WO2023130726A1 - 一种基于互质阵列的非对称大规模mimo信道估计方法 - Google Patents

一种基于互质阵列的非对称大规模mimo信道估计方法 Download PDF

Info

Publication number
WO2023130726A1
WO2023130726A1 PCT/CN2022/112628 CN2022112628W WO2023130726A1 WO 2023130726 A1 WO2023130726 A1 WO 2023130726A1 CN 2022112628 W CN2022112628 W CN 2022112628W WO 2023130726 A1 WO2023130726 A1 WO 2023130726A1
Authority
WO
WIPO (PCT)
Prior art keywords
array
uplink
antennas
frequency
matrix
Prior art date
Application number
PCT/CN2022/112628
Other languages
English (en)
French (fr)
Inventor
张军
陆佳程
蔡曙
王海荣
吕文俊
朱洪波
Original Assignee
南京邮电大学
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 南京邮电大学 filed Critical 南京邮电大学
Publication of WO2023130726A1 publication Critical patent/WO2023130726A1/zh

Links

Images

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04BTRANSMISSION
    • H04B7/00Radio transmission systems, i.e. using radiation field
    • H04B7/02Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas
    • H04B7/04Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas
    • H04B7/0413MIMO systems
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00Baseband systems
    • H04L25/02Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202Channel estimation
    • H04L25/0204Channel estimation of multiple channels
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00Baseband systems
    • H04L25/02Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202Channel estimation
    • H04L25/022Channel estimation of frequency response
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00Baseband systems
    • H04L25/02Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202Channel estimation
    • H04L25/024Channel estimation channel estimation algorithms
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02DCLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00Reducing energy consumption in communication networks
    • Y02D30/70Reducing energy consumption in communication networks in wireless communication networks

