CN103957178A

CN103957178A - Multi-channel digital pre-distortion processing method and system

Info

Publication number: CN103957178A
Application number: CN201410120758.9A
Authority: CN
Inventors: 李建强; 徐坤; 裴寅清; 陈皓; 戴一堂; 尹飞飞
Original assignee: Beijing University of Posts and Telecommunications
Current assignee: Beijing University of Posts and Telecommunications
Priority date: 2014-03-27
Filing date: 2014-03-27
Publication date: 2014-07-30

Abstract

The present invention proposes a multi-channel digital pre-distortion processing method, including: predicting that the frequency division multiplexing signals in each channel and the frequency division multiplexing signals in other channels are simultaneously loaded into the optical wireless link for transmission after frequency multiplexing Intermodulation distortion generated; Compensating the intermodulation distortion between the frequency division multiplexing signals in different channels. The present invention also provides a multi-channel digital pre-distortion processing system, including: a multi-channel digital pre-distortion module, used to establish a digital pre-processing model and perform digital pre-distortion processing on input signals according to the digital pre-processing model to compensate ROF Non-linear distortion in the link; a digital pre-distortion training module, used to solve the parameters of the digital pre-processing model. By adopting the multi-channel digital pre-distortion processing method and system disclosed in the present invention, the memory effect and nonlinear intermodulation distortion introduced by the ROF link to the frequency division multiplexing signal can be compensated, so that the whole system can realize linear transmission.

Description

Multi-channel digital predistortion processing method and system

技术领域technical field

本发明涉及通信信号处理技术领域，特别设计一种多信道数字预失真处理方法及系统。The invention relates to the technical field of communication signal processing, and particularly designs a multi-channel digital pre-distortion processing method and system.

背景技术Background technique

光载无线技术融合了无线通信灵活接入以及光纤通信高带宽、低成本、低损耗的优点，被广泛认为是未来无线接入网络最有吸引力的解决方案。光载无线链路可将任意无线信号调制到光载波上，在光纤中传输一段距离后再通过光电转换恢复成为原来的无线信号并直接发射到无线信道中。光纤链路损耗极低，因此所传输无线信号的覆盖范围即可得到极大的延伸。同时，相较传统无线通信系统中所使用的宏基站，基于光载无线技术的无线通信系统的远端天线单元从成本、能耗到体积都有极大的减小。因此光载无线技术的日益成熟为大规模微蜂窝无线通信系统的铺设提供了可能。Wireless over fiber technology combines the advantages of flexible wireless access and optical fiber communication with high bandwidth, low cost, and low loss, and is widely considered to be the most attractive solution for future wireless access networks. The wireless link over light can modulate any wireless signal onto the optical carrier, transmit it for a certain distance in the optical fiber, and then restore it to the original wireless signal through photoelectric conversion and directly transmit it to the wireless channel. Fiber optic links have extremely low loss, so the coverage of the transmitted wireless signal can be greatly extended. At the same time, compared with the macro base station used in the traditional wireless communication system, the remote antenna unit of the wireless communication system based on the radio-over-fiber technology is greatly reduced in terms of cost, energy consumption and volume. Therefore, the increasing maturity of the radio-over-optical technology provides the possibility for the laying of large-scale microcellular wireless communication systems.

然而，光载无线链路是一个非线性传输系统。当传输非恒包络信号时（如高阶QAM调制、OFDM调制信号、频分复用信号等），会对信号产生的非线性畸变，引起系统性能恶化。数字预失真是此问题的最佳解决方案之一。数字预失真采用FPGA,DSP处理芯片等数字信号处理器在数字域对待传信号引入一个特性与系统非线性失真特性恰好相反的失真，以实现整个系统的线性化输出。数字预失真因其成本低，效果好，灵活度高，调节精度高，兼容性强等的特点，已广泛的被商用通信系统所采用。如图1所示方案（IEEETRANSACTIONS ON CIRCUITS AND SYSTEMS—I:REGULARPAPERS,VOL.59,NO.3,MARCH2012pp664-672），即为一个典型的数字预失真方案的范例。然而，传统的数字预失真的方法可处理信号的带宽受限于数字信号处理器以及数模转换器的处理带宽（～100MHz）。因此其只适用于窄带单信道信号传输的应用场景。However, radio over optical link is a nonlinear transmission system. When transmitting non-constant envelope signals (such as high-order QAM modulation, OFDM modulation signals, frequency division multiplexing signals, etc.), nonlinear distortion will be generated on the signal, causing system performance to deteriorate. Digital predistortion is one of the best solutions to this problem. Digital pre-distortion uses digital signal processors such as FPGAs and DSP processing chips to introduce a distortion that is exactly opposite to the nonlinear distortion characteristics of the system in the digital domain to the signal to be transmitted, so as to achieve the linear output of the entire system. Due to its low cost, good effect, high flexibility, high adjustment accuracy, and strong compatibility, digital predistortion has been widely used in commercial communication systems. The scheme shown in Figure 1 (IEEETRANSACTIONS ON CIRCUITS AND SYSTEMS—I: REGULARPAPERS, VOL.59, NO.3, MARCH2012pp664-672) is an example of a typical digital predistortion scheme. However, the signal processing bandwidth of the traditional digital predistortion method is limited by the processing bandwidth of the digital signal processor and the digital-to-analog converter (~100MHz). Therefore, it is only applicable to the application scenario of narrowband single-channel signal transmission.

