[go: up one dir, main page]

CN110740005B - Uplink channel prediction method and prediction system based on path division multiple access - Google Patents

Uplink channel prediction method and prediction system based on path division multiple access Download PDF

Info

Publication number
CN110740005B
CN110740005B CN201911021624.0A CN201911021624A CN110740005B CN 110740005 B CN110740005 B CN 110740005B CN 201911021624 A CN201911021624 A CN 201911021624A CN 110740005 B CN110740005 B CN 110740005B
Authority
CN
China
Prior art keywords
channel
module
matrix
user
prediction
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201911021624.0A
Other languages
Chinese (zh)
Other versions
CN110740005A (en
Inventor
张川
冀贞昊
尤肖虎
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zijinshan Laboratory
Original Assignee
Network Communication and Security Zijinshan Laboratory
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Network Communication and Security Zijinshan Laboratory filed Critical Network Communication and Security Zijinshan Laboratory
Priority to CN201911021624.0A priority Critical patent/CN110740005B/en
Publication of CN110740005A publication Critical patent/CN110740005A/en
Application granted granted Critical
Publication of CN110740005B publication Critical patent/CN110740005B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B17/00—Monitoring; Testing
    • H04B17/30—Monitoring; Testing of propagation channels
    • H04B17/373—Predicting channel quality or other radio frequency [RF] parameters
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B7/00—Radio transmission systems, i.e. using radiation field
    • H04B7/02—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas
    • H04B7/04—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas
    • H04B7/0413—MIMO systems
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00—Baseband systems
    • H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202—Channel estimation
    • H04L25/024—Channel estimation channel estimation algorithms
    • H04L25/0242—Channel estimation channel estimation algorithms using matrix methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Power Engineering (AREA)
  • Quality & Reliability (AREA)
  • Electromagnetism (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

本发明提出了一种基于路径分多址的上行链路信道预测方法和预测系统。根据大规模天线阵列中存在的空间宽带效应和频率选择效应，提出了面向使用正交频分复用技术的大规模多输入多输出天线系统的新的信道预测VLSI架构，设计了上行信道链路中信道预测模块。对于导频阶段，本发明基于流水线和脉动阵列技术，设计了预处理、预搜索、用户分组和信道特征搜索等模块；对于上行信道预测阶段，本发明设计了每个用户的上行信道估计模块。所有模块只包含复数加法、复数乘法以及寄存器，不包含其他复杂运算模块。

Figure 201911021624

The present invention provides an uplink channel prediction method and prediction system based on path division multiple access. According to the spatial broadband effect and frequency selection effect existing in large-scale antenna arrays, a new channel prediction VLSI architecture for large-scale multiple-input multiple-output antenna systems using orthogonal frequency division multiplexing technology is proposed, and the uplink channel link is designed. Medium channel prediction module. For the pilot phase, the present invention designs modules such as preprocessing, pre-search, user grouping and channel feature search based on pipeline and systolic array technology; for the uplink channel prediction phase, the present invention designs an uplink channel estimation module for each user. All modules only include complex addition, complex multiplication and registers, and do not include other complex arithmetic modules.

Figure 201911021624

Description

一种基于路径分多址的上行链路信道预测方法及预测系统An uplink channel prediction method and prediction system based on path division multiple access

技术领域technical field

本发明属于下一代无线移动通信技术领域，涉及一种基于路径分多址的上行链路信道预测方法及预测系统。The invention belongs to the technical field of next-generation wireless mobile communication, and relates to an uplink channel prediction method and prediction system based on path division multiple access.

背景技术Background technique

2017年，HongxiangXie,Feifei Gao等人在“A Unified Transmission Strategyfor TDD/FDD Massive MIMO Systems With Spatial Basis Expansion Model”提出针对大规模多输入多输出系统的基于角分多址的信道预测算法，根据角分多址技术，信道在角度域上表现出极大的稀疏性，进而可以通过捕捉信道的非零元素来实现信道预测；同时根据角度的互益性，给出适用于TDD系统和FDD系统的下行信道特征获取方式。In 2017, HongxiangXie, Feifei Gao and others proposed an angular division multiple access-based channel prediction algorithm for massive multiple-input multiple-output systems in "A Unified Transmission Strategy for TDD/FDD Massive MIMO Systems With Spatial Basis Expansion Model". With the multiple access technology, the channel shows great sparsity in the angle domain, and then the channel prediction can be realized by capturing the non-zero elements of the channel; at the same time, according to the mutual benefit of the angle, the downlink suitable for the TDD system and the FDD system is given. Channel feature acquisition method.

2019年，Xiaozhen Liu等人在“Efficient Channel Estimator With Angle-Division Multiple Access”针对角分多址技术提出了对应的高效VLSI架构，实现了对于单一载波系统的大规模MIMO系统上下行信道预测。In 2019, Xiaozhen Liu et al. proposed a corresponding efficient VLSI architecture for the angle division multiple access technology in "Efficient Channel Estimator With Angle-Division Multiple Access", which realized the uplink and downlink channel prediction of massive MIMO systems for a single carrier system.

随着现代移动通信系统的发展，正交频分复用技术逐渐称被通信系统广泛采用。2018年，Bolei Wang,Feifei Gao等人在“Spatial-and Frequency-Wideband Effects inMillimeter-Wave Massive MIMO Systems”，提了基于空间宽带效应和频率选择效应的路径分复用方法，从理论上证明了基于路径分复用进行信道预测的可行性。With the development of modern mobile communication systems, orthogonal frequency division multiplexing technology is gradually said to be widely used in communication systems. In 2018, Bolei Wang, Feifei Gao and others proposed a path division multiplexing method based on spatial broadband effect and frequency selection effect in "Spatial-and Frequency-Wideband Effects in Millimeter-Wave Massive MIMO Systems". The feasibility of path division multiplexing for channel prediction.

发明内容SUMMARY OF THE INVENTION

本发明的目的在于提供一种基于路径分多址的上行链路信道预测方法。The purpose of the present invention is to provide an uplink channel prediction method based on path division multiple access.

本发明提供的基于路径分多址的上行链路信道预测方法，包括，The uplink channel prediction method based on path division multiple access provided by the present invention includes:

(1)导频阶段初始预测阶段：预处理模块对接收到的第p个用户的信号Yp处理得到初始估计信道矩阵

Figure GDA0003347119330000011
所述初始估计信道矩阵
Figure GDA0003347119330000012
分别输入预搜索模块、信道特征搜索模块；所述预搜索模块根据初始估计信道矩阵
Figure GDA0003347119330000013
进行初始搜索得到第p个用户的信道特征
Figure GDA0003347119330000014
将信道特征
Figure GDA0003347119330000015
分别输入信道特征搜索模块、信道特征分组模块，其中l为用户p的第l个路径；(1) Initial prediction stage of pilot stage: the preprocessing module processes the received signal Yp of the p-th user to obtain the initial estimated channel matrix
Figure GDA0003347119330000011
The initial estimated channel matrix
Figure GDA0003347119330000012
Input the pre-search module and the channel feature search module respectively; the pre-search module estimates the channel matrix according to the initial
Figure GDA0003347119330000013
Perform an initial search to obtain the channel characteristics of the pth user
Figure GDA0003347119330000014
channel characteristics
Figure GDA0003347119330000015
Input the channel feature search module and the channel feature grouping module respectively, where l is the lth path of user p;

