WO2017000752A1 - 一种fdd大规模mimo系统下行链路训练序列设计方法 - Google Patents

一种fdd大规模mimo系统下行链路训练序列设计方法 Download PDF

Info

Publication number
WO2017000752A1
WO2017000752A1 PCT/CN2016/085000 CN2016085000W WO2017000752A1 WO 2017000752 A1 WO2017000752 A1 WO 2017000752A1 CN 2016085000 W CN2016085000 W CN 2016085000W WO 2017000752 A1 WO2017000752 A1 WO 2017000752A1
Authority
WO
WIPO (PCT)
Prior art keywords
training sequence
channel
base station
time
downlink
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.)
Ceased
Application number
PCT/CN2016/085000
Other languages
English (en)
French (fr)
Inventor
王向阳
赵洋
王东
杨静雯
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.)
Southeast University
Original Assignee
Southeast University
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 Southeast University filed Critical Southeast University
Publication of WO2017000752A1 publication Critical patent/WO2017000752A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • 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

Definitions

  • the invention belongs to the technical field of wireless and mobile communication, and relates to a downlink training sequence design method for an FDD massive MIMO system.
  • the so-called massive MIMO technology is actually the evolution of MIMO technology.
  • the core idea is to establish multiple parallel data transmission channels in the same frequency and time resources by deploying a large number of transmit and receive antennas, thereby improving the overall system. Spectrum utilization efficiency and energy efficiency, while greatly increasing the data transmission rate.
  • Massive MIMO systems generally employ a channel estimation method based on training sequences.
  • Most large-scale MIMO studies assume that the system works in Time Division Duplex (TDD) mode instead of Frequency Division Duplex (FDD) mode because of the channel state in FDD mode.
  • Channel State Information (CSI) acquisition is more difficult.
  • TDD Time Division Duplex
  • FDD Frequency Division Duplex
  • CSI Channel State Information
  • each user needs to feed back CSI to the base station after estimating the channel.
  • the size of the huge channel vector occupies a large amount of uplink resources. It is assumed that the advantage of the large-scale MIMO system operating in the TDD mode is that the base station can acquire the downlink CSI using the reciprocity of the uplink and downlink channels of the TDD system.
  • the length of the training sequence is determined by the total number of users, so the number of antennas of the base station can be sufficiently large.
  • the reciprocity assumptions of the uplink and downlink channels are not strictly established. Therefore, in an actual communication system, a reciprocity calibration technique needs to be adopted to ensure the accuracy of the obtained downlink CSI.
  • the invention provides a downlink training sequence design method for an FDD massive MIMO system.
  • a base station equipped with a large-scale antenna array communicates with a single-antenna user terminal, and acquisition of downlink channel state information is an important issue.
  • the channel estimation of the downlink employs a training sequence based approach, which requires training sequences that are advantageous in terms of accuracy and time overhead.
  • the present invention employs continuous channel estimation based on Kalman filtering while taking advantage of the temporal and spatial correlation properties of the channel. During each coherence time, the user performs Kalman filtering and prediction to obtain the channel state information estimation value of the current coherence time, and gives the channel prediction of the next coherence time.
  • Some intermediate quantities generated during the Kalman filtering and prediction process ie some necessary statistical information, are passed by the user.
  • the uplink feeds back the necessary statistical information back to the base station to facilitate optimal training sequence design for the base station. It can be seen from the simulation analysis that, on the one hand, the invention can improve the accuracy of channel estimation; on the other hand, the downlink training duration of the FDD massive MIMO system can be greatly reduced, and thus more time is used for transmitting useful signals, This greatly improves link throughput and system capacity.
  • the base station is configured with N t antennas.
  • N t can take a very large value
  • the users in the cell are terminals configured with a single antenna
  • the base station is on the same time-frequency resource. Multiple users communicate wirelessly.
  • a scenario in which downlink data transmission is performed for a base station and a single antenna user which can be regarded as a multiple input single output (MISO) system, which models the downlink channel as a discrete time block fading channel and defines a coherence time length of T c (unit is channel use, symbol time), that is, the channel state information (CSI) remains unchanged for one coherence time and continues until the next coherence time.
  • MISO multiple input single output
  • n i (k) satisfies the standard complex Gaussian distribution (ie, n i (k) ⁇ CN(0, 1)) represents additive white Gaussian noise (AWGN) at the current time.
  • AWGN additive white Gaussian noise
  • a training sequence based approach can be employed and orthogonal training sequences are designed (the length of the training sequence is equal to the number of base station antennas), however, this can result in unacceptable time overhead.
  • Analysis of this MISO system shows that on the one hand, due to the large number of base station antennas, the channels between multiple antennas and users will not be completely independent, but there will be spatial correlation; on the other hand, because users will not Extremely high speeds move within the cell, and the geographic scene does not change dramatically, so the channel is not completely independent for two consecutive coherence times.
  • the channel can be written as follows:
  • represents the time correlation coefficient
  • g i is a set of N t -dimensional column vectors representing a complex Gaussian process, each element of which obeys the standard complex Gaussian distribution, ie And for different coherence times g i are completely independent.
  • the matrix R h in the equation is a channel spatial correlation matrix and can be modeled as an exponential model, ie
  • the parameter a (0 ⁇ a ⁇ 1) is the spatial correlation coefficient of the channel, and the larger a is, the stronger the correlation between the channels is.
  • the frame structure of the massive MIMO system is shown in Figure 1.
  • the coherence time T c is divided into three parts: the first part is the channel training stage, and the time length is recorded as T t ; the second part is the channel state information (CSI) feedback.
  • the third part is the useful information transmission phase, and its length of time is recorded as T d .
  • the length of the CSI feedback phase is negligible.
  • the T t -dimensional training signal received by the user can be expressed as:
  • y i,t X i,t H h i +n i,t ,(4)
  • X i,t [x i (1),...,x i (T t )]
  • n i,t [n i (1),...,n i (T t )] H is an additive white Gaussian noise and is a T t -dimensional column vector.
  • Equation (6) Represents the Kalman gain matrix.
  • the Kalman filter contains two parts of estimation and prediction, where equation (5) represents the channel estimation process of the i-th coherence time, and R i
  • the initial value of the Kalman filter is designed as: And R 0
  • the base station After the channel estimation is completed at the UE and fed back to the base station through the uplink, the base station wants the user to transmit a useful data signal and preprocess the transmitted data.
  • the signal sent by the base station can be expressed as:
  • the focus of the present invention is on the training sequence design of the downlink of the FDD massive MIMO system, and the design of the training sequence is based on the Minimized Estimated Mean Square Error (MMSE) criterion, and the statistical information required in the design is estimated by the UE. Calculated, so the channel estimation uses a closed-loop training mode.
  • MMSE Minimized Estimated Mean Square Error
  • the closed-loop training model is shown in Figure 2.
  • the base station sends the training sequence X i,t to the user; after receiving the corresponding training signal, the user obtains the channel estimate of the current coherent time by using Kalman filtering. The value is simultaneously predicted by the channel of the next coherent time; finally, the calculated channel prediction error covariance matrix R i+1
  • the training sequence X i,t is to be quantified, and R i
  • i can be further simplified to obtain:
  • the first step is to derive that step (a) is derived from the feature decomposition of R i
  • ⁇ i represents the diagonal matrix of the eigenvalues of R i
  • the Lagrangian multiplier method can be used, and the Lagrangian function written by the optimization problem is:
  • the Lagrangian function L( ⁇ , ⁇ i, l ) is separately biased to the variable ⁇ i, l and its value is 0.
  • the value of ⁇ i, l obtained at this time is:
  • the optimal value of ⁇ i,l is determined by the Lagrangian multiplier ⁇ .
  • the value of ⁇ can be calculated, and the value of the corresponding ⁇ i,l is also determined.
  • the iteration of the multiplier ⁇ is given by:
  • represents the step size in the iterative process, which determines the iteration speed of ⁇ ; in addition, the initial value of ⁇ is:
  • the optimal training sequence X i,t,opt of the i-th coherence time is determined by the eigenvector of the prediction error covariance matrix R i
  • the columns of X i,t,opt are mutually orthogonal.
  • the power factor multiplied by each column of X i,t,opt is obtained by the water injection algorithm, specifically for the eigenvalue of R i
  • FIG. 1 is a schematic diagram of a frame structure of a massive MIMO system
  • 2 is a schematic diagram of a concept of downlink closed-loop channel estimation
  • 3 is an estimated normalized mean squared error corresponding to three different channel estimation methods and different channel spatial correlation coefficients for the value of the downlink MIMO system downlink;
  • Figure 4 shows the normalized average of three different channel estimation methods in the downlink of a massive MIMO system. Receive signal-to-noise ratio and the effect of different channel spatial correlation coefficients on the value;
  • 5 is a diagram showing the normalized average received signal to noise ratio of the channel estimation method proposed in the present invention and the effect of different training sequence transmission signal to noise ratios and user moving speed on the value in the downlink of the massive MIMO system;
  • FIG. 6 is a diagram showing the effect of the normalized average received signal-to-noise ratio of the channel estimation method proposed in the present invention on the downlink of the massive MIMO system with the number of base station antennas and the length of different training sequences.
  • the Monte Carlo method is used to obtain the channel estimation mean square error of the first ten coherent times and Normalized average received signal to noise ratio.
  • Kalman filter-based closed-loop channel training method first determining the training sequence set, the user selects the best training sequence in the set and feeds its sequence back to the base station
  • the channel estimation mean square error and normalization The average received signal to noise ratio is also given by simulation.
  • Embodiment 1 Channel training scenario in a single coherence time
  • the method is simply single-shot training.
  • the traditional MMSE channel estimation method is adopted when channel estimation is performed in each coherent time.
  • the optimal training sequence design also satisfies the minimum mean square error criterion.
  • the mean square error of the corresponding single-shot training and the normalized received signal noise are shown in Figures 3 and 4.
  • Embodiment 2 Kalman Filter Based Closed Loop Channel Estimation Scene of a Preset Training Sequence Set
  • the channel estimation mean square error and the normalized average received signal-to-noise ratio of another closed-loop channel training method based on Kalman filtering are also given by simulation.
  • a closed-loop channel estimation method with an alternative set In order to distinguish the training mode and the training sequence design method proposed by the present invention, it is called a closed-loop channel estimation method with an alternative set.
  • the method differs from the present invention in that the base station prepares a training sequence in advance according to certain criteria. It contains 2 B alternative training sequences and B is the feedback length.
  • the method adopts a Kalman filter-based channel estimation method, and the user performs channel estimation and prediction in each coherence time, and based on the criterion of minimizing the estimated mean square error or maximizing the received signal to noise ratio.
  • the user adopts the Kalman filtering method, which not only considers the training sequence received at the current coherence time, but also considers the previous received value. Therefore, with the number of iterations Increased, the accuracy of Kalman filtering also increases.
  • Embodiment 3 Kalman filtering-based closed-loop channel estimation scenario for real-time design of training sequences
  • the training sequence design proposed by the present invention is real-time, that is, the optimal training sequence of the next coherence time is designed in each coherent time.
  • the design of the training sequence is done by the base station.
  • the user is responsible for feeding back the information necessary for the training sequence design (such as the channel prediction error covariance matrix R i
  • the number of base station antennas N t is set to 64
  • the training sequence time length T t 2
  • Figure 3 and Figure 4 are the performance comparisons of the estimation methods in the three examples. It can be seen from the figure that the design method of the proportional method of this example is more accurate. This is because the power distribution is performed during the design of the training sequence. Higher power is allocated than larger channels.
  • Figure 5 depicts the value of the normalized received signal to noise ratio of the channel estimation method of the present invention at different signal to noise ratios and different user movement rates.
  • the time correlation coefficient ⁇ decreases, that is, the correlation in the time dimension is weakened, and the accuracy of the channel estimation is also reduced.
  • the power of the training sequence is increased (the transmission signal-to-noise ratio becomes larger), the accuracy of the channel estimation is correspondingly improved, but the performance is not obvious in the figure.
  • the channel estimation method of this example has a limitation called saturation effect, that is, the upper bound of the received signal-to-noise ratio is only affected by the spatial correlation coefficient a, and the upper bound is also determined after a determination, so no matter the transmitted signal noise The ratio of normalized received signal-to-noise ratio does not increase significantly.
  • Figure 6 is a graph showing the normalized acceptance signal-to-noise ratio of the channel estimation method of the present invention in the ninth coherence time as a function of the number of base station antennas.