为了扩大数字预失真技术的应用范围，（专利申请号201310121508.2：一种多载波复用光载无线链路系统及其数字预失真方法）提出了一种多频点副载波复用光载无线链路的数字预失真方法。该方法是对副载波复用光载无线链路所传输的多路频分复用信号分别在基带进行数字预失真，然后再进行变频和频分复用。这样数字预失真的带宽仅取决于各路信号的带宽，而非整个频分复用信号的带宽。然而该“一种多载波复用光载无线链路系统及其数字预失真方法”所提出的数字预失真模型是无记忆效应的非线性模型。这种模型仅在光载无线链路的带宽远大于传输信号带宽的前提下才可较好补偿非线性失真。通常，基于外调制的光载无线链路的带宽较大（可达到40GHz以上），在传输现有分布于400MHz～3GHz的商用无线信号时，记忆效应可以忽略，因此可以采用以上方案。而受技术、工艺等限制，基于直调的光载无线链路的带宽普遍被限制在10GHz以内。在此情况下，系统传输400MHz～3GHz的多频点复用信号产生的记忆效应不可忽略。In order to expand the application range of digital pre-distortion technology, (Patent Application No. 201310121508.2: A multi-carrier multiplexed optical-carrier wireless link system and its digital pre-distortion method) proposed a multi-frequency subcarrier multiplexed optical-carrier wireless link way of digital predistortion. The method is to carry out digital pre-distortion on the baseband of multiple frequency division multiplexing signals transmitted by multiplexed light-carrying wireless links of sub-carriers, and then perform frequency conversion and frequency division multiplexing. In this way, the bandwidth of the digital predistortion only depends on the bandwidth of each signal, rather than the bandwidth of the entire frequency division multiplexed signal. However, the digital pre-distortion model proposed in "A Multi-carrier Multiplexing Optical Carrier Radio Link System and Its Digital Pre-Distortion Method" is a non-linear model without memory effect. This model can better compensate the nonlinear distortion only under the premise that the bandwidth of the radio-over-fiber link is much larger than the bandwidth of the transmission signal. Usually, the bandwidth of optical radio links based on external modulation is large (up to 40GHz or more), and the memory effect can be ignored when transmitting existing commercial wireless signals distributed in the range of 400MHz to 3GHz, so the above scheme can be adopted. However, due to limitations in technology and process, the bandwidth of direct modulation-based optical wireless links is generally limited to within 10 GHz. In this case, the memory effect produced by the system transmitting 400MHz-3GHz multi-frequency point multiplexing signals cannot be ignored.

发明内容Contents of the invention

本发明所要解决的技术问题是数字预失真处理中由于记忆效应未被考虑导致的不足。The technical problem to be solved by the present invention is that the memory effect is not taken into account in the digital pre-distortion process.

为此目的，本发明提出了一种多信道数字预失真处理方法，包括：For this purpose, the present invention proposes a kind of multi-channel digital predistortion processing method, comprising:

预测各信道中的频分复用信号与其他信道中的频分复用信号在频率复用后同时加载到光载无线链路传输产生的交调失真；Predict the intermodulation distortion caused by the frequency division multiplexing signal in each channel and the frequency division multiplexing signal in other channels being simultaneously loaded to the optical wireless link for transmission after frequency multiplexing;

对不同信道中的各所述频分复用信号间的交调失真进行补偿。Compensating for intermodulation distortion between said frequency division multiplexed signals in different channels.

优选的，不同信道中的各所述频分复用信号间的交调失真通过连续信号传输模型获得。Preferably, the intermodulation distortion between the frequency division multiplexed signals in different channels is obtained through a continuous signal transmission model.

优选的，所述连续信号传输模型为：Preferably, the continuous signal transmission model is:

v_out=∑_pK_p(v_in/g)^p v _out =∑ _p K _p (v _in /g) ^p

其中，v_in为待数字预失真处理的等效连续信号的瞬时输入，g为ROF链路的线性增益，K_p为记忆多项式归一化的p阶非线性系数，v_out为经数字预失真处理后的等效连续信号的瞬时输出。Among them, v _in is the instantaneous input of the equivalent continuous signal to be processed by digital predistortion, g is the linear gain of the ROF link, K _p is the p-order nonlinear coefficient normalized by the memory polynomial, v _out is the digital predistortion The instantaneous output of the processed equivalent continuous signal.