(2)信道特征预测分组和信道重构阶段：所述信道特征搜索模块对信道特征

Figure GDA0003347119330000021
进行精确预测，得到第p个用户的二维信道特征，并输出给重构模块；所述信道特征分组模块对信道特征进行分组并将分组后的信道特征存入分组寄存器，该分组后的信道特征用于指导上行链路信道预测模块进行预测；所述重构模块对二维信道特征进行重构，得到信道基矢量组w，信道基矢量组w包括第p个用户的Lp个路径的信道基矢量
Figure GDA0003347119330000022
所述信道基矢量组w输出至上行链路信道预测模块；(2) Channel feature prediction grouping and channel reconstruction stage: the channel feature search module
Figure GDA0003347119330000021
Carry out accurate prediction, obtain the two-dimensional channel characteristics of the pth user, and output them to the reconstruction module; the channel characteristic grouping module groups the channel characteristics and stores the grouped channel characteristics in the grouping register. The grouped channel characteristics The feature is used to guide the uplink channel prediction module to predict; the reconstruction module reconstructs the two-dimensional channel feature to obtain a channel basis vector group w, and the channel basis vector group w includes the channel of the Lp path of the pth user basis vector
Figure GDA0003347119330000022
the channel base vector group w is output to the uplink channel prediction module;

(3)上行信道预测阶段：所述上行链路信道预测模块根据接收到的信号Yp以及信道基矢量输出上行链路信道预测矩阵。(3) Uplink channel prediction stage: the uplink channel prediction module outputs an uplink channel prediction matrix according to the received signal Yp and the channel base vector.

优选地，所述步骤(1)中，所述预处理模块基于LS信道估计方式得到基矢量

Figure GDA0003347119330000023
将基矢量
Figure GDA0003347119330000024
进行串并转换得到初始估计信道矩阵
Figure GDA0003347119330000025
Preferably, in the step (1), the preprocessing module obtains the basis vector based on the LS channel estimation method
Figure GDA0003347119330000023
the basis vector
Figure GDA0003347119330000024
Perform serial-to-parallel conversion to obtain the initial estimated channel matrix
Figure GDA0003347119330000025

优选地，所述步骤(1)中，所述预搜索模块对初始估计信道矩阵

Figure GDA0003347119330000026
进行IFFT操作，得到在角度时延变换域中的信道预测矩阵：Preferably, in the step (1), the pre-search module initially estimates the channel matrix
Figure GDA0003347119330000026
Perform the IFFT operation to obtain the channel prediction matrix in the angular delay transform domain:

Figure GDA0003347119330000027
Figure GDA0003347119330000027

其中，

Figure GDA0003347119330000028
为M*M维傅里叶变换矩阵的转置共轭，
Figure GDA0003347119330000029
为N*N维傅里叶变换矩阵的共轭；对变换域矩阵
Figure GDA00033471193300000210
取模，比较出用户p的各路径的最大值二维坐标作为初始搜索得到的信道特征
Figure GDA00033471193300000211
in,
Figure GDA0003347119330000028
is the transposed conjugate of the M*M-dimensional Fourier transform matrix,
Figure GDA0003347119330000029
is the conjugate of the N*N-dimensional Fourier transform matrix; for the transform domain matrix
Figure GDA00033471193300000210
Take the modulo and compare the two-dimensional coordinates of the maximum value of each path of user p as the channel feature obtained by the initial search
Figure GDA00033471193300000211

优选地，所述步骤(2)中，所述信道特征搜索模块采用逐次二分迭代法在信道特征

Figure GDA00033471193300000212
附近搜索满足
Figure GDA00033471193300000213
的最大值坐标，输出二维信道特征，其中，
Figure GDA00033471193300000214
为M*M维傅里叶变换矩阵的转置共轭，
Figure GDA00033471193300000215
为N*N维傅里叶变换矩阵的共轭，ΨM，ΨN分别为角度域旋转因子和时延域选择因子，Θ为相移矩阵。Preferably, in the step (2), the channel feature search module adopts the successive bisection iteration method to search for the channel feature
Figure GDA00033471193300000212
Nearby Search Satisfaction
Figure GDA00033471193300000213
The maximum coordinate of , outputs the two-dimensional channel feature, where,
Figure GDA00033471193300000214
is the transposed conjugate of the M*M-dimensional Fourier transform matrix,
Figure GDA00033471193300000215
is the conjugate of the N*N-dimensional Fourier transform matrix, Ψ M , Ψ N are the rotation factor in the angle domain and the selection factor in the delay domain, respectively, and Θ is the phase shift matrix.

优选地，所述步骤(3)中，所述上行链路信道预测模块计算出用户p的每一条路径的信道增益Preferably, in the step (3), the uplink channel prediction module calculates the channel gain of each path of the user p

Figure GDA0003347119330000031
Figure GDA0003347119330000031

其中，Ep为第p个用户的限制功率，vec(Yp)为转化为向量输入的接收信号，

Figure GDA0003347119330000032
为用户p的各路径的最大值二维坐标作为初始搜索得到的信道特征；根据计算得到的信道增益，通过与对应的信道基矢量组w相乘并求和得到用户p的信道向量，再将其调整为矩阵形式得到最终的第p个用户的上行链路信道预测矩阵为Among them, E p is the limited power of the p-th user, vec(Y p ) is the received signal converted into vector input,
Figure GDA0003347119330000032
is the maximum two-dimensional coordinate of each path of user p as the channel feature obtained by the initial search; according to the calculated channel gain, the channel vector of user p is obtained by multiplying and summing the corresponding channel base vector group w, and then It is adjusted to matrix form to obtain the final uplink channel prediction matrix of the pth user as

Figure GDA0003347119330000033
其中，
Figure GDA00033471193300000311
为第p个用户对应的路径集，对每一个路径的信道基矢量：
Figure GDA0003347119330000034
其中，Θp，I为用户p的第l条路径的相移矩阵，bp，l为频域方向矢量，ap，l为角度方向矩阵。
Figure GDA0003347119330000033
in,
Figure GDA00033471193300000311
For the path set corresponding to the pth user, the channel basis vector for each path:
Figure GDA0003347119330000034
Among them, Θ p, I is the phase shift matrix of the lth path of user p, b p, l is the frequency domain direction vector, and a p, l is the angle direction matrix.

本发明还提供一种基于路径分多址的上行链路信道预测系统，包括预处理模块、预搜索模块、信道特征搜索模块、信道特征分组模块、重构模块、上行链路信道预测模块；The present invention also provides an uplink channel prediction system based on path division multiple access, comprising a preprocessing module, a presearch module, a channel feature search module, a channel feature grouping module, a reconstruction module, and an uplink channel prediction module;

所述预处理模块用于对接收到的第p个用户的信号Yp处理得到初始估计信道矩阵

Figure GDA0003347119330000035
并分别输入预搜索模块、信道特征搜索模块；The preprocessing module is used to process the received signal Yp of the pth user to obtain an initial estimated channel matrix
Figure GDA0003347119330000035
And input the pre-search module and the channel feature search module respectively;

所述预搜索模块用于根据初始估计信道矩阵

Figure GDA0003347119330000036
进行初始搜索得到第p个用户的信道特征
Figure GDA0003347119330000037
将信道特征
Figure GDA0003347119330000038
分别输入信道特征搜索模块、信道特征分组模块，其中l为用户p的第l个路径；The pre-search module is used to estimate the channel matrix according to the initial
Figure GDA0003347119330000036
Perform an initial search to obtain the channel characteristics of the pth user
Figure GDA0003347119330000037
channel characteristics
Figure GDA0003347119330000038
Input the channel feature search module and the channel feature grouping module respectively, where l is the lth path of user p;

所述信道特征搜索模块用于对信道特征

Figure GDA0003347119330000039
进行精确预测，得到第p个用户的二维信道特征，并输出给重构模块；The channel feature search module is used to
Figure GDA0003347119330000039
Accurate prediction is performed to obtain the two-dimensional channel characteristics of the pth user, and output to the reconstruction module;