Definitions

  • the invention relates to an asymmetric massive MIMO (Multiple Input Multiple Output, MIMO) channel estimation method based on a coprime array, belonging to the technical field of wireless communication.
  • MIMO Multiple Input Multiple Output
  • the technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide an asymmetric large-scale channel estimation method based on a coprime array in broadband multi-frequency point communication, which avoids the problems caused by the different numbers of antennas in the uplink and downlink arrays.
  • the difference in angular resolution capability effectively ensures the reliability and accuracy of the reconstructed downlink channel information.
  • the present invention provides an asymmetric massive MIMO channel estimation method based on a coprime array, comprising the following steps:
  • the system includes a base station equipped with ultra-large-scale antennas and K users with single antennas; the number of base station antennas is M, and all antennas of the base station have transmission radio frequency chains Only N (N ⁇ M, K ⁇ N) receiving radio frequency links can be connected to N antennas for uplink receiving signals, and users and base stations use Q frequency points for communication.
  • N N ⁇ M, K ⁇ N
  • the base station receives and estimates the uplink channel information of some antennas, transforms it into the frequency domain and performs screening and rearrangement to construct a virtual linear uniform array;
  • the present invention designs a partial-antenna channel estimation method based on a coprime array under an asymmetric transceiver architecture, using partial antennas to effectively and reliably restore the channel information of the complete array to overcome the rate loss caused by the difference between the uplink and downlink channels.
  • the communication system proposed by the present invention first determine the number of antennas enabled in the uplink and downlink and the location of the specific activated antenna in the uplink according to the nested coprime array; then the pilot signals received on the activated part of the uplink antenna Fourier transform, which separates the frequency domain channel information of each frequency point in the frequency domain, and performs autocorrelation processing on the frequency domain signals of each frequency point, extracts virtual array element signals, screens, and rearranges them to form a virtual linear array Uniform array; then stretch the virtual array information on all frequency points into a vector, construct an observation matrix, and obtain the solution vector according to the iterative formula of the ADMM (Alternating Direction Method of Multipliers, ADMM) optimization framework, sort the solution vector and select the path , get the direction of arrival, reconstruct the uplink array manifold according to the direction of arrival, and obtain the path gain by least squares; finally, reconstruct the complete downlink array according to the previously estimated direction of arrival and path gain.
  • ADMM Alternating Direction Method of Multipliers
  • the present invention uses a coprime array to avoid asymmetry in channel information and resolution caused by different numbers of antennas in the uplink and downlink arrays, and the channel information estimated in the uplink can be directly used for downlink precoding in time-division duplex mode, effectively ensuring The communication rate of the asymmetric transceiver architecture is improved.
  • the downlink array uses all antennas to send signals to maximize the communication rate; the uplink only uses some of the antennas activated according to the coprime array arrangement to receive signals, so as to reduce the pressure on uplink signal reception and decoding.
  • the specific operation is as follows:
  • the receiving and transmitting radio frequency chains of M antennas are designed separately, and there are M transmitting radio frequency links and N receiving radio frequency chains.
  • S102 Select the uplink receiving antenna according to the coprime array, and connect the receiving radio frequency chain.
  • the interval d between the antennas is ⁇ min /2, where ⁇ min is the wavelength corresponding to the subcarrier with the highest frequency in each frequency point.
  • the interval d between the antennas is ⁇ min /2, where ⁇ min is the wavelength corresponding to the subcarrier with the highest frequency in each frequency point.
  • the base station receives and estimates the uplink channel information of some antennas, transforms it into the frequency domain and performs screening and rearrangement, and constructs a virtual linear uniform array, which specifically includes the following steps:
  • the column vector y i whose length is M is the frequency domain signal at the i-th frequency point
  • [ ⁇ ] T means to take the transpose
  • the specific selection method is as follows:
  • the antenna numbers are recorded as 0, 1, 2,..., M-1, and the antenna array numbers activated in the uplink are recorded as a set of length N ⁇ p 1 ,p 2 ,...,p N ⁇ , arrange this set into an N ⁇ N matrix T c by column:
  • Each element in the autocorrelation matrix R q is related to One-to-one correspondence among elements, that is, the element [R q ] i, j in row i, column j in R q is the element at position [R tab ] i, j in the virtual array.
  • step S3 the group least absolute shrinkage selection operator based on compressed sensing is used to construct an estimation problem, and the problem is solved under the ADMM optimization framework, thereby estimating the direction of arrival, which specifically includes the following steps:
  • a i,1 is the steering vector of the first frequency point on the i grid point
  • a i,Q is the steering vector of the Qth frequency point on the i grid point
  • a i,q is the steering vector of the qth frequency point on the i-th grid point
  • the length of e 1 is M
  • j is the imaginary number unit
  • i is the order of grid points
  • i is the order of grid points
  • ⁇ q is the wavelength corresponding to the q-th frequency point
  • ADMM Alternating Direction Method of Multipliers
  • the column vector of length Q(w+1) is the target variable to be solved, representing the energy of each frequency point on each estimated grid point
  • x 1 is The subvector composed of (1-1)Q+1 to 1Q elements
  • x w+1 is The subvector composed of (w+1-1)Q+1 to w+1Qth elements
  • x i is The subvector composed of (i-1)Q+1th to iQth elements
  • z 1 is The subvector composed of (1-1)Q+1 to 1Q elements
  • z w+1 is The subvector composed of (w+1-1)Q+1 to w+1Qth elements
  • z i is A subvector composed of (i-1)Q+1 to iQth elements
  • auxiliary variables ⁇ t is the penalty coefficient
  • 2 means to take the 2-norm of the target vector.
  • the k+1 iteration formula of the above problem is:
  • I is the identity matrix, is the auxiliary variable
  • u 1 is The subvector composed of (1-1)Q+1 to 1Q elements
  • u w+1 is The subvector composed of (w+1-1)Q+1 to w+1Qth elements
  • u i is The subvector composed of (i-1)Q+1 to iQth elements of , u i is defined like x i , the superscript ( ) (k) indicates the variable value of the kth iteration
  • is the iteration step size
  • Indicates to take the real part of the complex number is the vector produced by the k+1th iteration
  • the subvector composed of (i-1)Q+1th to iQth elements of , the convergence condition is:
  • is the convergence threshold, which is a small constant greater than 0;
  • step S303 calculates the energy distribution vector on each estimated grid point (spatial direction) And for the vector
  • the elements in are sorted, and the sorted for select middle front elements such that:
  • is the path restoration threshold , and it satisfies 0 ⁇ 1,
  • is the path restoration threshold , and it satisfies 0 ⁇ 1,
  • is the path restoration threshold , and it satisfies 0 ⁇ 1,
  • the corresponding subscripts in form a set Indicates that the first selected element in the above formula is in in the original order, Indicates that the second selected element in the above formula is in in the original order, Indicates that the first selected elements in in the original order, Indicates that the lth selected element in the above formula is in The original order in , in the space
  • the direction of arrival of a path is written in vector form as:
  • step S4 the uplink partial array manifold matrix is reconstructed according to the estimated direction of arrival, and the path gain is estimated according to the observed instantaneous information in the frequency domain, which specifically includes the following steps:
  • p 1 , p 2 , p n , p N are the order corresponding to the selected antenna in the uplink array, p n ⁇ 0,1,2,...,M-1 ⁇ , i ⁇ L;
  • step S4 according to the path gain estimated in S402 and the direction of arrival estimated in S303, the complete uplink channel is reconstructed and symmetric to the downlink channel, and the specific operation is as follows:
  • the present invention proposes a downlink channel estimation method based on a coprime array under asymmetric massive MIMO (Multi-Input Multi-Output, MIMO).
  • MIMO Multi-Input Multi-Output
  • MIMO Multi-Input Multi-Output
  • the invention solves the problem that the recovered uplink channel cannot be directly used for downlink channel precoding by utilizing the feature of high angular resolution of the coprime array.
  • the present invention effectively utilizes part of the antennas to estimate the complete downlink channel in the asymmetric architecture, and has obvious improvements in reducing uplink receiving pressure and improving channel recovery accuracy.
  • the present invention adopts the above technical scheme and has the following technical effects:
  • the present invention makes full use of the sparse multipath characteristics of high-frequency signals in space under the asymmetric transceiver architecture, transforms the original channel estimation problem into a parameter estimation problem, and utilizes the group minimum absolute contraction selection operator in ADMM (Alternating Direction Method of Multipliers, ADMM) optimization framework to quickly and accurately estimate the direction of arrival, while achieving the overall channel estimation accuracy, it also ensures the sparse characteristics of the channel model and prevents errors caused by subsequent overfitting.
  • ADMM Alternating Direction Method of Multipliers
  • the present invention introduces a coprime array into the uplink array of an asymmetric structure, which effectively eliminates the difference in the resolution capability of the uplink and downlink arrays, ensures the accuracy of channel reconstruction and the effectiveness of channel information available for downlink precoding, thereby improving system transfer rate.
  • Fig. 1 is a flowchart of the present invention.
  • Fig. 2 is a schematic diagram of selection of uplink receiving antennas in the present invention.
  • This embodiment proposes an asymmetric massive MIMO channel estimation method based on a coprime array, as shown in FIG. 1 , including the following steps:
  • the system includes a base station equipped with ultra-large-scale antennas (the number of antennas is M) and K users with single antennas. All antennas of the base station have transmitting radio frequency links, and only N (N ⁇ M, K ⁇ N) receiving radio frequency links can be connected to N antennas for uplink receiving signals, and users communicate with the base station using Q frequency points.
  • the interval d between the antennas is ⁇ min /2, where ⁇ min is the wavelength corresponding to the subcarrier with the highest frequency in each frequency point.
  • the base station receives and estimates the uplink channel information of some antennas, transforms it into the frequency domain and performs screening and rearrangement to construct a virtual linear uniform array. Include the following steps:
  • the antenna numbers are recorded as 0, 1, 2,..., M-1, and the antenna array numbers activated in the uplink are recorded as a set of length N ⁇ p 1 ,p 2 ,...,p N ⁇ , arrange this set into an N ⁇ N matrix T c by column:
  • Each element in the autocorrelation matrix R q is related to One-to-one correspondence among elements, that is, the element [R q ] i, j in row i, column j in R q is the element at position [R tab ] i, j in the virtual array.
  • the column vector y i whose length is M is the frequency domain signal at the i-th frequency point
  • [ ⁇ ] T means to take the transpose
  • a i,1 is the steering vector of the first frequency point on the i grid point
  • a i,Q is the steering vector of the Qth frequency point on the i grid point
  • a i,q is the steering vector of the qth frequency point on the i-th grid point
  • the length of e 1 is M
  • j is the imaginary number unit
  • i is the order of grid points
  • i is the order of grid points
  • ⁇ q is the wavelength corresponding to the q-th frequency point
  • ADMM Alternating Direction Method of Multipliers
  • the column vector of length Q(w+1) is the target variable to be solved, representing the energy of each frequency point on each estimated grid point
  • x 1 is The subvector composed of (1-1)Q+1 to 1Q elements
  • x w+1 is The subvector composed of (w+1-1)Q+1 to w+1Qth elements
  • x i is The subvector composed of (i-1)Q+1th to iQth elements
  • z 1 is The subvector composed of (1-1)Q+1 to 1Q elements
  • z w+1 is The subvector composed of (w+1-1)Q+1 to w+1Qth elements
  • z i is A subvector composed of (i-1)Q+1 to iQth elements
  • auxiliary variables ⁇ t is the penalty coefficient
  • 2 means to take the 2-norm of the target vector.
  • the k+1 iteration formula of the above problem is:
  • I is the identity matrix, is the auxiliary variable
  • u 1 is The subvector composed of (1-1)Q+1 to 1Q elements
  • u w+1 is The subvector composed of (w+1-1)Q+1 to w+1Qth elements
  • u i is The subvector composed of (i-1)Q+1 to iQth elements of , u i is defined like x i , the superscript ( ) (k) indicates the variable value of the kth iteration
  • is the iteration step size
  • Indicates to take the real part of the complex number is the vector produced by the k+1th iteration
  • the subvector composed of (i-1)Q+1th to iQth elements of , the convergence condition is:
  • is the convergence threshold, which is a small constant greater than 0;
  • step S303 calculates the energy distribution vector on each estimated grid point (spatial direction) And for the vector
  • the elements in are sorted, and the sorted for select middle front elements such that:
  • is the path restoration threshold , and it satisfies 0 ⁇ 1,
  • is the path restoration threshold , and it satisfies 0 ⁇ 1,
  • is the path restoration threshold , and it satisfies 0 ⁇ 1,
  • the corresponding subscripts in form a set Indicates that the first selected element in the above formula is in in the original order, Indicates that the second selected element in the above formula is in in the original order, Indicates that the first selected elements in in the original order, Indicates that the lth selected element in the above formula is in The original order in , in the space
  • the direction of arrival of a path is written in vector form as:
  • p 1 , p 2 , p n , p N are the order corresponding to the selected antenna in the uplink array, p n ⁇ 0,1,2,...,M-1 ⁇ , i ⁇ L;
  • step S4 according to the path gain estimated in S402 and the direction of arrival estimated in S303, the complete uplink channel is reconstructed and symmetric to the downlink channel, and the specific operation is as follows:

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Power Engineering (AREA)
  • Radio Transmission System (AREA)