优选的，对各信道中频分复用信号进行数字预失真处理采用的数学模型为：Preferably, the mathematical model used for digital pre-distortion processing of frequency division multiplexing signals in each channel is:

${Z Z}_{i i} ((n no)) = = \underset{q q}{Σ Σ} \underset{p p}{Σ Σ} {K K}_{pq pq} \frac{{X x}_{i i} ((n no - - q q))}{| | {X x}_{i i} ((n no - - q q)) | |} \frac{11}{22^{p p - - 11}} \underset{{f f}_{i i} = = {Σ Σ}_{j j}^{p p} {f f}_{ij ij}}{Σ Σ} {Π Π}_{j j = = 11}^{p p} | | {X x}_{{i i}_{j j}} ((n no - - q q)) | |$

其中，X_i(n)为第i个子载波上信号的复幅度，f_i为第i个子载波上信号的载波频率，K_pq为记忆多项式第p阶非线性、第q阶记忆效应项的归一化系数，Z_i(n)为经过数字预失真处理后的输出。Among them, X _i (n) is the complex amplitude of the signal on the i-th subcarrier, f _i is the carrier frequency of the signal on the i-th sub-carrier, K _pq is the normalization of the p-th order nonlinearity and q-th order memory effect term of the memory polynomial normalization coefficient, Z _i (n) is the output after digital pre-distortion processing.

优选的，各信道中频分复用信号进行数字预失真处理采用的数学模型的系数K_pq用最小均方根法得出。Preferably, the coefficient K _pq of the mathematical model used for the digital pre-distortion processing of the frequency division multiplexing signal in each channel is obtained by the least mean square method.

优选的，在各信道中所述频分复用信号上变频前进行所述多信道数字预失真处理方法。Preferably, the multi-channel digital pre-distortion processing method is performed before the frequency-division multiplexing signal in each channel is up-converted.

本发明还公开了一种多信道数字预失真处理系统，包括：The invention also discloses a multi-channel digital pre-distortion processing system, comprising:

多信道数字预失真模块，用以建立数字预处理模型并根据所述数字预处理模型对输入信号进行数字预失真处理以补偿ROF链路中的非线性失真；A multi-channel digital pre-distortion module, used to establish a digital pre-processing model and perform digital pre-distortion processing on the input signal according to the digital pre-processing model to compensate for nonlinear distortion in the ROF link;

数字预失真训练模块，用以对所述数字预处理模型的参数进行求解。The digital predistortion training module is used to solve the parameters of the digital preprocessing model.

优选的，所述数字预处理模型为：Preferably, the digital preprocessing model is:

优选的，所述数字预失真训练模块使用的数学模型为所述数字预处理模型，所述数字预失真训练模块通过对已知的训练序列进行对比获取所述数字预处理模型的参数。Preferably, the mathematical model used by the digital pre-distortion training module is the digital pre-processing model, and the digital pre-distortion training module obtains the parameters of the digital pre-processing model by comparing known training sequences.

优选的，所述数字预失真训练模块设置于ROF链路反馈回路上，所述数字预失真训练模块以ROF链路的归一化输出为输入且所述数字预失真训练模块的输出与所述多信道数字预失真模块的输出之差收敛。Preferably, the digital pre-distortion training module is set on the ROF link feedback loop, the digital pre-distortion training module takes the normalized output of the ROF link as input and the output of the digital pre-distortion training module and the The difference between the outputs of the multi-channel digital pre-distortion module converges.

通过采用本发明所公开的多信道数字预失真处理方法及系统可补偿ROF链路对频分复用信号引入的记忆效应和非线性交调失真，使整个系统实现线性传输，并且可适用于任何光载无线链路，是副载波复用光载无线链路的通用数字预失真方法。By adopting the multi-channel digital pre-distortion processing method and system disclosed in the present invention, the memory effect and nonlinear intermodulation distortion introduced by the ROF link to the frequency division multiplexing signal can be compensated, so that the entire system can realize linear transmission, and can be applied to any The optical wireless link is a general digital pre-distortion method for multiplexing the optical wireless link with subcarriers.

附图说明Description of drawings

通过参考附图会更加清楚的理解本发明的特征和优点，附图是示意性的而不应理解为对本发明进行任何限制，在附图中：The features and advantages of the present invention will be more clearly understood by referring to the accompanying drawings, which are schematic and should not be construed as limiting the invention in any way. In the accompanying drawings:

图1示出了现有技术中的数字预失真框图。Fig. 1 shows a block diagram of digital predistortion in the prior art.

图2示出了基于基带数字预失真的副载波复用光载无线链路框图。Fig. 2 shows a block diagram of a wireless link over optical with subcarrier multiplexing based on baseband digital predistortion.

图3示出了多信道数字预失真实验系统框图。Fig. 3 shows a block diagram of a multi-channel digital pre-distortion experiment system.

图4示出了频分复用信号接收频谱图。FIG. 4 shows a frequency division multiplexing signal reception spectrum diagram.

图5示出了两路OFDM-64QAM信号经过/未经过多信道预失真处理的EVM对比图。FIG. 5 shows a comparison diagram of EVM of two channels of OFDM-64QAM signals with or without multi-channel predistortion processing.