所述信道特征分组模块用于对信道特征进行分组并将分组后的信道特征存入分组寄存器，该分组后的信道特征用于指导上行链路信道预测模块进行预测；The channel feature grouping module is used to group the channel features and store the grouped channel features into the grouping register, and the grouped channel features are used to instruct the uplink channel prediction module to predict;

所述重构模块用于对二维信道特征进行重构，得到信道基矢量组w，信道基矢量组w包括第p个用户的Lp个路径的信道基矢量

Figure GDA00033471193300000310
输出所述信道基矢量组w至上行链路信道预测模块；The reconstruction module is used to reconstruct the two-dimensional channel characteristics to obtain a channel basis vector group w, where the channel basis vector group w includes the channel basis vectors of the Lp paths of the pth user
Figure GDA00033471193300000310
outputting the channel base vector group w to the uplink channel prediction module;

所述上行链路信道预测模块用于根据接收到的信号Yp以及信道基矢量输出上行链路信道预测矩阵。The uplink channel prediction module is configured to output the uplink channel prediction matrix according to the received signal Yp and the channel base vector.

优选地，所述预处理模块基于LS信道估计方式得到基矢量

Figure GDA0003347119330000041
将基矢量
Figure GDA0003347119330000042
进行串并转换得到初始估计信道矩阵
Figure GDA0003347119330000043
Preferably, the preprocessing module obtains the basis vector based on the LS channel estimation method
Figure GDA0003347119330000041
the basis vector
Figure GDA0003347119330000042
Perform serial-to-parallel conversion to obtain the initial estimated channel matrix
Figure GDA0003347119330000043

优选地，所述预搜索模块对初始估计信道矩阵

Figure GDA0003347119330000044
进行IFFT操作，得到在角度时延变换域中的信道预测矩阵：Preferably, the pre-search module initially estimates the channel matrix
Figure GDA0003347119330000044
Perform the IFFT operation to obtain the channel prediction matrix in the angular delay transform domain:

Figure GDA0003347119330000045
Figure GDA0003347119330000045

其中，

Figure GDA0003347119330000046
为M*M维傅里叶变换矩阵的转置共轭，
Figure GDA0003347119330000047
为N*N维傅里叶变换矩阵的共轭；对变换域矩阵
Figure GDA0003347119330000048
取模，比较出用户p的各路径的最大值二维坐标作为初始搜索得到的信道特征
Figure GDA0003347119330000049
in,
Figure GDA0003347119330000046
is the transposed conjugate of the M*M-dimensional Fourier transform matrix,
Figure GDA0003347119330000047
is the conjugate of the N*N-dimensional Fourier transform matrix; for the transform domain matrix
Figure GDA0003347119330000048
Take the modulo and compare the two-dimensional coordinates of the maximum value of each path of user p as the channel feature obtained by the initial search
Figure GDA0003347119330000049

优选地，所述信道特征搜索模块采用逐次二分迭代法在信道特征

Figure GDA00033471193300000410
附近搜索满足
Figure GDA00033471193300000411
的最大值坐标，输出二维信道特征，其中，
Figure GDA00033471193300000412
为M*M维傅里叶变换矩阵的转置共轭，
Figure GDA00033471193300000413
为N*N维傅里叶变换矩阵的共轭，ΨM，ΨN分别为角度域旋转因子和时延域选择因子，Θ为相移矩阵。Preferably, the channel feature search module adopts the successive bisection iteration method in the channel feature
Figure GDA00033471193300000410
Nearby Search Satisfaction
Figure GDA00033471193300000411
The maximum coordinate of , outputs the two-dimensional channel feature, where,
Figure GDA00033471193300000412
is the transposed conjugate of the M*M-dimensional Fourier transform matrix,
Figure GDA00033471193300000413
is the conjugate of the N*N-dimensional Fourier transform matrix, Ψ M , Ψ N are the rotation factor in the angle domain and the selection factor in the delay domain, respectively, and Θ is the phase shift matrix.

优选地，所述上行链路信道预测模块计算出用户p的每一条路径的信道增益Preferably, the uplink channel prediction module calculates the channel gain of each path of user p

Figure GDA00033471193300000414
Figure GDA00033471193300000414

其中，Ep为第p个用户的限制功率，vec(Yp)为转化为向量输入的接收信号，

Figure GDA00033471193300000415
为用户p的各路径的最大值二维坐标作为初始搜索得到的信道特征；根据计算得到的信道增益，通过与对应的信道基矢量组w相乘并求和得到用户p的信道向量，再将其调整为矩阵形式得到最终的第p个用户的上行链路信道预测矩阵为Among them, E p is the limited power of the p-th user, vec(Y p ) is the received signal converted into vector input,
Figure GDA00033471193300000415
is the maximum two-dimensional coordinate of each path of user p as the channel feature obtained by the initial search; according to the calculated channel gain, the channel vector of user p is obtained by multiplying and summing the corresponding channel base vector group w, and then It is adjusted to matrix form to obtain the final uplink channel prediction matrix of the pth user as

Figure GDA00033471193300000416
其中，
Figure GDA00033471193300000417
为第p个用户对应的路径集，对每一个路径的信道基矢量：
Figure GDA0003347119330000051
其中，Θp，I为用户p的第l条路径的相移矩阵，bp，l为频域方向矢量，ap，l为角度方向矩阵。
Figure GDA00033471193300000416
in,
Figure GDA00033471193300000417
For the path set corresponding to the pth user, the channel basis vector for each path:
Figure GDA0003347119330000051
Among them, Θ p, I is the phase shift matrix of the lth path of user p, b p, l is the frequency domain direction vector, and a p, l is the angle direction matrix.

本发明在硬件上实现了路径分多址信道预测方式，并且不存在高复杂度运算模块，同时架构中采用流水线和脉动阵列，大大提高了系统的效率。The invention realizes the path division multiple access channel prediction mode in hardware, and there is no high-complexity operation module, and meanwhile, pipeline and systolic array are adopted in the structure, which greatly improves the efficiency of the system.

附图说明Description of drawings

图1为信道预处理模块示意图；1 is a schematic diagram of a channel preprocessing module;

图2为逐次二分搜索示意图；Fig. 2 is a schematic diagram of successive binary search;

图3为预搜索模块示意图；3 is a schematic diagram of a pre-search module;

图4为信道特征搜索模块示意图；4 is a schematic diagram of a channel feature search module;

图5为信道特征分组模块示意图；5 is a schematic diagram of a channel feature grouping module;

图6为第p个用户上行链路信道预测模块；Fig. 6 is the p-th user uplink channel prediction module;

图7为分组子模块示意图；7 is a schematic diagram of a grouping submodule;

图8为本发明整体框架示意图。FIG. 8 is a schematic diagram of the overall framework of the present invention.

具体实施方式Detailed ways

下面结合附图和具体实例对本发明作进一步说明。The present invention will be further described below in conjunction with the accompanying drawings and specific examples.

本发明针对使用正交频分复用技术和面向大规模多输入多输出天线系统，在考虑空间宽带效应以及频率宽带效应的双宽带效应下，提出基于路径分多址的上行链路信道预测VLSI架构。Aiming at using the orthogonal frequency division multiplexing technology and facing the large-scale multiple-input multiple-output antenna system, the present invention proposes an uplink channel prediction VLSI based on path division multiple access under the consideration of the space broadband effect and the double broadband effect of the frequency broadband effect. Architecture.