Abstract

本发明在非对称大规模MIMO下提出了一种基于互质阵列的下行信道估计方法。首先,构建基于互质阵列的上下行非对称收发系统模型,并关注阵列宽带信号带来的频域方向偏移;其次,进行上行接收估计出上行信道,并由此恢复路径数,到达角,路径增益等信道参数;最后,利用上行信道恢复出的信道参数重建下行信道;对于宽带信号还可以利用不同频点上的群稀疏特性提高信道估计精度。本发明利用互质阵列角度分辨率高的特点,解决了恢复出的上行信道无法直接用于下行信道预编码的问题。本发明在非对称架构中有效地利用部分天线估计完整的下行信道,在降低上行链路接收压力和提高信道恢复准确性方面具有明显改善。

Description

一种基于互质阵列的非对称大规模MIMO信道估计方法 技术领域
本发明涉及一种基于互质阵列的非对称大规模MIMO(Multiple Input Multiple Output,MIMO)信道估计方法,属于无线通信技术领域。
背景技术
据了解,在广泛部署第五代移动通信系统中,大规模MIMO技术得到了进一步应用与拓展,显示了MIMO阵列在速率,频谱效率以及可靠性方面的优越性能。为满足不断增长的速率和可靠性需求,人们希望进一步扩大MIMO阵列的规模,但随之而来的硬件成本,数据处理负担以及功耗要求限制了全数字大规模MIMO技术的进一步发展。尽管已经有许多文献提出了各种替代方案,如混合波束赋形,低精度数模/模数转换器,天线选择等技术,但这些技术都不可避免地牺牲了部分通信性能,这在对速率要求越来越高的多样化网络业务中是令人难以接收的。为了在保证用户体验的同时,尽可能降低基站的成本,近期有学者提出了一种新型的全数字非对称收发架构。这种架构的非对称性体现在上下行收发过程的差异,即通过解耦发送与接收的射频链路,允许上行仅用部分天线进行接收,下行全部天线都可用于发送信号。这种设计理念的初衷源于上下行通信需求的不同。下行通信中速率是影响用户体验的最直观指标,然而上行的数据吞吐量相比下行要小很多。为此上行仅需部分天线即可满足基站信息搜集的需求。这种设计方式由于上下行阵列天线数目不同容易带来角度分辨能力的差异,会导致下行信道信息不可靠、不准确。
发明内容
本发明所要解决的技术问题是,克服现有技术的不足而提供一种宽带多频点通信中的基于互质阵列的非对称大规模信道估计方法,避免了上下行阵列由于天线数目不同带来的角度分辨能力差异,有效地保证了重建出的下行信道信息可靠性与准确性。
本发明提供一种基于互质阵列的非对称大规模MIMO信道估计方法,包括以下步骤:
S1、构建基于互质阵列的上下行非对称大规模MIMO系统,该系统包括一个配有超大规模天线的基站和K个单天线的用户;基站天线数为M,基站所有天线都有发送射频链路,只有N(N<<M,K≤N)根接收射频链路可连接到N根天线用于上行接收信号,用户与基站用Q个频点进行通信。数目关系为:M=mn+1,N=m+n-1,其中m<n,且m,n互质;按照互质阵列的方式选取上行接收天线,连接接收射频链;
S2、基站方接收并估计得到部分天线的上行信道信息,将其变换到频域并进行筛选重排后,构造虚拟线性均匀阵列;
S3、利用空间稀疏特性构造基于压缩感知的群最小绝对收缩选择算子,估计来波方向;
S4、根据估计出的来波方向重构部分阵列流形矩阵,并根据后续观测到的频域瞬时信息估计路径增益;
S5、依据估计出的路径增益,来波方向重构完整的上行信道,根据互易性将其对称至下行信道。
本发明在非对称收发架构下设计一种基于互质阵列的部分天线信道估计方法,使用部分天线来有效且可靠地恢复完整阵列信道信息,用于克服上下行信道差异性带来的速率损失。本发明在提出的通信系统中,首先按照嵌套互质阵列的方式确定上下行各自启用的 天线数目以及上行具体激活的天线所在位置;然后将被激活的部分上行天线上接收的导频信号进行傅里叶变换,在频域上将各个频点的频域信道信息分离开来,并对各个频点的频域信号做自相关处理,提取虚拟阵元信号、筛选、重排后形成虚拟线性均匀阵列;接着将所有频点上的虚拟阵列信息拉伸成向量,构造观测矩阵,按照ADMM(Alternating Direction Method of Multipliers,ADMM)优化框架的迭代公式求得解向量,对解向量排序选取路径后,得到来波方向,依据来波方向重构上行阵列流形,用最小二乘求得路径增益;最后依据先前估计出来的来波方向和路径增益重建完整的下行阵列。本发明利用互质阵列避免了上下行阵列由于天线数不同导致的信道信息和分辨率上的非对称性,上行估计出的信道信息在时分双工模式下可以直接用于下行预编码,有效保证了非对称收发架构的通信速率。
本发明进一步优化的技术方案如下所示:
所述步骤S1中,下行阵列使用所有天线发送信号,以最大化通信速率;上行仅使用部分按照互质阵列排布方式激活的天线接收信号,以减少上行信号接收与解码压力。具体操作如下:
S101、M根天线的接收和发送射频链分开设计电路,有M条发送射频链路和N条接收射频链,数目关系为:M=mn+1,N=m+n-1,其中m<n,且m,n互质。
S102、根据互质阵列的方式选取上行接收天线,连接接收射频链,具体选取方式为:选中起始的连续n根,第2n+1根,第3n+1根,…,直至第mn+1根,共计N=m+n-1根。天线之间间隔d为λ min/2,其中λ min是各个频点中频率最大的子载波对应的波长。
所述步骤S1中,选取上行接收天线的具体方式为:选中起始的 连续n根,第2n+1根,第3n+1根,…,直至第mn+1根,共计N=m+n-1根天线(m,n分别是互质的两个数,用于设计非对称阵列上下行天线的个数)。天线之间的间隔d为λ min/2,其中λ min是各个频点中频率最大的子载波所对应的波长。
所述步骤S2中,基站方接收并估计得到部分天线的上行信道信息,将其变换到频域并进行筛选重排后,构造虚拟线性均匀阵列,具体包括以下步骤:
S201、在连续的P个符号时间内,对上行阵列收到的P组Q个离散时域信号分别进行Q点FFT(Fast Fourier Transform,FFT)变换,得到P组Q个频点上的频域信号,记N×1的列向量x p,q为第p组信号中第q个频点对应的频域信号;
S202、对每个频点的P个频域信号进行自相关处理,得到Q个N×N的自相关矩阵:
Figure PCTCN2022112628-appb-000001
在R q中将重复位置的阵元进行分类(上三角对应负滞后部分,故舍去)、平均后,按照顺序将其排列成长为M×1的列向量y q,[·] H表示共轭转置,[·] T表示转置;
S203、将所有频点上的虚拟阵列信号排成一个QM的列向量:
Figure PCTCN2022112628-appb-000002
其中,长为M的列向量y i是第i个频点上的频域信号,[·] T表示取转置。
所述步骤S202中,具体选取方式如下:
将完整的阵列从一端开始,天线序号记为0,1,2,…,M-1,其中上行激活的天线阵列序号记为一个长为N的集合{p 1,p 2,…,p N},将此集 合按列排出一个N×N的矩阵T c
Figure PCTCN2022112628-appb-000003
自相关矩阵R q中各元素与
Figure PCTCN2022112628-appb-000004
中元素一一对应,即R q中第i行第j列元素[R q] i,j即为虚拟阵列中第[R tab] i,j个位置上的元素。
所述步骤S3中,利用基于压缩感知的群最小绝对收缩选择算子构造估计问题,在ADMM优化框架下求解该问题,从而估计来波方向,具体包括以下步骤:
S301、在预先设定的入射角区间[θ lr]上构造QM×Q(w+1)的观测矩阵
Figure PCTCN2022112628-appb-000005
θ lr分别是左角度边界和右角度边界,并满足0≤θ lr≤π,这两个边界值可以通过DFT(Discrete Fourier Transform,DFT)等角度域检测手段获得;若不进行检测则θ l=0,θ r=π。w是在估计区间[θ lr]上划分的格点个数,w越大估计结果越精确,但复杂度也提示,QM×Q的子矩阵A i是第i个格点上的观测矩阵,
Figure PCTCN2022112628-appb-000006
是用于估计噪声的矩阵,各子观测矩阵按如下方式生成:
Figure PCTCN2022112628-appb-000007