具体实施方式Detailed ways

下面将结合附图对本发明的实施例进行详细描述。Embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

本发明提供了一种针对传输频分复用无线信号的光载无线链路（副载波复用光载无线链路）的多信道数字预失真处理方法，且该数字处理在每路信号上变频之前进行。在该数字预失真处理中又进行：预测各信道中的频分复用信号与其他信道中的频分复用信号在频率复用后同时加载到光载无线链路传输产生的交调失真；对不同信道中的各所述频分复用信号间的交调失真进行补偿。The present invention provides a multi-channel digital predistortion processing method for an optical-carrying wireless link (sub-carrier multiplexing optical-carrying wireless link) for transmitting frequency-division multiplexing wireless signals, and the digital processing is frequency-converted on each signal carried out before. In the digital pre-distortion processing, it is further carried out: predicting the intermodulation distortion generated by the frequency-division multiplexing signal in each channel and the frequency-division multiplexing signal in other channels being simultaneously loaded into the optical-carrier wireless link for transmission after frequency multiplexing; Compensating for intermodulation distortion between said frequency division multiplexed signals in different channels.

此外，本发明还提供了一种多信道数字预失真处理系统，该系统包括：In addition, the present invention also provides a multi-channel digital pre-distortion processing system, which includes:

多信道数字预失真模块，用以建立数字预处理模型并根据这一数字预处理模型对输入信号进行数字预失真处理以补偿ROF链路中的非线性失真；A multi-channel digital pre-distortion module is used to establish a digital pre-processing model and perform digital pre-distortion processing on the input signal according to the digital pre-processing model to compensate for nonlinear distortion in the ROF link;

数字预失真训练模块，用以对数字预处理模型的参数进行求解。The digital predistortion training module is used to solve the parameters of the digital preprocessing model.

如图2所示，为基于基带数字预失真的副载波复用光载无线链路框图。其中，X_i(n)(i=1～I)表示每个信道所承载的信号上变频前的基带信号，Y_i(n)为每路信号经过光载无线链路的输出并下变频后接收到的信号。在对原始基带信号X_i(n)(i=1～I)上变频及频率复用之前，将其通过多信道数字预失真模块进行数字预失真处理使每路信号分别产生一个非线性失真，输出Z_i(n)。之后，以Z_i(n)代替X_i(n)作为副载波复用光载无线链路的输入，使其非线性输出Y_i(n)与原始信号X_i(n)成为线性关系，即Y_i=g_iX_i,其中g_i表示每路信号的增益。As shown in FIG. 2 , it is a block diagram of a radio-over-optical link with subcarrier multiplexing based on baseband digital predistortion. Among them, X _i (n) (i=1～I) represents the baseband signal before up-conversion of the signal carried by each channel, and Y _i (n) is the output of each signal through the optical wireless link and after down-conversion received signal. Before the original baseband signal X _i (n) (i=1～I) is up-converted and frequency multiplexed, it is subjected to digital pre-distortion processing through a multi-channel digital pre-distortion module so that each signal generates a nonlinear distortion respectively, Output Z _i (n). Afterwards, replace Xi (n) with Z _i ₍ n) as the input of subcarrier multiplexed radio-over-optical link, so that its nonlinear output Y _i (n) has a linear relationship with the original signal _Xi (n), namely Y _i =g _i X _i , where g _i represents the gain of each signal.

作为一种优选的实施方式，多信道数字预失真模块的系数是通过间接学习法获得的。在数字预失真模块开始对待传信号进行数字预失真处理以前，先发送一个已知的训练序列，并由置于ROF链路反馈回路上的数字预失真训练模块对其输入输出进行对比,来获得数字预失真模块的系数。数字预失真训练模块以ROF链路的归一化输出Yi/gi为输入，通过对模块系数的训练获得输出使其与多信道数字预失真模块输出之差收敛，即使最小。每次训练之后，多信道数字预失真模块将完全复制多信道数字预失真训练模块的系数，作为对待传信号进行预失真的完整数学模型。As a preferred implementation manner, the coefficients of the multi-channel digital pre-distortion module are obtained through an indirect learning method. Before the digital pre-distortion module starts digital pre-distortion processing on the signal to be transmitted, a known training sequence is sent first, and the input and output of the digital pre-distortion training module placed on the feedback loop of the ROF link are compared to obtain Coefficients of the digital predistortion block. The digital predistortion training module takes the normalized output Yi/gi of the ROF link as input, and obtains the output by training the module coefficients Make it converge with the difference between the output of the multi-channel digital pre-distortion module, even if minimum. After each training, the multi-channel digital pre-distortion module will completely copy the coefficients of the multi-channel digital pre-distortion training module as a complete mathematical model for pre-distorting the signal to be transmitted.

其中，多信道数字预失真模块的数学模型是基于通用的记忆多项式建立的，Among them, the mathematical model of the multi-channel digital pre-distortion module is established based on a general memory polynomial,

$z z ((n no)) = = \underset{q q}{Σ Σ} \underset{p p}{Σ Σ} {K K}_{pq pq} x x ((n no - - q q)) {| | x x ((n no - - q q)) | |}^{p p - - 11} - - - - - - ((11))$

其中x(n)与z(n)是数字预失真模块的输入与输出，K_pq是记忆多项式第p阶非线性、第q阶记忆效应项的归一化系数。Among them, x(n) and z(n) are the input and output of the digital pre-distortion module, and K _pq is the normalization coefficient of the p-th order nonlinearity and the q-th order memory effect term of the memory polynomial.