从图1可以看到，接收到的输入数据Yp从数据缓冲器开始分两条路径分别进入UL估计模块(即上行链路信道预测模块)和LS信道估计模块，LS信道估计模块的输出

Figure GDA0003347119330000052
经过串并转换器后以信道矩阵的格式输出到信道特征搜索模块和预搜索模块。结合图8的整体预测框图，对于导频阶段，本发明基于流水线和脉动阵列技术，设计了预处理、预搜索、用户分组和信道特征搜索等模块；对于上行信道预测阶段，本发明设计了每个用户的上行信道估计模块，所有模块只包含复数加法、复数乘法以及寄存器，不包含其他复杂运算模块。从第1个用户到第p个用户接收到的信号为M*N维信号Yp，其中M为发射天线数，N为正交频分复用的子载波数。首先ImadBarhumi等人在文献“Optimal Training Design for MIMO OFDMSystemsin Mobile Wireless Channels”(期刊名称IEEE Trans Signal Processing，发表时间2003年，)中设计的基于LS信道估计方式得到初始估计得到的信道矩阵基矢量
Figure GDA0003347119330000061
通过将得到的基矢量
Figure GDA0003347119330000062
进行串并转换得到对应的初始估计信道矩阵
Figure GDA0003347119330000063
从而，将初始估计信道矩阵
Figure GDA0003347119330000064
输入到预搜索模块进行变换域信道特征搜索，同时作为输入数据流数据输出到信道特征搜索模块进行精确地信道特征搜索。As can be seen from Figure 1, the received input data Yp is divided into two paths from the data buffer to enter the UL estimation module (ie the uplink channel prediction module) and the LS channel estimation module respectively. The output of the LS channel estimation module
Figure GDA0003347119330000052
After the serial-to-parallel converter, it is output to the channel feature search module and the pre-search module in the format of the channel matrix. Combined with the overall prediction block diagram of Fig. 8, for the pilot stage, the present invention designs modules such as preprocessing, pre-search, user grouping and channel feature search based on pipeline and systolic array technology; for the uplink channel prediction stage, the present invention designs each module. Upstream channel estimation modules for each user, all modules only include complex addition, complex multiplication and registers, and do not include other complex operation modules. The signals received from the first user to the pth user are M*N dimensional signals Y p , where M is the number of transmit antennas, and N is the number of sub-carriers for OFDM. First, Imad Barhumi et al. in the document "Optimal Training Design for MIMO OFDM Systems in Mobile Wireless Channels" (journal name IEEE Trans Signal Processing, published in 2003) designed the channel matrix basis vector based on the initial estimation based on the LS channel estimation method.
Figure GDA0003347119330000061
The basis vector obtained by
Figure GDA0003347119330000062
Perform serial-to-parallel conversion to obtain the corresponding initial estimated channel matrix
Figure GDA0003347119330000063
Thus, the initial estimated channel matrix will be
Figure GDA0003347119330000064
Input to the pre-search module for channel feature search in transform domain, and output to the channel feature search module as input data stream data for accurate channel feature search.

如图3所示，经过预处理模块得到的M*N维的信道矩阵

Figure GDA0003347119330000065
输入到预搜索模块中，经过预搜索模块的处理得到信道特征初始估计值，最终，输出初始信道特征到信道特征分组模块进行信道特征分组和信道特征搜索模块进行精确搜索。As shown in Figure 3, the M*N-dimensional channel matrix obtained by the preprocessing module
Figure GDA0003347119330000065
The input is input into the pre-search module, and the initial estimated value of the channel characteristics is obtained through the processing of the pre-search module. Finally, the initial channel characteristics are output to the channel characteristic grouping module for channel characteristic grouping and channel characteristic search module for accurate search.

当初始估计信道矩阵

Figure GDA0003347119330000066
输入到预搜索模块时，首先进行IFFT操作，得到在角度时延变换域中的信道预测矩阵：When initially estimating the channel matrix
Figure GDA0003347119330000066
When input to the pre-search module, the IFFT operation is first performed to obtain the channel prediction matrix in the angle-delay transform domain:

Figure GDA0003347119330000067
Figure GDA0003347119330000067

其中，

Figure GDA0003347119330000068
为M*M维傅里叶变换矩阵的转置共轭，
Figure GDA0003347119330000069
为N*N维傅里叶变换矩阵的共轭。in,
Figure GDA0003347119330000068
is the transposed conjugate of the M*M-dimensional Fourier transform matrix,
Figure GDA0003347119330000069
is the conjugate of the N*N-dimensional Fourier transform matrix.

本发明为了降低硬件实现复杂度，首先参考FFT架构对列分别进行IFFT运算，其次对输出矩阵进行转置再输入原有的IFFT模块进行列IFFT运算，最终输出变换域矩阵

Figure GDA00033471193300000610
随后，通过取模模块对变换域矩阵
Figure GDA00033471193300000611
取模，比较出每个用户的各路径的最大值二维坐标作为初始搜索得到的信道特征
Figure GDA00033471193300000612
l为第p个用户的第l个路径，
Figure GDA00033471193300000613
为角度域的初始预测值，
Figure GDA00033471193300000614
为时延域的初始预测值，并将每一个用户所对应的信道特征集合作为
Figure GDA00033471193300000615
接下来，将信道特征集合
Figure GDA00033471193300000616
分别输入到信道特征搜索模块中进行精确搜索和输入到信道特征分组模块中进行信道特征分组。In order to reduce the complexity of hardware implementation, the present invention firstly performs IFFT operations on the columns with reference to the FFT architecture, then transposes the output matrix and then inputs the original IFFT module to perform column IFFT operations, and finally outputs a transform domain matrix
Figure GDA00033471193300000610
Then, the transform domain matrix is
Figure GDA00033471193300000611
Take the modulo and compare the maximum two-dimensional coordinates of each path of each user as the channel feature obtained by the initial search
Figure GDA00033471193300000612
l is the lth path of the pth user,
Figure GDA00033471193300000613
is the initial predicted value of the angle domain,
Figure GDA00033471193300000614
is the initial prediction value in the delay domain, and the channel feature set corresponding to each user is used as
Figure GDA00033471193300000615
Next, set the channel features
Figure GDA00033471193300000616
They are respectively input into the channel feature search module for precise search and into the channel feature grouping module for channel feature grouping.