其中省略号位置全部为0,a i,1是第1个频点在第i个格点上的导向矢量,a i,Q是第Q个频点在第i个格点上的导向矢量,a i,q是第q个频点在第i个格点上的导向矢量,且:
Figure PCTCN2022112628-appb-000008
e 1=[1,0,…,0] T
其中,e 1长度为M,j是虚数单位,i是格点次序,
Figure PCTCN2022112628-appb-000009
是格点间距大小,
Figure PCTCN2022112628-appb-000010
表示第i个格点所代表的角度,λ q是第q个频点对应的波长;
S302、依据压缩感知理论在ADMM(Alternating Direction Method of Multipliers,ADMM)框架下写出L21范数约束的线性回归问题:
Figure PCTCN2022112628-appb-000011
Figure PCTCN2022112628-appb-000012
其中,长度为Q(w+1)的列向量
Figure PCTCN2022112628-appb-000013
是待求解的目标变量,代表着各个频点在各个估计格点上的能量大小,x 1
Figure PCTCN2022112628-appb-000014
的第(1-1)Q+1到第1Q个元素组成的子向量,x w+1
Figure PCTCN2022112628-appb-000015
的第(w+1-1)Q+1到第w+1Q个元素组成的子向量,x i
Figure PCTCN2022112628-appb-000016
的第(i-1)Q+1到第iQ个元素组成的子向量,z 1
Figure PCTCN2022112628-appb-000017
的第(1-1)Q+1到第1Q个元素组成的子向量,z w+1
Figure PCTCN2022112628-appb-000018
的第(w+1-1)Q+1到第w+1Q个元素组成的子向量,z i
Figure PCTCN2022112628-appb-000019
的第(i-1)Q+1到第iQ个元素组成的子向量,辅助变量
Figure PCTCN2022112628-appb-000020
λ t是惩罚系数,||·|| 2表示取目标向量的2范数。上述问题的第k+1次迭代公式为:
Figure PCTCN2022112628-appb-000021
Figure PCTCN2022112628-appb-000022
for i=1,2,…,w+1,
Figure PCTCN2022112628-appb-000023
其中,I是单位矩阵,
Figure PCTCN2022112628-appb-000024
是辅助变量,u 1
Figure PCTCN2022112628-appb-000025
的第(1-1)Q+1到第1Q个元素组成的子向量,u w+1
Figure PCTCN2022112628-appb-000026
的第(w+1-1)Q+1到第w+1Q个元素组成的子向量,u i
Figure PCTCN2022112628-appb-000027
的第(i-1)Q+1到第iQ个元素组成 的子向量,u i定义类似x i,上标(·) (k)表示第k次迭代的变量值,ρ是迭代步长,可取定值,
Figure PCTCN2022112628-appb-000028
表示取复数的实数部分,
Figure PCTCN2022112628-appb-000029
是第k+1次迭代产生的向量
Figure PCTCN2022112628-appb-000030
的第(i-1)Q+1到第iQ个元素组成的子向量,收敛条件为:
Figure PCTCN2022112628-appb-000031
Figure PCTCN2022112628-appb-000032
是目标变量
Figure PCTCN2022112628-appb-000033
第k次迭代求得的值,ε为收敛门限,是一个较小的大于0的常数;
S303、根据步骤S302中得到的解
Figure PCTCN2022112628-appb-000034
计算各个估计格点(空间方向)上的能量分布向量
Figure PCTCN2022112628-appb-000035
并对向量
Figure PCTCN2022112628-appb-000036
中的元素排序,记排序后的
Figure PCTCN2022112628-appb-000037
Figure PCTCN2022112628-appb-000038
选取
Figure PCTCN2022112628-appb-000039
中前
Figure PCTCN2022112628-appb-000040
个元素,使其满足:
Figure PCTCN2022112628-appb-000041
其中,η是路径恢复门限,且满足0<η<1,||·|| 1表示取目标向量的1范数,[·] i表示向量的第i个元素,记恢复出的
Figure PCTCN2022112628-appb-000042
个元素在向量
Figure PCTCN2022112628-appb-000043
中对应的下标构成集合
Figure PCTCN2022112628-appb-000044
表示上式中第1个被选出的元素在
Figure PCTCN2022112628-appb-000045
中的原始顺序,
Figure PCTCN2022112628-appb-000046
表示上式中第2个被选出的元素在
Figure PCTCN2022112628-appb-000047
中的原始顺序,
Figure PCTCN2022112628-appb-000048
表示上式中第
Figure PCTCN2022112628-appb-000049
个被选出的元素在
Figure PCTCN2022112628-appb-000050
中的原始顺序,
Figure PCTCN2022112628-appb-000051
表示上式中第l个被选出的元素在
Figure PCTCN2022112628-appb-000052
中的原始顺序,空间中
Figure PCTCN2022112628-appb-000053
条路径的来波方向写成向量形式为:
Figure PCTCN2022112628-appb-000054
所述步骤S4中,根据估计出的来波方向重构上行部分阵列流形矩阵,并根据观测到的频域瞬时信息估计路径增益,具体包括以下步骤:
S401、将实际N根接收天线上的时域信号进行傅里叶变换,得到Q个长度为N的频域列向量信号,记为
Figure PCTCN2022112628-appb-000055
q表示第q个频点,构造各个频点上的上行接收天线的部分阵列流形矩阵
Figure PCTCN2022112628-appb-000056
(大小为
Figure PCTCN2022112628-appb-000057
),q表示第q个频点:
Figure PCTCN2022112628-appb-000058
其中,
Figure PCTCN2022112628-appb-000059
均是实际上行部分激活天线的导向矢量:
Figure PCTCN2022112628-appb-000060
p 1,p 2,p n,p N都是上行阵列中被选中的天线对应的次序,p n∈{0,1,2,…,M-1},i∈L;
S402、对Q个频点依此求解路径增益,
Figure PCTCN2022112628-appb-000061
条路径在第q个频点上的增益向量
Figure PCTCN2022112628-appb-000062
为:
Figure PCTCN2022112628-appb-000063
其中,
Figure PCTCN2022112628-appb-000064
的各个元素对应每一条路径上的路径增益,κ是一个极小的常数,用于确保求逆过程中矩阵的非奇异性。
所述步骤S4中,依据S402中估计出的路径增益,S303中估计出的来波方向重构完整的上行信道并将其对称至下行信道,具体操作如下:
按下式分别重构Q个频点上完整的
Figure PCTCN2022112628-appb-000065
阵列流形矩阵:
Figure PCTCN2022112628-appb-000066
for q=1,2,…,Q,
其中,
Figure PCTCN2022112628-appb-000067
是第q个频点在第1个角度上的完整导向矢量,
Figure PCTCN2022112628-appb-000068
第q个频点在第2个角度上的完整导向矢量,
Figure PCTCN2022112628-appb-000069
第q个频点在第L个角度上的完整导向矢量,
Figure PCTCN2022112628-appb-000070
是第q个频点在第i个角度上的完整导向矢量:
Figure PCTCN2022112628-appb-000071
重建所有天线上M×1的完整信道:
Figure PCTCN2022112628-appb-000072
for q=1,2,…,Q。
在时分双工模式下,
Figure PCTCN2022112628-appb-000073
就是第q个频点上的下行信道矩阵;至此,完整的下行信道为:
Figure PCTCN2022112628-appb-000074
式中,
Figure PCTCN2022112628-appb-000075
是第1个频点上的下行信道矩阵,
Figure PCTCN2022112628-appb-000076
是第Q个频点上的下行信道矩阵。
总之,本发明在非对称大规模MIMO(Multi-Input Multi-Output,MIMO)下提出了一种基于互质阵列的下行信道估计方法。首先,构建基于互质阵列的上下行非对称收发系统模型,并关注阵列宽带信号带来的频域方向偏移;其次,进行上行接收估计出上行信道,并由此恢复路径数,到达角,路径增益等信道参数;最后,利用上行信道恢复出的信道参数重建下行信道;对于宽带信号还可以利用不同频点上的群稀疏特性提高信道估计精度。本发明利用互质阵列角度分辨率高的特点,解决了恢复出的上行信道无法直接用于下行信道预编码的问题。本发明在非对称架构中有效地利用部分天线估计完整的下行信道,在降低上行链路接收压力和提高信道恢复准确性方面具有明显改善。
本发明采用以上技术方案与现有技术相比,具有以下技术效果:
(1)本发明在非对称收发架构下充分利用了高频信号在空间中的稀疏多径特性,将原本的信道估计问题转化为参数估计问题,利用群最小绝对收缩选择算子在ADMM(Alternating Direction Method of Multipliers,ADMM)优化框架下快速准确地估计来波方向,在达到 整体信道估计精度的同时,还确保了信道模型的稀疏特性,防止后续过拟合带来的误差。
(2)本发明在非对称架构的上行阵列中引入互质阵列,有效地消除了上下行阵列的分辨能力差异,确保了重建信道的准确性和下行预编码可用信道信息的有效性,从而提升系统的传输速率。
附图说明
图1为本发明的流程图。
图2为本发明中上行接收天线的选取示意图。
具体实施方式
下面结合附图对本发明的技术方案做进一步的详细说明:本实施例在以本发明技术方案为前提下进行实施,给出了详细的实施方式和具体的操作过程,但本发明的保护权限不限于下述的实施例。
本实施例提出了一种基于互质阵列的非对称大规模MIMO信道估计方法,如图1所示,包括以下步骤:
S1、构建基于互质阵列的上下行非对称大规模MIMO系统,该系统包括一个配有超大规模天线(天线数为M)的基站和K个单天线的用户。基站所有天线都有发送射频链路,只有N(N<<M,K≤N)根接收射频链路可连接到N根天线用于上行接收信号,用户与基站用Q个频点进行通信。数目关系为:M=mn+1,N=m+n-1,其中m<n,且m,n互质。按照互质阵列的方式选取上行接收天线,连接接收射频链,具体选取方式为:选中起始的连续n根,第2n+1根,第3n+1根,…,直至第mn+1根,共计N=m+n-1根天线(m,n分别是互质的两个数,用于设计非对称阵列上下行天线的个数)。天线之间的间隔d为λ min/2,其中λ min是各个频点中频率最大的子载波所对应的波长。
S2、基站方接收并估计得到部分天线的上行信道信息,将其变换 到频域并进行筛选重排后,构造虚拟线性均匀阵列。包括以下步骤:
S201、在连续的P个符号时间内,对上行阵列收到的P组Q个离散时域信号分别进行Q点FFT(Fast Fourier Transform,FFT)变换,得到P组Q个频点上的频域信号。记N×1的列向量x p,q为第p组信号中第q个频点对应的频域信号。
S202、对每个频点的P个频域信号进行自相关处理,得到Q个N×N的自相关矩阵:
Figure PCTCN2022112628-appb-000077
for q=1,…,Q,
在R q中将重复位置的阵元进行分类(上三角对应负滞后部分,故舍去)、平均后,按照顺序将其排列成长为M×1的列向量y q,[·] H表示共轭转置,[·] T表示转置。具体选取方式如下:
将完整的阵列从一端开始,天线序号记为0,1,2,…,M-1,其中上行激活的天线阵列序号记为一个长为N的集合{p 1,p 2,…,p N},将此集合按列排出一个N×N的矩阵T c
Figure PCTCN2022112628-appb-000078
自相关矩阵R q中各元素与
Figure PCTCN2022112628-appb-000079
中元素一一对应,即R q中第i行第j列元素[R q] i,j即为虚拟阵列中第[R tab] i,j个位置上的元素。
S203、将所有频点上的虚拟阵列信号排成一个QM的列向量:
Figure PCTCN2022112628-appb-000080
其中,长为M的列向量y i是第i个频点上的频域信号,[·] T表示取转置。
S3、利用空间稀疏特性构造基于压缩感知的群最小绝对收缩选 择算子,估计来波方向。包括以下步骤:
S301、在预先设定的入射角区间[θ lr]上构造QM×Q(w+1)的观测矩阵
Figure PCTCN2022112628-appb-000081
θ lr分别是左角度边界和右角度边界,并满足0≤θ lr≤π,这两个边界值可以通过DFT(Discrete Fourier Transform,DFT)等角度域检测手段获得;若不进行检测则θ l=0,θ r=π。w是在估计区间[θ lr]上划分的格点个数,w越大估计结果越精确,但复杂度也提示,QM×Q的子矩阵A i是第i个格点上的观测矩阵,
Figure PCTCN2022112628-appb-000082
是用于估计噪声的矩阵,各子观测矩阵按如下方式生成:
Figure PCTCN2022112628-appb-000083
其中省略号位置全部为0,a i,1是第1个频点在第i个格点上的导向矢量,a i,Q是第Q个频点在第i个格点上的导向矢量,a i,q是第q个频点在第i个格点上的导向矢量,且:
Figure PCTCN2022112628-appb-000084
e 1=[1,0,…,0] T
其中,e 1长度为M,j是虚数单位,i是格点次序,
Figure PCTCN2022112628-appb-000085
是格点间距大小,
Figure PCTCN2022112628-appb-000086
表示第i个格点所代表的角度,λ q是第q个频点对应的波长;
S302、依据压缩感知理论在ADMM(Alternating Direction Method of Multipliers,ADMM)框架下写出L21范数约束的线性回归问题:
Figure PCTCN2022112628-appb-000087
Figure PCTCN2022112628-appb-000088
其中,长度为Q(w+1)的列向量
Figure PCTCN2022112628-appb-000089
是待求解的目标变 量,代表着各个频点在各个估计格点上的能量大小,x 1
Figure PCTCN2022112628-appb-000090
的第(1-1)Q+1到第1Q个元素组成的子向量,x w+1
Figure PCTCN2022112628-appb-000091
的第(w+1-1)Q+1到第w+1Q个元素组成的子向量,x i
Figure PCTCN2022112628-appb-000092
的第(i-1)Q+1到第iQ个元素组成的子向量,z 1
Figure PCTCN2022112628-appb-000093