数字预失真处理模块的等效连续信号模型可表示为ROF链路连续信号非线性模型的反函数，写为The equivalent continuous signal model of the digital predistortion processing module can be expressed as the inverse function of the ROF link continuous signal nonlinear model, written as

v_out=∑_pK_p(v_in/g)^p (2)其中v_in and v_out是DPD模块的等效连续信号的瞬时输入与输出，g是ROF链路的线性增益，K_p是归一化的p阶非线性系数。v _out =∑ _p K _p (v _in /g) ^p (2) where v _in and v _out are the instantaneous input and output of the equivalent continuous signal of the DPD module, g is the linear gain of the ROF link, and K _p is the normalized Normalized p-order nonlinear coefficients.

公式（1）和（2）的不同点在于，公式（1）所描述的是离散的基带采样信号模型，而（2）所描述的是连续信号模型。因此（2）可用来描述在副载波复用光载无线链路上传输的带有多载波的频分复用信号的传输模型。一个带有I个载波的频分复用信号在第(n-1)^th～n^th个采样区间的表达式为The difference between formulas (1) and (2) is that formula (1) describes a discrete baseband sampling signal model, while (2) describes a continuous signal model. Therefore (2) can be used to describe the transmission model of the frequency division multiplexing signal with multiple carriers transmitted on the subcarrier multiplexing optical-carrying wireless link. The expression of a frequency division multiplexing signal with I carrier in the (n-1) ^th to n ^th sampling interval is

其中X_i(n)，f_i和分别表示第i个子载波上信号的复幅度，载波频率和初始相位。将（1）和（3）带入（2），推导出v_out(n)为where X _i (n), f _i and represent the complex amplitude, carrier frequency and initial phase of the signal on the i-th subcarrier, respectively. Substituting (1) and (3) into (2), deduce v _out (n) as

其中(1)中的K_pq代替了(2)中的K_p以将记忆效应考虑其中.注意到，式（4）中满足p>1及不相等的项即为对交调失真的补偿项。Among them, K _pq in (1) replaces K _p in (2) to take the memory effect into account. Note that p>1 and The unequal term is the compensation term for intermodulation distortion.

提取式（4）中每个载波（即每个cos）的复幅度，我们即可得到每个信道数字预失真的数学模型，即Extracting the complex amplitude of each carrier (that is, each cos) in formula (4), we can get the mathematical model of digital predistortion for each channel, namely

${Z Z}_{i i} ((n no)) = = \underset{q q}{Σ Σ} \underset{p p}{Σ Σ} {K K}_{pq pq} \frac{{X x}_{i i} ((n no - - q q))}{| | {X x}_{i i} ((n no - - q q)) | |} \frac{11}{22^{p p - - 11}} \underset{{f f}_{i i} = = {Σ Σ}_{j j}^{p p} {f f}_{ij ij}}{Σ Σ} {Π Π}_{j j = = 11}^{p p} | | {X x}_{{i i}_{j j}} ((n no - - q q)) | | - - - - - - ((55))$

多信道数字预失真模块及数字预失真训练模块均采用公式（5）作为数学模型。而数字预失真训练模块可采用简单的最小均方根法（LMS法）对模型系数K_pq进行训练。Both the multi-channel digital pre-distortion module and the digital pre-distortion training module use formula (5) as a mathematical model. The digital predistortion training module can use a simple least mean square method (LMS method) to train the model coefficient K _pq .

首先，我们将ROF链路的输出Y _i进行归一化，得到序列First, we normalize the output Y _i of the ROF link to obtain the sequence

${u u}_{i i}^{pq pq ((m m))} ((n no)) = = \frac{{Y Y}_{i i} ((n no - - q q))}{| | {Y Y}_{i i} ((n no - - q q)) | |} \underset{{f f}_{i i} = = {Σ Σ}_{j j}^{p p} &PlusMinus; &PlusMinus; {f f}_{{i i}_{j j}}}{{Π Π}_{j j = = 11}^{p p}} | | \frac{11}{{g g}_{i i}} {Y Y}_{{i i}_{j j}} ((n no - - q q)) | | - - - - - - ((66))$

其中上标（m）表示满足的第m种组合。由组成的集合U_i，where the superscript (m) indicates that the The mth combination of . Depend on The set U _i composed of,

$U_{i} = [u_{i}^{10 (1)}, \cdot \cdot \cdot, u_{i}^{1 Q (1)}, \frac{1}{2^{2}} u_{i}^{30 (1)}, \cdot \cdot \cdot, \frac{1}{2^{2}} u_{i}^{30 (M)}, \frac{1}{2^{2}} u_{i}^{3 Q (1)}, \cdot \cdot \cdot, \frac{1}{2^{2}} u_{i}^{3 Q (M)}, \cdot \cdot \cdot \cdot \cdot \cdot, \frac{1}{2^{P - 1}} u_{i}^{PQ (L)}], u_{i}^{pq (m)} = {[u_{i}^{pq (m)} (0), \cdot \cdot \cdot, u_{i}^{pq (m)} (N - 1)]}^{T},$ 当收敛时应有 $u_{i} = [u_{i}^{10 (1)}, &Center Dot; &Center Dot; &Center Dot;, u_{i}^{1 Q (1)}, \frac{1}{2^{2}} u_{i}^{30 (1)}, &Center Dot; &Center Dot; \cdot, \frac{1}{2^{2}} u_{i}^{30 (m)}, \frac{1}{2^{2}} u_{i}^{3 Q (1)}, &Center Dot; &Center Dot; &Center Dot;, \frac{1}{2^{2}} u_{i}^{3 Q (m)}, &Center Dot; &Center Dot; &Center Dot; &Center Dot; &Center Dot; \cdot, \frac{1}{2^{P - 1}} u_{i}^{PQ (L)}], u_{i}^{pq (m)} = {[u_{i}^{pq (m)} (0), \cdot \cdot \cdot, u_{i}^{pq (m)} (N - 1)]}^{T},$ when convergent should have