由于信道中的干扰问题，初步预测的得到的信道特征只能精确到整数位(在恶劣的信道条件下，甚至整数位将发生偏移)，所以需要采用二维旋转措施，将变换域中的得到的矩阵，分别左乘和右乘角度旋转和时延旋转因子，再进行二维搜索。考虑到定步长二维搜索问满足实际需要进行的搜索次数与步长成反比，所以本发明采用基于逐次二分反馈搜索，对根据搜索树的每一层不断进行迭代，达到更高精度的搜索。如图4所示，在信道特征搜索模块中，对逐次二分搜索的方法提出相应的VLSI结构，通过控制反馈回路不断进行循环，并调整寄存器1和寄存器2中的存储的搜索节点数据和当前步长。在这个模块中，存在ζ个子模块，对应着本层次需要搜索的点数。在每个子模块中，首先对初始估计信道矩阵乘以相移矩阵Θ(根据寄存器中存储的上一次迭代的信息生成的)的共轭，再进行二维旋转(根据寄存器中存储的上一次迭代的信息生成)和二维离散傅里叶反变换，为了降低复杂度，我们只需要计算初始估计信道矩阵中，在预搜索模块得到初始信道特征点处的，变换域中旋转之后信道值。之后，选取ζ个子模块中输出最大的那个模块中的信道特征作为本次迭代输出，进而选择就此输出或进行下一次迭代。最终，经过满足迭代精度的迭代次数之后，输出精确的二维信道特征。Due to the interference problem in the channel, the initially predicted channel characteristics can only be accurate to integer bits (in poor channel conditions, even integer bits will be offset), so it is necessary to use two-dimensional rotation measures to convert the The obtained matrix is left-multiplied and right-multiplied by the angle rotation and delay rotation factors, respectively, and then a two-dimensional search is performed. Considering that the number of searches performed to meet the actual needs of a two-dimensional search with a fixed step size is inversely proportional to the step size, the present invention adopts a search based on successive binary feedback, and iterates continuously according to each layer of the search tree to achieve a higher-precision search. . As shown in Figure 4, in the channel feature search module, the corresponding VLSI structure is proposed for the successive binary search method, and the loop is continuously looped through the control feedback loop, and the stored search node data and the current step in register 1 and register 2 are adjusted. long. In this module, there are ζ sub-modules, corresponding to the number of points to be searched in this level. In each sub-module, the initial estimated channel matrix is first multiplied by the conjugate of the phase shift matrix Θ (generated from the information stored in the register of the previous iteration), and then a two-dimensional rotation is performed (based on the previous iteration stored in the register). information generation) and two-dimensional inverse discrete Fourier transform, in order to reduce the complexity, we only need to calculate the initial estimated channel matrix, in the pre-search module to obtain the initial channel feature points, the channel value after rotation in the transform domain. After that, the channel feature in the module with the largest output among the ζ sub-modules is selected as the output of this iteration, and then the output is selected or the next iteration is performed. Finally, after the number of iterations satisfying the iterative precision, accurate two-dimensional channel features are output.

信道特征集合

Figure GDA0003347119330000071
输入到信道特征预测模块后，要进行精确预测，结合图2的迭代过程要进行逐层搜索，在原有的坐标(已得到的信道特征集合)附近搜索满足
Figure GDA0003347119330000072
的最大值坐标，其中ΨM，ΨN分别为角度域旋转因子和时延域选择因子，Θ为相移矩阵。通过前文所述的逐次二分迭代以及其对应的迭代硬件架构，可以最终得到第p个用户满足精度的精确信道二维集合
Figure GDA0003347119330000073
。如图2所示，所示，从根节点开始，搜索逐步精确，在第一个层次中继承节点位Pre-0,同时在此层次中最大节点位Max-1，则Max-1节点成为根节点，进而派生出下一个层次。每个层次根节点派生出三个子节点，其中每个子节点取值为上一层次的根节点值与相邻节点的平均，计算的范围每一层将控制在生成出的两个节点中，那么在搜索的范围只控制在原有比较派生节点与根节点三个点上。通过从层级1到层次n,搜索的精度每层将提高50％，在保证精度的条件下，相比于固定长度的搜索，大大减少了搜索点数。channel feature set
Figure GDA0003347119330000071
After input to the channel feature prediction module, accurate prediction is required, and a layer-by-layer search is performed in combination with the iterative process in Figure 2, and the search is performed near the original coordinates (the obtained channel feature set) to satisfy
Figure GDA0003347119330000072
The maximum coordinates of , where Ψ M , Ψ N are the rotation factor in the angle domain and the selection factor in the delay domain, respectively, and Θ is the phase shift matrix. Through the successive bisection iterations described above and its corresponding iterative hardware architecture, the precise two-dimensional set of channels that satisfy the accuracy of the pth user can be finally obtained.
Figure GDA0003347119330000073
. As shown in Figure 2, as shown, starting from the root node, the search is gradually accurate, and the node position Pre-0 is inherited in the first level, while the largest node in this level is Max-1, then the Max-1 node becomes the root node, which in turn derives the next level. The root node of each level derives three sub-nodes, and each sub-node is the average of the value of the root node of the previous level and the adjacent nodes. The calculation range of each layer will be controlled in the two generated nodes, then The scope of the search is only controlled at the three points of the original comparison of the derived node and the root node. By going from level 1 to level n, the accuracy of the search will be improved by 50% for each layer, and the number of search points is greatly reduced compared to the fixed-length search under the condition of guaranteed accuracy.

另一方面，与信道特征精确搜索同时进行的是信道特征的分组。经过初始搜索的信道特征将输入信道特征分组模块，对于多用户并行输入的情况，我们需要进行并行分组，首先，根据每个用户的一维展开坐标进行排序，根据并行和串行输入分别可以采取现有的并行排序网络排序或串行冒泡排序等策略进行排序，排序之后的用户信道信息将通过设计的分组脉动阵列，比较各路径坐标的欧几里得距离与设置阈值之间的关系，在对应的处理模块输出相应用户。如图5所示，每个用户的信道特征并行输入到信道特征分组模块，通过排序网络以及分组模块，其中，分组模块由n个分组子模块组成的脉动阵列组成，每个子模块对应一个分组的输出，同时，将每个用户的分组信息储存在每一个用户的分组信息寄存器中，将进一步指导上行链路信道预测模块对信道预测进行分组训练。如图7所示，由n个分组子模块组成了适用于n分组的脉动阵列，通过计算前后输入的信道特征的几何距离，与分组阈值Ω进行比较，其中分组阈值Ω表征系统对各分组之间重叠的容忍程度，阈值设置越小，系统分组精度越高，训练成本量越少，但是会导致分组数过多造成后续负担，具体阈值选择由系统实际需求进行调整。通过比较的结果，选择是输出到下一个模块还是就在此模块输出，若排序后相邻的两个元素距离大于阈值，那种可以认为它们不重叠，可以放在同一个组中进行训练。若小于阈值，那么三态门导通，将此信息输出到下一个分组子模块中进行比较。进而在每一个子模块输出不同组的分组信息。On the other hand, the grouping of channel features is performed concurrently with the precise search of channel features. The channel features after the initial search will be input into the channel feature grouping module. For the case of multi-user parallel input, we need to perform parallel grouping. First, sort according to the one-dimensional expansion coordinates of each user. According to the parallel and serial input, we can take The existing parallel sorting network sorting or serial bubble sorting and other strategies are used for sorting, and the user channel information after sorting will pass through the designed grouping systolic array to compare the relationship between the Euclidean distance of each path coordinate and the set threshold. The corresponding user is output in the corresponding processing module. As shown in Figure 5, the channel characteristics of each user are input into the channel characteristic grouping module in parallel, through the sorting network and the grouping module, wherein the grouping module is composed of a systolic array composed of n grouping sub-modules, each sub-module At the same time, the grouping information of each user is stored in the grouping information register of each user, which will further instruct the uplink channel prediction module to perform grouping training on channel prediction. As shown in Figure 7, a systolic array suitable for n groups is composed of n grouping sub-modules. By calculating the geometric distance of the channel features input before and after, and comparing with the grouping threshold Ω, the grouping threshold Ω represents the system's effect on each grouping. The smaller the threshold setting is, the higher the system grouping accuracy will be, and the lower the training cost will be. However, it will cause too many groups to cause subsequent burdens. The specific threshold selection is adjusted according to the actual needs of the system. By comparing the results, choose whether to output to the next module or just this module. If the distance between two adjacent elements after sorting is greater than the threshold, it can be considered that they do not overlap and can be placed in the same group for training. If it is less than the threshold, the tri-state gate is turned on, and this information is output to the next grouping sub-module for comparison. Further, different groups of grouping information are output in each sub-module.