的第(1-1)Q+1到第1Q个元素组成的子向量,z w+1
Figure PCTCN2022112628-appb-000094
的第(w+1-1)Q+1到第w+1Q个元素组成的子向量,z i
Figure PCTCN2022112628-appb-000095
的第(i-1)Q+1到第iQ个元素组成的子向量,辅助变量
Figure PCTCN2022112628-appb-000096
λ t是惩罚系数,||·|| 2表示取目标向量的2范数。上述问题的第k+1次迭代公式为:
Figure PCTCN2022112628-appb-000097
Figure PCTCN2022112628-appb-000098
for i=1,2,…,w+1,
Figure PCTCN2022112628-appb-000099
其中,I是单位矩阵,
Figure PCTCN2022112628-appb-000100
是辅助变量,u 1
Figure PCTCN2022112628-appb-000101
的第(1-1)Q+1到第1Q个元素组成的子向量,u w+1
Figure PCTCN2022112628-appb-000102
的第(w+1-1)Q+1到第w+1Q个元素组成的子向量,u i
Figure PCTCN2022112628-appb-000103
的第(i-1)Q+1到第iQ个元素组成的子向量,u i定义类似x i,上标(·) (k)表示第k次迭代的变量值,ρ是迭代步长,可取定值,
Figure PCTCN2022112628-appb-000104
表示取复数的实数部分,
Figure PCTCN2022112628-appb-000105
是第k+1次迭代产生的向量
Figure PCTCN2022112628-appb-000106
的第(i-1)Q+1到第iQ个元素组成的子向量,收敛条件为:
Figure PCTCN2022112628-appb-000107
Figure PCTCN2022112628-appb-000108
是目标变量
Figure PCTCN2022112628-appb-000109
第k次迭代求得的向量,ε为收敛门限,是一个较小的大于0的常数;
S303、根据步骤S302中得到的解
Figure PCTCN2022112628-appb-000110
计算各个估计格点(空间方向)上的能量分布向量
Figure PCTCN2022112628-appb-000111
并对向 量
Figure PCTCN2022112628-appb-000112
中的元素排序,记排序后的
Figure PCTCN2022112628-appb-000113
Figure PCTCN2022112628-appb-000114
选取
Figure PCTCN2022112628-appb-000115
中前
Figure PCTCN2022112628-appb-000116
个元素,使其满足:
Figure PCTCN2022112628-appb-000117
其中,η是路径恢复门限,且满足0<η<1,||·|| 1表示取目标向量的1范数,[·] i表示向量的第i个元素,记恢复出的
Figure PCTCN2022112628-appb-000118
个元素在向量
Figure PCTCN2022112628-appb-000119
中对应的下标构成集合
Figure PCTCN2022112628-appb-000120
表示上式中第1个被选出的元素在
Figure PCTCN2022112628-appb-000121
中的原始顺序,
Figure PCTCN2022112628-appb-000122
表示上式中第2个被选出的元素在
Figure PCTCN2022112628-appb-000123
中的原始顺序,
Figure PCTCN2022112628-appb-000124
表示上式中第
Figure PCTCN2022112628-appb-000125
个被选出的元素在
Figure PCTCN2022112628-appb-000126
中的原始顺序,
Figure PCTCN2022112628-appb-000127
表示上式中第l个被选出的元素在
Figure PCTCN2022112628-appb-000128
中的原始顺序,空间中
Figure PCTCN2022112628-appb-000129
条路径的来波方向写成向量形式为:
Figure PCTCN2022112628-appb-000130
S4、根据估计出的来波方向重构部分阵列流形矩阵,并根据后续观测到的频域瞬时信息估计路径增益。包括以下步骤:
S401、将实际N根接收天线上的时域信号进行傅里叶变换,得到Q个长度为N的频域列向量信号,记为
Figure PCTCN2022112628-appb-000131
q表示第q个频点,构造各个频点上的上行接收天线的部分阵列流形矩阵
Figure PCTCN2022112628-appb-000132
(大小为
Figure PCTCN2022112628-appb-000133
),q表示第q个频点:
Figure PCTCN2022112628-appb-000134
其中,
Figure PCTCN2022112628-appb-000135
均是实际上行部分激活天线的导向矢量:
Figure PCTCN2022112628-appb-000136
p 1,p 2,p n,p N都是上行阵列中被选中的天线对应的次序, p n∈{0,1,2,…,M-1},i∈L;
S402、对Q个频点依此求解路径增益,
Figure PCTCN2022112628-appb-000137
条路径在第q个频点上的增益向量
Figure PCTCN2022112628-appb-000138
为:
Figure PCTCN2022112628-appb-000139
其中,
Figure PCTCN2022112628-appb-000140
的各个元素对应每一条路径上的路径增益,κ是一个极小的常数,用于确保求逆过程中矩阵的非奇异性。
S5、依据估计出的路径增益,来波方向重构完整的上行信道,根据互易性将其对称至下行信道。具体操作如下:
所述步骤S4中,依据S402中估计出的路径增益,S303中估计出的来波方向重构完整的上行信道并将其对称至下行信道,具体操作如下:
按下式分别重构Q个频点上完整的
Figure PCTCN2022112628-appb-000141
阵列流形矩阵:
Figure PCTCN2022112628-appb-000142
其中,
Figure PCTCN2022112628-appb-000143
是第q个频点在第1个角度上的完整导向矢量,
Figure PCTCN2022112628-appb-000144
第q个频点在第2个角度上的完整导向矢量,
Figure PCTCN2022112628-appb-000145
第q个频点在第L个角度上的完整导向矢量,
Figure PCTCN2022112628-appb-000146
是第q个频点在第i个角度上的完整导向矢量:
Figure PCTCN2022112628-appb-000147
重建所有天线上M×1的完整信道:
Figure PCTCN2022112628-appb-000148
for q=1,2,…,Q。
在时分双工模式下,
Figure PCTCN2022112628-appb-000149
就是第q个频点上的下行信道矩阵;至此,完整的下行信道为:
Figure PCTCN2022112628-appb-000150
式中,
Figure PCTCN2022112628-appb-000151
是第1个频点上的下行信道矩阵,
Figure PCTCN2022112628-appb-000152
是第Q个频点上的 下行信道矩阵。
以上所述,仅为本发明中的具体实施方式,但本发明的保护范围并不局限于此,任何熟悉该技术的人在本发明所揭露的技术范围内,可理解想到的变换或替换,都应涵盖在本发明的包含范围之内,因此,本发明的保护范围应该以权利要求书的保护范围为准。