z_i=U_ik_i (7)其中，z_i=[Z_i(0),Z_i(1),...,Z_i(N-1)]^T, $k_{i} = {[K_{i}^{10 (1)}, \cdot \cdot \cdot, K_{i}^{1 q (1)}, K_{i}^{30 (1)}, \cdot \cdot \cdot, K_{i}^{30 (M)}, K_{i}^{3 Q (1)}, \cdot \cdot \cdot, K_{i}^{3 Q (M)} \cdot \cdot \cdot, \cdot \cdot \cdot, K_{i}^{PQ (L)}]}^{T} .$ 则通过LMS法可求得使收敛的k_i，z _i =U _i k _i (7) among them, z _i =[Z _i (0),Z _i (1),...,Z _i (N-1)] ^T , $k_{i} = {[K_{i}^{10 (1)}, &Center Dot; &Center Dot; &Center Dot;, K_{i}^{1 q (1)}, K_{i}^{30 (1)}, &Center Dot; \cdot \cdot, K_{i}^{30 (m)}, K_{i}^{3 Q (1)}, \cdot \cdot \cdot, K_{i}^{3 Q (m)} &Center Dot; &Center Dot; &Center Dot;, &Center Dot; &Center Dot; &Center Dot;, K_{i}^{PQ (L)}]}^{T} .$ Then the LMS method can be obtained so that convergent k _i ,

${\overset{^^}{k k}}_{i i} = = {(({U u}_{i i}^{H h} {U u}_{11}))}^{- - 11} {U u}_{i i}^{H h} {z z}_{i i} - - - - - - ((88))$

其中(.)^H表示集合的共轭转置。此处的即为我们所要求的式（5）各项系数的集合。将中的各项带入式（5），即可得到每路信号的基带数字预失真补偿模型。采用此模型对待传信号进行数字预失真，可补偿ROF链路引入的记忆效应和非线性交调失真。以此，提高频分复用信号所有频率分量的传输质量以及抑制邻信道干扰，同时能够提高系统传输信号的动态范围。where (.) ^H denotes the conjugate transpose of the set. here That is, the set of coefficients of formula (5) we require. Will The terms in (5) can be used to obtain the baseband digital pre-distortion compensation model of each signal. Using this model to carry out digital predistortion on the signal to be transmitted can compensate the memory effect and nonlinear intermodulation distortion introduced by the ROF link. In this way, the transmission quality of all frequency components of the frequency division multiplexing signal is improved and adjacent channel interference is suppressed, and the dynamic range of the system transmission signal can be improved at the same time.

本发明对频分复用信号上变频及频分复用之前的基带信号分别进行数字预失真处理，而忽略了频分复用后各路信号间的空白频段。因此在本发明中，数字预失真的处理带宽和数模转换器的带宽仅仅取决于每路信号的基带带宽而不是整个频分复用信号的带宽。这大大降低了数字预失真的硬件设备要求，使低成本的DSP,FPGA以及DAC等器件可用于宽带的射频频分复用光载无线系统中。The present invention performs digital predistortion processing on frequency division multiplexing signal up-conversion and baseband signal before frequency division multiplexing, and ignores blank frequency bands between signals after frequency division multiplexing. Therefore, in the present invention, the processing bandwidth of the digital predistortion and the bandwidth of the digital-to-analog converter only depend on the baseband bandwidth of each signal rather than the bandwidth of the entire frequency division multiplexing signal. This greatly reduces the hardware requirements for digital pre-distortion, so that low-cost DSP, FPGA, and DAC devices can be used in broadband radio frequency frequency division multiplexing wireless systems.

作为一个优选的实施方式，如图3所示为本发明在一个双信道的副载波复用光载无线系统中的非线性失真补偿系统的实验框图：As a preferred embodiment, as shown in Figure 3, it is an experimental block diagram of the nonlinear distortion compensation system of the present invention in a dual-channel subcarrier multiplexing radio-over-optical system:

两路随机的OFDM-64QAM信号（信号调制格式符合IEEE802.11g的物理层定义）在MATLAB中产生，并进行数字预失真处理。经过数字预失真的OFDM信号被输入到两个矢量信号发生源（VSG）中，实现数模转换及上变频。两个VSG之间通过10MHz的reference和trigger信号相连，以实现严格的符号同步。接下来，这两路信号分别被上变频到2.412GHz和3.6GHz的载波上，并用一个微波合路器合路，加载到直调的ROF链路上。在ROF链路的输出端，一个矢量信号分析仪（VSA）用于信号接收，解调以及将数模转换后的信号回传电脑以提取ROF链路的非线性系数。Two random OFDM-64QAM signals (the signal modulation format conforms to the physical layer definition of IEEE802.11g) are generated in MATLAB and processed by digital predistortion. The digitally predistorted OFDM signal is input to two vector signal generators (VSG) to realize digital-to-analog conversion and up-conversion. The two VSGs are connected through a 10MHz reference and trigger signal to achieve strict symbol synchronization. Next, the two signals are up-converted to 2.412GHz and 3.6GHz carriers respectively, combined with a microwave combiner, and loaded onto the directly modulated ROF link. At the output of the ROF link, a vector signal analyzer (VSA) is used for signal reception, demodulation, and the digital-to-analog converted signal is sent back to the computer to extract the nonlinear coefficients of the ROF link.

在本例中，我们考虑到3阶非线性及2阶记忆效应，因此，两个信道，f₁和f₂，上信号的数字预失真补偿模型的系数数组，和分别有9项。我们分别利用两个有8192个采样点的训练序列的输入与ROF链路输出值代入（8），即可求出和的值。In this example, we consider the third-order nonlinearity and the second-order memory effect, therefore, two channels, f ₁ and f ₂ , the coefficient array of the digital predistortion compensation model of the signal, and There are 9 items respectively. We use the input of the two training sequences with 8192 sampling points and the output value of the ROF link into (8), and we can find and value.

我们将上面得到的和代入数字预失真模块，用其对另外两路承载于2.412GHz和3.6GHz的20MHz带宽的64QAM-OFDM信号进行预失真，并在ROF输出端观察接收信号质量。图4是ROF系统接收信号的频谱图。由图4（a）可见，两路OFDM-64QAM信号的频谱分别占用了2.412GHz和3.6GHz的频点，此副载波复用光载无线系统的带宽约1.2GHz，这是传统DPD技术无法处理的带宽。然而采用本发明提出的多信道DPD技术，信号的ACP得到了极大的抑制。如图4（b），图4（c）所示，经过DPD的两路信号的ACP分别被抑制了14dB和15dB。We will get the above and Substitute into the digital pre-distortion module, use it to pre-distort the other two 64QAM-OFDM signals with 20MHz bandwidth carried at 2.412GHz and 3.6GHz, and observe the received signal quality at the ROF output. FIG. 4 is a spectrum diagram of a signal received by the ROF system. It can be seen from Figure 4(a) that the frequency spectrums of the two OFDM-64QAM signals occupy the frequency points of 2.412GHz and 3.6GHz respectively, and the bandwidth of this subcarrier multiplexed optical wireless system is about 1.2GHz, which cannot be handled by traditional DPD technology. bandwidth. However, by adopting the multi-channel DPD technology proposed by the present invention, the ACP of the signal is greatly suppressed. As shown in Figure 4(b) and Figure 4(c), the ACPs of the two signals passing through the DPD are suppressed by 14dB and 15dB respectively.

同时，我们还利用VSA解出了64QAM信号的EVM和星座图，如图5所示。由图可见，经过本发明处理过的信号接收到的EVM要比未处理的信号好很多。其中在输入功率为-8dBm时，经过本发明处理的两路信号的接收EVM分别为2.12%和2.33%，未经过处理的接收信号的EVM分别为6.13%和6.34%。从星座图中也可看出，经过本发明处理的信号星座图要优于未经过处理的信号接收的星座图。综上可知，采用本发明的多信道数字预失真处理方法及系统接收信号的EVM、邻信道泄漏功率（ACP）以及动态范围都得到了大幅度提高。At the same time, we also use VSA to solve the EVM and constellation diagram of the 64QAM signal, as shown in Figure 5. It can be seen from the figure that the received EVM of the signal processed by the present invention is much better than that of the unprocessed signal. When the input power is -8dBm, the received EVMs of the two signals processed by the present invention are 2.12% and 2.33% respectively, and the EVMs of the unprocessed received signals are 6.13% and 6.34% respectively. It can also be seen from the constellation diagram that the signal constellation diagram processed by the present invention is better than the received constellation diagram of the unprocessed signal. It can be seen from the above that the EVM, adjacent channel leakage power (ACP) and dynamic range of the received signal of the multi-channel digital pre-distortion processing method and system of the present invention are greatly improved.

采用本发明的技术方案可补偿ROF链路对频分复用信号引入的记忆效应和非线性交调失真，使整个系统实现线性传输，有效提高了频分复用信号所有频率分量的传输质量，抑制了邻信道干扰、增大了系统传输信号的动态范围。且由于本发明是针对频分复用的每路信号分别进行数字预失真，这大大降低了数字预失真的硬件设备要求，使低成本的DSP、FPGA以及DAC等器件可用于宽带的射频频分复用光载无线系统中。The technical solution of the present invention can compensate the memory effect and nonlinear intermodulation distortion introduced by the ROF link to the frequency division multiplexing signal, so that the entire system can realize linear transmission, effectively improving the transmission quality of all frequency components of the frequency division multiplexing signal, The adjacent channel interference is suppressed, and the dynamic range of the system transmission signal is increased. And because the present invention carries out digital pre-distortion for each channel signal of frequency division multiplexing, this greatly reduces the hardware equipment requirement of digital pre-distortion, makes devices such as low-cost DSP, FPGA and DAC can be used for the radio frequency frequency division of broadband In the multiplexing radio-over-fiber system.