经过初始估计的信道特征输入到信道特征分组模块之后，首先要经过排序网络进行并行排序，如图5和7所示经过排序之后的用户特征输入到分组脉动阵列中，在每一个脉动阵列子模块中，将计算输入的用户信道特征坐标之间的几何距离，根据与分组阈值Ω比较的结果，在不同的组输出不同用户，进而实现对各信道特征的分组，分组信息将被存储进每一个用户信道特征分组寄存器中，在寄存器中存储的信道特征将指导下一阶段上行链路信道预测模块进行分组训练，图7中的UL估计器即上行链路信道预测模块。After the initial estimated channel features are input into the channel feature grouping module, they must first be sorted in parallel through the sorting network. As shown in Figures 5 and 7, the sorted user features are input into the grouped systolic array. In the calculation, the geometric distance between the input user channel feature coordinates will be calculated, and according to the result of comparing with the grouping threshold Ω, different users will be output in different groups, and then the grouping of each channel feature will be realized, and the grouping information will be stored in each In the user channel feature grouping register, the channel features stored in the register will guide the next stage uplink channel prediction module to perform grouping training, and the UL estimator in FIG. 7 is the uplink channel prediction module.

经过信道特征搜索模块精确搜索的信道信息将进入信道重构模块，得到重构后的第p个用户的信道基矢量组w，信道基矢量组w包括第p个用户的Lp个路径的信道基矢量

Figure GDA0003347119330000081
对每一个路径的信道基矢量：
Figure GDA0003347119330000082
其中，Θp，I为用户p的第l条路径的相移矩阵，bp，l为频域方向矢量，ap，l为角度方向矩阵，反映信道特征。之后，信道基矢量组w将与接收到的信号Yp同时输入到上行链路信道预测模块中，在信道特征分组模块的寄存器中存储的分组信息指导下进行上行信道预测。其中，分组信息将把同一组的用户分组模块安排到一起训练，同时可以调整导频训练集，由于同一组内的用户特征不重叠，所以可以共用导频序列，进而提高训练速度，适应快变信道。The channel information accurately searched by the channel feature search module will enter the channel reconstruction module, and the reconstructed channel basis vector group w of the pth user will be obtained. The channel basis vector group w includes the channel basis of the Lp paths of the pth user. vector
Figure GDA0003347119330000081
Channel basis vector for each path:
Figure GDA0003347119330000082
Among them, Θ p, I is the phase shift matrix of the lth path of user p, b p, l is the frequency domain direction vector, and a p, l is the angle direction matrix, reflecting the channel characteristics. After that, the channel base vector group w is input into the uplink channel prediction module simultaneously with the received signal Yp , and the uplink channel prediction is performed under the guidance of the grouping information stored in the register of the channel feature grouping module. Among them, the grouping information will arrange the user grouping modules of the same group to train together, and at the same time, the pilot training set can be adjusted. Since the user characteristics in the same group do not overlap, the pilot sequence can be shared, thereby improving the training speed and adapting to rapid changes. channel.

如图6所示，每个用户将使用一个上行链路信道预测模块，通过输入对应的接收到的信号Yp和信道重构出的信道基矢量组w，输出最终的第p个用户的上行链路的信道预测矩阵，即，将输入信号Yp进行串行转化，将展开的信号vec(Yp)与第p个用户的信道基矢量组w中的每一个路径的信道基矢量的转置共轭做矢量相乘得到第p个用户的第l个路径的信道增益，之后，将信道增益与对应的信道基矢量相乘得到第p个用户的第l个路径的信道矢量，将Lp个路径相加，并将矢量分隔进行输出，最终得到第p个用户的最终上行链路信道预测矩阵

Figure GDA0003347119330000091
接收机得到信道预测矩阵作为冲击响应，为下一步信号的解调提供信息。在上行链路信道预测模块中，首先要预测计算出每一用户的每一条路径的信道增益As shown in Figure 6, each user will use an uplink channel prediction module to output the final uplink channel of the pth user by inputting the corresponding received signal Yp and the channel basis vector group w reconstructed from the channel. The channel prediction matrix of the link, that is, the input signal Y p is serially converted, and the unrolled signal vec(Y p ) is converted to the channel basis vector of each path in the channel basis vector group w of the pth user. Set the conjugate to do vector multiplication to obtain the channel gain of the lth path of the pth user, and then multiply the channel gain by the corresponding channel base vector to obtain the channel vector of the lth path of the pth user. The paths are added, and the vectors are separated for output, and finally the final uplink channel prediction matrix of the pth user is obtained.
Figure GDA0003347119330000091
The receiver obtains the channel prediction matrix as the impulse response, which provides information for the demodulation of the next signal. In the uplink channel prediction module, firstly, the channel gain of each path of each user should be predicted and calculated

Figure GDA0003347119330000092
Figure GDA0003347119330000092

其中，Ep为第p个用户的限制功率，vec(Yp)为转化为向量输入的接收信号。Among them, E p is the limited power of the p-th user, and vec(Y p ) is the received signal converted into a vector input.

根据计算得到的信道增益，通过与对应的信道基矢量组w相乘并求和得到每个用户的信道向量，再将其重新调整为矩阵形式(使用shape来表示)得到最终的第p个用户的上行链路信道预测矩阵为According to the calculated channel gain, the channel vector of each user is obtained by multiplying and summing the corresponding channel base vector group w, and then readjusting it into a matrix form (represented by shape) to obtain the final p-th user The uplink channel prediction matrix of is

Figure GDA0003347119330000093
其中，
Figure GDA0003347119330000094
为第p个用户对应的路径集。
Figure GDA0003347119330000093
in,
Figure GDA0003347119330000094
is the path set corresponding to the pth user.

本发明基于面向大规模多输入多输出天线系统并结合目前主流的正交频分复用技术，根据路径分多址接入方式设计了一种低复杂度的VLSI架构，可以高效实现上行链路信道预测。在本发明中，首次将设计出适用于路径分多址的信道预测架构，路径分多址考虑联合时延和角度域进行信道预测，通过精确预测波达方向、路径增益等参数，进行信道重构。但是本身算法存在大规模矩阵求傅里叶反变换，高精度二维特征搜索等复杂算法，难以在实际应用中部署。本发明针对原有算法中存在高复杂度运算难以在硬件上高速实施的问题，第一次提出路径分多址对应的VLSI硬件架构，以方便算法在FPGA或ASIC上进行硬件实现。在原有的信道预测和跟踪算法中，大多数均采用对信道矩阵的直接预测和跟踪，由于在大规模多输入多输出天线系统中，信道矩阵将随着天线数的增多而迅速扩大，这导致原有的算法难以高效地实现。本发明采取基于路径分多址地预测只需计算出每一条路径的路径参数，进而重构出这一路径对应的信道矩阵，这种方式大大降低了计算量。同时，本发明主要设计的VLSI架构中，采用了将模块流水线化并将原有的算法中的二维搜索算法用逐次二分反馈搜索取代，使得参数更为准确的被搜索，进而使得信道预测效率大幅度提高，而对应的硬件复杂度由于单路反馈不断复用模块而并没有显著增加，降低了硬件成本。Based on the large-scale multiple-input multiple-output antenna system and the current mainstream orthogonal frequency division multiplexing technology, the present invention designs a low-complexity VLSI architecture according to the path division multiple access mode, which can efficiently realize the uplink channel prediction. In the present invention, a channel prediction architecture suitable for path division multiple access is designed for the first time. Path division multiple access considers joint time delay and angle domain for channel prediction, and accurately predicts parameters such as direction of arrival, path gain, etc., to perform channel reconstruction. structure. However, there are complex algorithms such as large-scale matrix inverse Fourier transform and high-precision two-dimensional feature search, which are difficult to deploy in practical applications. Aiming at the problem that high-complexity operations in the original algorithm are difficult to implement on hardware at high speed, the present invention proposes a VLSI hardware architecture corresponding to path division multiple access for the first time, so as to facilitate the hardware implementation of the algorithm on FPGA or ASIC. In the original channel prediction and tracking algorithms, most of them use direct prediction and tracking of the channel matrix. In a large-scale multiple-input multiple-output antenna system, the channel matrix will expand rapidly with the increase of the number of antennas, which leads to The original algorithm is difficult to implement efficiently. The present invention only needs to calculate the path parameters of each path based on path division multiple access prediction, and then reconstruct the channel matrix corresponding to this path, which greatly reduces the amount of calculation. At the same time, in the VLSI architecture mainly designed by the present invention, the modules are pipelined and the two-dimensional search algorithm in the original algorithm is replaced by successive binary feedback search, so that the parameters are searched more accurately, and the channel prediction efficiency is improved. It is greatly improved, and the corresponding hardware complexity does not increase significantly due to the continuous multiplexing of modules for single-channel feedback, which reduces the hardware cost.