Claims (7)

  1. 一种基于互质阵列的非对称大规模MIMO信道估计方法,其特征在于,包括以下步骤:
    S1、构建基于互质阵列的上下行非对称大规模MIMO系统,该系统包括一个配有超大规模天线的基站和K个单天线的用户;基站天线数为M,基站所有天线都有发送射频链路,只有N根接收射频链路可连接到N根天线用于上行接收信号,用户与基站用Q个频点进行通信,数目关系为:M=mn+1,N=m+n-1,其中m<n,且m,n互质;按照互质阵列的方式选取上行接收天线,连接接收射频链;
    S2、基站方接收并估计得到部分天线的上行信道信息,将其变换到频域并进行筛选重排后,构造虚拟线性均匀阵列;
    S3、利用空间稀疏特性构造基于压缩感知的群最小绝对收缩选择算子,估计来波方向;
    S4、根据估计出的来波方向重构部分阵列流形矩阵,并根据后续观测到的频域瞬时信息估计路径增益;
    S5、依据估计出的路径增益,来波方向重构完整的上行信道,根据互易性将其对称至下行信道。
  2. 根据权利要求1所述一种基于互质阵列的非对称大规模MIMO信道估计方法,其特征在于,所述步骤S1中,选取上行接收天线的具体方式为:选中起始的连续n根,第2n+1根,第3n+1根,…,直至第mn+1根,共计N=m+n-1根天线;天线之间的间隔d为λ min/2, 其中λ min是各个频点中频率最大的子载波所对应的波长。
  3. 根据权利要求2所述一种基于互质阵列的非对称大规模MIMO信道估计方法,其特征在于,所述步骤S2中,基站方接收并估计得到部分天线的上行信道信息,将其变换到频域并进行筛选重排后,构造虚拟线性均匀阵列,具体包括以下步骤:
    S201、在连续的P个符号时间内,对上行阵列收到的P组Q个离散时域信号分别进行Q点FFT变换,得到P组Q个频点上的频域信号,记N×1的列向量x p,q为第p组信号中第q个频点对应的频域信号;
    S202、对每个频点的P个频域信号进行自相关处理,得到Q个N×N的自相关矩阵:
    Figure PCTCN2022112628-appb-100001
    在R q中将重复位置的阵元进行分类、平均后,按照顺序将其排列成长为M×1的列向量y q
    S203、将所有频点上的虚拟阵列信号排成一个QM的列向量:
    Figure PCTCN2022112628-appb-100002
    其中,长为M的列向量y i是第i个频点上的频域信号。
  4. 根据权利要求3所述一种基于互质阵列的非对称大规模MIMO信道估计方法,其特征在于,所述步骤S202中,具体选取方式如下:
    将完整的阵列从一端开始,天线序号记为0,1,2,...,M-1,其中上行激活的天线阵列序号记为一个长为N的集合{p 1,p 2,...,p N},将此集合按列排出一个N×N的矩阵T c
    Figure PCTCN2022112628-appb-100003
    自相关矩阵R q中各元素与
    Figure PCTCN2022112628-appb-100004
    中元素一一对应,即R q中第i行第j列元素[R q] i,j即为虚拟阵列中第[R tab] i,j个位置上的元素。
  5. 根据权利要求4所述一种基于互质阵列的非对称大规模MIMO信道估计方法,其特征在于,所述步骤S3中,利用基于压缩感知的群最小绝对收缩选择算子构造估计问题,在ADMM优化框架下求解该问题,从而估计来波方向,具体包括以下步骤:
    S301、在预先设定的入射角区间[θ lr]上构造QM×Q(w+1)的观测矩阵
    Figure PCTCN2022112628-appb-100005
    θ lr分别是左角度边界和右角度边界,并满足0≤θ lr≤π,w是在估计区间[θ lr]上划分的格点个数,QM×Q的子矩阵A i是第i个格点上的观测矩阵,
    Figure PCTCN2022112628-appb-100006
    是用于估计噪声的矩阵,各子观测矩阵按如下方式生成:
    Figure PCTCN2022112628-appb-100007
    a i,q是第q个频点在第i个格点上的导向矢量,且:
    Figure PCTCN2022112628-appb-100008
    e 1=[1,0,...,0] T
    其中,e 1长度为M,j是虚数单位,i是格点次序,
    Figure PCTCN2022112628-appb-100009
    是格点间距大小,
    Figure PCTCN2022112628-appb-100010
    表示第i个格点所代表的角度,λ q是第q个频点对应的波长;
    S302、依据压缩感知理论在ADMM框架下写出L21范数约束的线性回归问题:
    Figure PCTCN2022112628-appb-100011
    Figure PCTCN2022112628-appb-100012
    其中,长度为Q(w+1)的列向量
    Figure PCTCN2022112628-appb-100013
    是待求解的目标变量,代表着各个频点在各个估计格点上的能量大小,x i
    Figure PCTCN2022112628-appb-100014
    的第(i-1)Q+1到第iQ个元素组成的子向量,z i
    Figure PCTCN2022112628-appb-100015
    的第(i-1)Q+1到第iQ个元素组成的子向量,辅助变量
    Figure PCTCN2022112628-appb-100016
    λ t是惩罚系数,上述问题的第k+1次迭代公式为:
    Figure PCTCN2022112628-appb-100017
    Figure PCTCN2022112628-appb-100018
    Figure PCTCN2022112628-appb-100019
    其中,I是单位矩阵,
    Figure PCTCN2022112628-appb-100020
    是辅助变量,u i
    Figure PCTCN2022112628-appb-100021
    的第(i-1)Q+1到第iQ个元素组成的子向量,ρ是迭代步长,
    Figure PCTCN2022112628-appb-100022
    是第k+1次迭代产生的向量
    Figure PCTCN2022112628-appb-100023
    的第(i-1)Q+1到第iQ个元素组成的子向量,收敛条件为:
    Figure PCTCN2022112628-appb-100024
    Figure PCTCN2022112628-appb-100025
    是目标变量
    Figure PCTCN2022112628-appb-100026
    第k次迭代求得的向量,ε为收敛门限,是一个较小的大于0的常数;
    S303、根据步骤S302中得到的解
    Figure PCTCN2022112628-appb-100027
    计算各个估计 格点上的能量分布向量
    Figure PCTCN2022112628-appb-100028
    并对向量
    Figure PCTCN2022112628-appb-100029
    中的元素排序,记排序后的
    Figure PCTCN2022112628-appb-100030
    Figure PCTCN2022112628-appb-100031
    选取
    Figure PCTCN2022112628-appb-100032
    中前
    Figure PCTCN2022112628-appb-100033
    个元素,使其满足:
    Figure PCTCN2022112628-appb-100034
    其中,η是路径恢复门限,且满足0<η<1,记恢复出的
    Figure PCTCN2022112628-appb-100035
    个元素在向量
    Figure PCTCN2022112628-appb-100036
    中对应的下标构成集合
    Figure PCTCN2022112628-appb-100037
    表示上式中第l个被选出的元素在
    Figure PCTCN2022112628-appb-100038
    中的原始顺序,空间中
    Figure PCTCN2022112628-appb-100039
    条路径的来波方向写成向量形式为:
    Figure PCTCN2022112628-appb-100040
  6. 根据权利要求5所述一种基于互质阵列的非对称大规模MIMO信道估计方法,其特征在于,所述步骤S4中,根据估计出的来波方向重构上行部分阵列流形矩阵,并根据观测到的频域瞬时信息估计路径增益,具体包括以下步骤:
    S401、将实际N根接收天线上的时域信号进行傅里叶变换,得到Q个长度为N的频域列向量信号,记为
    Figure PCTCN2022112628-appb-100041
    q表示第q个频点,构造各个频点上的上行接收天线的部分阵列流形矩阵
    Figure PCTCN2022112628-appb-100042
    q表示第q个频点:
    Figure PCTCN2022112628-appb-100043
    其中,
    Figure PCTCN2022112628-appb-100044
    均是实际上行部分激活天线的导向矢量:
    Figure PCTCN2022112628-appb-100045
    p n是上行阵列中被选中的天线对应的次序,p n∈{0,1,2,...,M-1},i∈L;
    S402、对Q个频点依此求解路径增益,
    Figure PCTCN2022112628-appb-100046
    条路径在第q个频点上的增益向量
    Figure PCTCN2022112628-appb-100047
    为:
    Figure PCTCN2022112628-appb-100048
    其中,
    Figure PCTCN2022112628-appb-100049
    的各个元素对应每一条路径上的路径增益,κ是一个极小的常数,用于确保求逆过程中矩阵的非奇异性。
  7. 根据权利要求6所述一种基于互质阵列的非对称大规模MIMO信道估计方法,其特征在于,所述步骤S4中,依据S402中估计出的路径增益,S303中估计出的来波方向重构完整的上行信道并将其对称至下行信道,具体操作如下:
    按下式分别重构Q个频点上完整的
    Figure PCTCN2022112628-appb-100050
    阵列流形矩阵:
    Figure PCTCN2022112628-appb-100051
    其中,
    Figure PCTCN2022112628-appb-100052
    是第q个频点在第i个角度上的完整导向矢量:
    Figure PCTCN2022112628-appb-100053
    重建所有天线上M×1的完整信道:
    Figure PCTCN2022112628-appb-100054
    在时分双工模式下,
    Figure PCTCN2022112628-appb-100055
    就是第q个频点上的下行信道矩阵;至此,完整的下行信道为:
    Figure PCTCN2022112628-appb-100056
    式中,
    Figure PCTCN2022112628-appb-100057
    是第1个频点上的下行信道矩阵,
    Figure PCTCN2022112628-appb-100058
    是第Q个频点上的下行信道矩阵。
PCT/CN2022/112628 2022-01-07 2022-08-16 一种基于互质阵列的非对称大规模mimo信道估计方法 WO2023130726A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210014946.8A CN114401171B (zh) 2022-01-07 2022-01-07 一种基于互质阵列的非对称大规模mimo信道估计方法
CN202210014946.8 2022-01-07