本发明的多信道数字预失真处理方法及系统可适用于包括基于直调制以及外调制的任何光载无线链路，有效克服了基于直调的光载无线链路中多频点复用信号产生的记忆效应所产生的问题。The multi-channel digital pre-distortion processing method and system of the present invention can be applied to any optical wireless link based on direct modulation and external modulation, effectively overcoming the problem of multi-frequency point multiplexing signal generation in optical wireless links based on direct modulation problems caused by memory effects.

虽然结合附图描述了本发明的实施方式，但是本领域技术人员可以在不脱离本发明的精神和范围的情况下做出各种修改和变型，这样的修改和变型均落入由所附权利要求所限定的范围之内。Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present invention. within the bounds of the requirements.

Claims

1. A multi-channel digital predistortion processing method, characterized in that, comprising:

Predict the intermodulation distortion caused by the frequency division multiplexing signal in each channel and the frequency division multiplexing signal in other channels being simultaneously loaded to the optical wireless link for transmission after frequency multiplexing;

Compensating for intermodulation distortion between said frequency division multiplexed signals in different channels.

2. The multi-channel digital predistortion processing method according to claim 1, characterized in that, the intermodulation distortion between each of the frequency division multiplexing signals in different channels is obtained by a continuous signal transmission model.

3. multi-channel digital predistortion processing method according to claim 2, is characterized in that, described continuous signal transmission model is:

v _out =∑ _p K _p (v _in /g) ^p

Among them, v _in is the instantaneous input of the equivalent continuous signal to be processed by digital predistortion, g is the linear gain of the ROF link, K _p is the p-order nonlinear coefficient normalized by the memory polynomial, v _out is the digital predistortion The instantaneous output of the processed equivalent continuous signal.

4. multi-channel digital pre-distortion processing method according to claim 1, is characterized in that, the mathematical model that digital pre-distortion processing is carried out to frequency division multiplexing signal in each channel is:

{Z Z}_{i i} ((n no)) = = \underset{q q}{Σ Σ} \underset{p p}{Σ Σ} {K K}_{pq pq} \frac{{X x}_{i i} ((n no - - q q))}{| | {X x}_{i i} ((n no - - q q)) | |} \frac{11}{22^{p p - - 11}} \underset{{f f}_{i i} = = {Σ Σ}_{j j}^{p p} {f f}_{ij ij}}{Σ Σ} {Π Π}_{j j = = 11}^{p p} | | {X x}_{{i i}_{j j}} ((n no - - q q)) | |

Among them, X _i (n) is the complex amplitude of the signal on the i-th subcarrier, f _i is the carrier frequency of the signal on the i-th sub-carrier, K _pq is the normalization of the p-th order nonlinearity and q-th order memory effect term of the memory polynomial normalization coefficient, Z _i (n) is the output after digital pre-distortion processing.

5. multi-channel digital pre-distortion processing method according to claim 4, is characterized in that, the coefficient _K of the mathematical model that digital pre-distortion processing adopts in each channel frequency-division multiplexing signal obtains with the least mean square method.

6. The multi-channel digital pre-distortion processing method according to claim 1, characterized in that the multi-channel digital pre-distortion processing method is performed before the frequency division multiplexing signal is up-converted in each channel.

7. A multi-channel digital pre-distortion processing system, characterized in that, comprising:

A multi-channel digital pre-distortion module, used to establish a digital pre-processing model and perform digital pre-distortion processing on the input signal according to the digital pre-processing model to compensate for nonlinear distortion in the ROF link;

The digital predistortion training module is used to solve the parameters of the digital preprocessing model.

8. multi-channel digital pre-distortion processing system according to claim 7, is characterized in that, described digital pre-processing model is:

{Z Z}_{i i} ((n no)) = = \underset{q q}{Σ Σ} \underset{p p}{Σ Σ} {K K}_{pq pq} \frac{{X x}_{i i} ((n no - - q q))}{| | {X x}_{i i} ((n no - - q q)) | |} \frac{11}{22^{p p - - 11}} \underset{{f f}_{i i} = = {Σ Σ}_{j j}^{p p} {f f}_{ij ij}}{Σ Σ} {Π Π}_{j j = = 11}^{p p} | | {X x}_{{i i}_{j j}} ((n no - - q q)) | |

9. The multi-channel digital pre-distortion processing system according to claim 8, wherein the mathematical model used by the digital pre-distortion training module is the digital pre-processing model, and the digital pre-distortion training module passes through the The parameters of the digital preprocessing model are obtained by comparing the known training sequences.

10. multi-channel digital pre-distortion processing system according to claim 7, is characterized in that, described digital pre-distortion training module is arranged on the ROF link feedback loop, and described digital pre-distortion training module uses the normalization of ROF link The normalization output is input and the difference between the output of the digital pre-distortion training module and the output of the multi-channel digital pre-distortion module converges.