Claims (10)

1. A method for predicting uplink channel based on path division multiple access (TDMA), comprising:
(1) pilot phase initial prediction phase: the preprocessing module processes the received signal Yp of the p-th user to obtain an initial estimation channel matrix
Figure FDA0003350464250000011
The initial estimated channel matrix
Figure FDA0003350464250000012
Respectively inputting the data into a pre-searching module and a channel characteristic searching module; the pre-search module estimates a channel matrix based on an initial estimate
Figure FDA0003350464250000013
Initial search is carried out to obtain the channel characteristics of the p-th user
Figure FDA0003350464250000014
Characterizing channels
Figure FDA0003350464250000015
Respectively inputting a channel characteristic searching module and a channel characteristic grouping module, wherein l is the l-th path of a user p;
(2) channel characteristic prediction grouping and channel reconstruction stage: the channel characteristic searching module is used for searching the channel characteristics
Figure FDA0003350464250000016
Carrying out accurate prediction to obtain the two-dimensional channel characteristics of the p-th user and outputting the two-dimensional channel characteristics to a reconstruction module; the channel characteristic grouping module is used for grouping channel characteristics and storing the grouped channel characteristics into a grouping register, and the grouped channel characteristics are used for guiding the uplink channel prediction module to predict; the reconstruction module reconstructs the two-dimensional channel characteristics to obtain a channel base vector group w, wherein the channel base vector group w comprises channel base vectors of Lp paths of the p-th user
Figure FDA0003350464250000017
The channel base vector group w is output to an uplink channel prediction module;
(3) and an uplink channel prediction stage: and the uplink channel prediction module outputs an uplink channel prediction matrix according to the received signal Yp and the channel base vector.
2. The path division multiple access-based uplink channel prediction method of claim 1, wherein: in the step (1), the preprocessing module obtains a base vector based on an LS channel estimation mode
Figure FDA0003350464250000018
The base vector
Figure FDA0003350464250000019
Performing serial-to-parallel conversion to obtain an initial estimation channel matrix
Figure FDA00033504642500000110
3. The path division multiple access-based uplink channel prediction method of claim 1, wherein: in the step (1), the pre-search module initially estimates the channel matrix
Figure FDA00033504642500000111
Performing IFFT operation to obtain a channel prediction matrix in an angle time delay transformation domain:
Figure FDA00033504642500000112
wherein,
Figure FDA00033504642500000113
is the transposed conjugate of an M x M dimensional fourier transform matrix,
Figure FDA00033504642500000114
conjugate of N x N dimensional Fourier transform matrix; for transform domain matrix
Figure FDA00033504642500000115
Modulus is taken, and the maximum two-dimensional coordinates of each path of the user p are compared to be used as channel characteristics obtained by initial search
Figure FDA00033504642500000116
4. The path division multiple access-based uplink channel prediction method of claim 1, wherein: in the step (2), the channel characteristic searching module adopts a successive binary iteration method to search the channel characteristics
Figure FDA0003350464250000021
Neighborhood search satisfaction
Figure FDA0003350464250000022
Outputs a two-dimensional channel profile, wherein,
Figure FDA0003350464250000023
is the transposed conjugate of an M x M dimensional fourier transform matrix,
Figure FDA0003350464250000024
is a conjugate of an N x N dimensional Fourier transform matrix, ΨM，ΨNThe angle domain rotation factor and the time delay domain selection factor are respectively, and theta is a phase shift matrix.
5. The path division multiple access-based uplink channel prediction method of claim 1, wherein: in the step (3), the uplink channel prediction module calculates a channel gain of each path of the user p
Figure FDA0003350464250000025
Wherein M is the number of transmitting antennas, N is the number of subcarriers of OFDM, MN is the number of subcarriers of OFDM of M antennas, EpLimiting power for the p-th user, vec (Y)p) In order to convert the received signal into a vector input,
Figure FDA0003350464250000026
taking the maximum two-dimensional coordinates of each path of the user p as channel characteristics obtained by initial search; according to the channel gain obtained by calculation, the channel vector of the user p is obtained by multiplying and summing the channel gain with the corresponding channel base vector group w, and then the channel vector is adjusted to a matrix form to obtain the final uplink channel prediction matrix of the p-th user
Figure FDA0003350464250000027
Wherein shape is a matrix dimension transformation function,
Figure FDA0003350464250000028
for the path set corresponding to the p-th user, for the channel base vector of each path:
Figure FDA0003350464250000029
wherein, thetap，lPhase-shift matrix for the l-th path of user p, bp，lIs a frequency domain direction vector, ap，lIs an angular direction matrix.
6. An uplink channel prediction system based on path division multiple access, characterized by: the device comprises a preprocessing module, a pre-searching module, a channel characteristic grouping module, a reconstruction module and an uplink channel prediction module;
the preprocessing module is used for processing the received signal Yp of the p-th user to obtain an initial estimation channel matrix
Figure FDA0003350464250000031
Respectively inputting the data into a pre-searching module and a channel characteristic searching module;
the pre-search module is used for estimating a channel matrix according to the initial estimation
Figure FDA0003350464250000032
Initial search is carried out to obtain the channel characteristics of the p-th user
Figure FDA0003350464250000033
Characterizing channels
Figure FDA0003350464250000034
Respectively inputting a channel characteristic searching module and a channel characteristic grouping module, wherein l is the l-th path of a user p;
the channel characteristic searching module is used for searching the channel characteristics
Figure FDA0003350464250000035
Carrying out accurate prediction to obtain the two-dimensional channel characteristics of the p-th user and outputting the two-dimensional channel characteristics to a reconstruction module;
the channel characteristic grouping module is used for grouping channel characteristics and storing the grouped channel characteristics into a grouping register, and the grouped channel characteristics are used for guiding the uplink channel prediction module to predict;
the reconstruction module is used for reconstructing the two-dimensional channel characteristics to obtain a channel base vector group w, wherein the channel base vector group w comprises channel base vectors of Lp paths of the p-th user
Figure FDA0003350464250000036
Outputting the channel base vector group w to an uplink channel prediction module;
the uplink channel prediction module is configured to output an uplink channel prediction matrix according to the received signal Yp and the channel base vector.
7. The path division multiple access based uplink channel prediction system of claim 6, wherein: the preprocessing module obtains a base vector based on an LS channel estimation mode
Figure FDA0003350464250000037
The base vector
Figure FDA0003350464250000038
Performing serial-to-parallel conversion to obtain an initial estimation channel matrix
Figure FDA0003350464250000039
8. The path division multiple access based uplink channel prediction system of claim 6, wherein: the pre-search module is used for initially estimating a channel matrix
Figure FDA00033504642500000310
Performing IFFT operation to obtain a channel prediction matrix in an angle time delay transformation domain:
Figure FDA00033504642500000311
wherein,
Figure FDA00033504642500000312
is the transposed conjugate of an M x M dimensional fourier transform matrix,
Figure FDA00033504642500000313
conjugate of N x N dimensional Fourier transform matrix; for transform domain matrix
Figure FDA00033504642500000314
Modulus is obtained, and the maximum two-dimensional coordinates of each path of the user p are compared and are used as the coordinates obtained by initial searchChannel characteristics
Figure FDA00033504642500000315
9. The path division multiple access based uplink channel prediction system of claim 6, wherein: the channel characteristic searching module adopts a successive dichotomy iteration method to search the channel characteristics
Figure FDA00033504642500000316
Neighborhood search satisfaction
Figure FDA0003350464250000041
Outputs a two-dimensional channel profile, wherein,
Figure FDA0003350464250000042
is the transposed conjugate of an M x M dimensional fourier transform matrix,
Figure FDA0003350464250000043
is a conjugate of an N x N dimensional Fourier transform matrix, ΨM，ΨNThe angle domain rotation factor and the time delay domain selection factor are respectively, and theta is a phase shift matrix.
10. The path division multiple access based uplink channel prediction system of claim 6, wherein: the uplink channel prediction module calculates the channel gain of each path of the user p
Figure FDA0003350464250000044
Wherein M is the number of transmitting antennas, N is the number of subcarriers of OFDM, MN is the number of subcarriers of OFDM of M antennas, EpLimiting power for the p-th user, vec (Y)p) In order to convert the received signal into a vector input,
Figure FDA0003350464250000045
taking the maximum two-dimensional coordinates of each path of the user p as channel characteristics obtained by initial search; according to the channel gain obtained by calculation, the channel vector of the user p is obtained by multiplying and summing the channel gain with the corresponding channel base vector group w, and then the channel vector is adjusted to a matrix form to obtain the final uplink channel prediction matrix of the p-th user
Figure FDA0003350464250000046
Wherein shape is a matrix dimension transformation function,
Figure FDA0003350464250000047
for the path set corresponding to the p-th user, for the channel base vector of each path:
Figure FDA0003350464250000048
wherein, thetap，lPhase-shift matrix for the l-th path of user p, bp，lIs a frequency domain direction vector, ap，lIs an angular direction matrix.
CN201911021624.0A 2019-10-25 2019-10-25 Uplink channel prediction method and prediction system based on path division multiple access Active CN110740005B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911021624.0A CN110740005B (en) 2019-10-25 2019-10-25 Uplink channel prediction method and prediction system based on path division multiple access