Publications (1)

Publication Number Publication Date
WO2023130726A1 true WO2023130726A1 (zh) 2023-07-13

Family

ID=81228735

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/112628 WO2023130726A1 (zh) 2022-01-07 2022-08-16 一种基于互质阵列的非对称大规模mimo信道估计方法

Country Status (2)

Country Link
CN (1) CN114401171B (zh)
WO (1) WO2023130726A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114401171B (zh) * 2022-01-07 2023-07-14 南京邮电大学 一种基于互质阵列的非对称大规模mimo信道估计方法
CN116094876B (zh) * 2023-03-10 2023-08-29 南京邮电大学 一种基于非对称架构的正交时频空系统的信道估计方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180159638A1 (en) * 2016-12-07 2018-06-07 Qatar University Method of identifying faulty antenna elements in massive uniform linear antenna arrays
CN109257309A (zh) * 2018-10-24 2019-01-22 东南大学 一种高性能的大规模mimo下行链路传输信道估计方法
CN111107023A (zh) * 2019-09-23 2020-05-05 南京邮电大学 在大规模mimo中基于光滑范数的压缩感知信道估计方法
CN111665468A (zh) * 2020-06-08 2020-09-15 浙江大学 基于单比特量化信号虚拟域统计量重建的互质阵列波达方向估计方法
CN114401171A (zh) * 2022-01-07 2022-04-26 南京邮电大学 一种基于互质阵列的非对称大规模mimo信道估计方法

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI609577B (zh) * 2016-01-29 2017-12-21 財團法人工業技術研究院 執行上行鏈路通道估測的方法及使用所述方法的基地台
CN109450500A (zh) * 2018-11-16 2019-03-08 南京邮电大学 基于嵌套阵列的毫米波联合多普勒和信道估计方法
US11728571B2 (en) * 2019-07-12 2023-08-15 Arizona Board Of Regents On Behalf Of Arizona State University Large intelligent surfaces with sparse channel sensors
CN111193679B (zh) * 2020-01-09 2022-04-19 中国传媒大学 一种基于互质阵列系统的信道估计方法及系统
US11855692B2 (en) * 2020-01-28 2023-12-26 Phase Sensitive Innovations, Inc. Phased-array mapping for beamspace processing and beamspace processor

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180159638A1 (en) * 2016-12-07 2018-06-07 Qatar University Method of identifying faulty antenna elements in massive uniform linear antenna arrays
CN109257309A (zh) * 2018-10-24 2019-01-22 东南大学 一种高性能的大规模mimo下行链路传输信道估计方法
CN111107023A (zh) * 2019-09-23 2020-05-05 南京邮电大学 在大规模mimo中基于光滑范数的压缩感知信道估计方法
CN111665468A (zh) * 2020-06-08 2020-09-15 浙江大学 基于单比特量化信号虚拟域统计量重建的互质阵列波达方向估计方法
CN114401171A (zh) * 2022-01-07 2022-04-26 南京邮电大学 一种基于互质阵列的非对称大规模mimo信道估计方法

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
"Master's Thesis", 27 May 2020, CHONGQING UNIVERSITY OF POSTS AND TELECOMMUNICATIONS, CN, article WANG, QIANG: "Multiple-Input Multiple-Output (MIMO) Research on Estimation Algorithm of Radar Direction of Arrival", pages: 1 - 61, XP009547100, DOI: 10.27675/d.cnki.gcydx.2020.000364 *
SHI JUNPENG; HU GUOPING; ZHANG XIAOFEI; SUN FENGGANG; ZHENG WANG; XIAO YU: "Generalized Co-Prime MIMO Radar for DOA Estimation With Enhanced Degrees of Freedom", IEEE SENSORS JOURNAL, IEEE, USA, vol. 18, no. 3, 1 February 2018 (2018-02-01), USA, pages 1203 - 1212, XP011675618, ISSN: 1530-437X, DOI: 10.1109/JSEN.2017.2782746 *

Also Published As

Publication number Publication date
CN114401171A (zh) 2022-04-26
CN114401171B (zh) 2023-07-14

Similar Documents

Publication Publication Date Title
WO2023130726A1 (zh) 一种基于互质阵列的非对称大规模mimo信道估计方法
CN113225276B (zh) 一种面向智能反射面辅助通信系统的半盲信道估计方法
Ali et al. Millimeter wave beam-selection using out-of-band spatial information
CN1695271B (zh) 包括虚拟天线单元的天线阵
CN113179231B (zh) 一种毫米波大规模mimo系统中波束空间信道估计方法
CN109005133B (zh) 双稀疏多径信道模型及基于此模型的信道估计方法
CN114726687B (zh) 一种智能反射面辅助毫米波大规模mimo系统的信道估计方法
CN112929962B (zh) 定位方法、装置、计算机设备和存储介质
CN114189408A (zh) 一种用于智能反射面辅助毫米波系统的信道估计方法
CN115022134B (zh) 基于非迭代重构网络的毫米波大规模mimo系统信道估计方法及系统
CN104333407A (zh) 一种信道信息的获取方法及装置
CN113411107A (zh) 一种基于波束束的毫米波大规模mimo系统宽带信道估计方法
CN114338301A (zh) 一种基于压缩感知的ris辅助毫米波系统的信道估计方法
CN107092005A (zh) 一种时空联合的阵列处理方法及装置
WO2016183957A1 (zh) 一种天线通道的降阶方法及装置
CN114172546B (zh) Ris辅助mimo系统中一种多参数迭代估计方法
CN110212951B (zh) 一种基于Butler矩阵的大规模MIMO信道估计方法
Srivastava et al. Sparse Bayesian learning (SBL)-based frequency-selective channel estimation for millimeter wave hybrid MIMO systems
CN114726686A (zh) 一种均匀面阵毫米波大规模mimo信道估计方法
CN110636018B (zh) 一种网格补偿大规模mimo信道估计方法
CN115065446B (zh) 一种基于流形学习的信道信息反馈方法
CN116056118A (zh) 基于主被动混合智能超表面的无线通信传输方法及系统
CN114244658B (zh) 大规模mimo系统中的基于倍角估计的信道估计方法
WO2022166477A1 (zh) 定位方法、装置、基站、计算机设备和存储介质
CN113242193B (zh) 一种低训练开销的混合大规模mimo-ofdm系统信道估计方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22918182

Country of ref document: EP

Kind code of ref document: A1