with the input signal of the frequency band;

updating the DPD coefficient by using the synchronized signal;

wherein, updating the DPD coefficients using the synchronized signals includes:

synchronously completing amplitude calibration after respectively feeding back signals of all frequency bands to form feedback signals of an F + A frequency band, wherein the range of the F frequency band is 1880MHz-1920MHz, and the range of the A frequency band is 2010MHz-2025 MHz;

carrying out information superposition on each frequency band signal to form an F + A signal, and then carrying out peak value factor removal (CFR) processing;

carrying out DPD coefficient updating processing on the F + A signal fed back by the power amplifier and the F + A signal after CFR;

updating the DPD channel by using the updated coefficient;

or,

updating DPD coefficients by using the signals synchronized by each frequency band and the input signals of the frequency band respectively;

respectively carrying out peak factor removal (CFR) on the input signals of each frequency band;

and carrying out DPD updating on the signals after each frequency band CFR and the coefficients after the frequency band updating.

2. The method of claim 1, wherein the filtering the rf signals of each frequency band separately comprises:

respectively acquiring radio frequency signals of each frequency band through switch setting;

respectively obtaining signals of each frequency band, and entering the signals into a mixer;

and (4) passing the mixed signal through an intermediate frequency feedback anti-aliasing filter.

3. The method of claim 1, wherein collecting the non-linear intermodulation distortion for each frequency band comprises:

the ADC samples the filtered signal;

down-converting the sampled signals of each frequency band to zero frequency;

and filtering and extracting the down-converted signal.

4. The method of claim 1, wherein each frequency band feedback uses a different Nyquist zone.

5. A digital predistortion processing apparatus, comprising:

the filtering module is used for respectively filtering the radio frequency signals of each frequency band;

the acquisition module is used for acquiring the nonlinear intermodulation distortion of each frequency band;

the synchronization module is used for synchronizing the nonlinear intermodulation distortion of each frequency band with the input signal of the frequency band;

the updating module is used for updating the DPD coefficient by using the synchronized signal;

wherein, the update module includes:

the power amplifier feedback device is used for synchronously completing amplitude calibration after respectively feeding back signals of each frequency band to form feedback signals of an F + A frequency band, wherein the range of the F frequency band is 1880MHz-1920MHz, and the range of the A frequency band is 2010MHz-2025 MHz;

the superposition and CFR processor is used for carrying out information superposition on the signals of all frequency bands to form an F + A signal and then carrying out peak factor removal CFR processing;

the coefficient updating module is used for carrying out DPD coefficient updating processing on the F + A signal fed back by the power amplifier and the F + A signal after the CFR;

the channel updating module is used for updating the DPD channel by using the updated coefficient;

or,

a coefficient updating module, which is used for updating DPD coefficients by using the signals synchronized by each frequency band and the input signals of the frequency band;

a peak factor removal CFR processor for performing CFR on the input signals of each frequency band;

and the frequency band updating module is used for carrying out DPD updating on the signals after each frequency band CFR and the coefficients after the frequency band updating.

6. The apparatus of claim 5, wherein the filtering module comprises: the switch is used for controlling the passing of the radio frequency signals of each frequency band;

a mixer connected to the switch for mixing the passed signals;

and the intermediate frequency feedback anti-aliasing filter is connected with the mixer and used for filtering the mixed signal.

7. The apparatus of claim 5, wherein the acquisition module comprises:

the ADC is used for sampling the filtered signal;

the down converter is used for carrying out down conversion on the sampled signals of each frequency band to zero frequency;

and the filter is used for filtering and extracting the down-converted signal.

Images