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911021624.0A CN110740005B (en) 2019-10-25 2019-10-25 Uplink channel prediction method and prediction system based on path division multiple access

Publications (2)

Publication Number Publication Date
CN110740005A CN110740005A (en) 2020-01-31
CN110740005B true CN110740005B (en) 2021-12-24

Family

ID=69271275

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911021624.0A Active CN110740005B (en) 2019-10-25 2019-10-25 Uplink channel prediction method and prediction system based on path division multiple access

Country Status (1)

Country Link
CN (1) CN110740005B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111427537B (en) * 2020-03-17 2023-06-30 云南大学 Pulse array parallel ordering method and device based on FPGA
CN112019461A (en) * 2020-07-24 2020-12-01 深圳信息职业技术学院 Channel prediction method, wireless communication system, and storage device
CN113541747B (en) * 2021-07-01 2022-07-08 网络通信与安全紫金山实验室 Massive MIMO detection method, device and storage medium
CN116527167A (en) * 2022-01-24 2023-08-01 华为技术有限公司 Channel characteristic acquisition method and related device

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104798062A (en) * 2012-11-21 2015-07-22 相干逻辑公司 Processing system with interspersed processors with multi-layer interconnect
WO2017000752A1 (en) * 2015-07-01 2017-01-05 东南大学 Downlink training sequence design method of fdd large-scale mimo system
CN108736935A (en) * 2018-05-18 2018-11-02 东南大学 A kind of general down and out options method for extensive mimo system signal detection
CN108964726A (en) * 2018-09-03 2018-12-07 东南大学 A kind of extensive MIMO uplink transmission channels estimation method of low complex degree

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104798062A (en) * 2012-11-21 2015-07-22 相干逻辑公司 Processing system with interspersed processors with multi-layer interconnect
WO2017000752A1 (en) * 2015-07-01 2017-01-05 东南大学 Downlink training sequence design method of fdd large-scale mimo system
CN108736935A (en) * 2018-05-18 2018-11-02 东南大学 A kind of general down and out options method for extensive mimo system signal detection
CN108964726A (en) * 2018-09-03 2018-12-07 东南大学 A kind of extensive MIMO uplink transmission channels estimation method of low complex degree

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
MIMO-OFDM中的高阶Viterbi译码器的研究;孙元华等;《电子技术》;20091025(第10期);全文 *

Also Published As

Publication number Publication date
CN110740005A (en) 2020-01-31

Similar Documents

Publication Publication Date Title
CN110740005B (en) Uplink channel prediction method and prediction system based on path division multiple access
CN101674160B (en) Signal detection method and device for multiple-input-multiple-output wireless communication system
US20230113600A1 (en) Mixing coefficient data for processing mode selection
Yang et al. Channel estimation for practical IRS-assisted OFDM systems
EP3058689B1 (en) An improved lattice-reduction-aided k-best algorithm for low complexity and high performance communications
CN115065432B (en) Method and system for modeling and obtaining relevant channel information of sky-wave large-scale MIMO triple-beam base channel
CN107005504A (en) Method and device for the data in the tree searching and detecting cordless communication network by reducing complexity
JP2009100478A (en) Wireless communication device
Dong et al. Blind demixing for low-latency communication
TWI436605B (en) Lattice reduction architecture and method and detection system thereof
CN108964725A (en) The sparse estimation method of channel parameter in the extensive MIMO network of time-varying
Wang et al. Underwater acoustic sparse channel estimation based on DW-SACoSaMP reconstruction algorithm
Haq et al. Deep neural network augmented wireless channel estimation for preamble-based ofdm phy on zynq system on chip
Jing et al. Transceiver beamforming for over-the-air computation in massive MIMO systems
Payami et al. Deep learning-based channel estimation for massive MIMO-OTFS communication systems
CN103532609A (en) MIMO (Multiple Input Multiple Output) signal detection method and detector on basis of K-Best algorithm
Tang et al. Channel Estimation for XL-MIMO Systems with Decentralized Baseband Processing: Integrating Local Reconstruction with Global Refinement
CN109067674A (en) A kind of channel estimation methods based on selected slant stack pilot tone
CN107770104A (en) A kind of channel estimation pilot optimization method and device based on compressed sensing
Yang et al. An improved least squares (LS) channel estimation method based on CNN for OFDM systems.
CN105553614B (en) Integrating device based on signal detection algorithm
Amiri et al. A high throughput configurable SDR detector for multi-user MIMO wireless systems
Tolba et al. A meta learner autoencoder for channel state information feedback in massive MIMO systems
CN108471333A (en) A Channel Estimation Method in Underwater Communication
CN107104714B (en) A MIMO Detection Method Without QR Decomposition

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
CP03 Change of name, title or address

Address after: No. 9 Mozhou East Road, Nanjing City, Jiangsu Province, 211111

Patentee after: Zijinshan Laboratory

Country or region after: China

Address before: No. 9 Mozhou East Road, Jiangning Economic Development Zone, Jiangning District, Nanjing City, Jiangsu Province

Patentee before: Purple Mountain Laboratories

Country or region before: China

CP03 Change of name, title or address