Landscapes

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

Abstract

本发明提出一种FDD大规模MIMO系统下行链路信道估计训练序列的设计方法。在此系统中,配置大规模多天线阵列的基站与单天线终端用户进行通信。本发明利用多天线MIMO信道的时间与空间相关特性,基于卡尔曼滤波进行连续信道估计。在每个相干时间,用户进行卡尔曼滤波和预测,得到当前相干时间内的信道状态信息估计值,并且给出下一个相干时间的信道预测值。在卡尔曼滤波和预测过程中产生了中间变量的统计信息,用户通过上行链路将必要的统计信息反馈回基站,以便基站进行最优训练序列设计。通过仿真分析可以看出,本方案能够提高信道估计的准确性,同时减少系统下行链路训练序列占用时长,从而提高了链路吞吐率与系统容量。

Description

一种FDD大规模MIMO系统下行链路训练序列设计方法 技术领域
本发明属于无线及移动通信技术领域,涉及一种FDD大规模MIMO系统下行链路训练序列设计方法。
背景方法
所谓大规模MIMO技术,实际上是MIMO技术的演进,其核心思想是通过部署海量的发射天线和接收天线,在相同的频率、时间资源内建立多个并行的数据传输通道,进而提升整个系统的频谱利用效率和能量效率,同时大幅地提高数据的传输速率。
大规模MIMO系统一般采用基于训练序列的信道估计方法。大部分的大规模MIMO的研究都假设系统工作于时分双工(Time Division Duplex,TDD)模式下,而非频分双工(Frequency Division Duplex,FDD)模式,其原因是FDD模式下,信道状态信息(Channel State Information,CSI)获取较为困难。首先,由于无线传输信道的相干时间有限,在FDD大规模MIMO系统中,当散射较丰富时,如果信道服从独立同分布的平坦瑞利衰落,训练长度等于发送天线数的正交训练序列可以实现估计性能和数据速率之间的最优折中。由于基站侧部署了大量的发送天线,如果最优的训练序列长度等于发送天线数目,这将占用大量珍贵的下行链路资源。如果基站发送天线数量过大,训练序列长度甚至会超过相干时间,导致基于训练序列的信道估计算法失效。其次,每个用户在估计信道后需向基站反馈CSI,巨大的信道向量的维度会占用大量的上行链路资源。而假设大规模MIMO系统工作于TDD模式下的好处在于,基站能够用TDD系统上下行链路信道的互易性获取下行链路的CSI。在这种情况下,训练序列的长度由用户总数目决定,因此基站的天线数量可以足够大。然而,由于下行和上行射频链路存在差异,上下行链路信道的互易性假设并不严格成立。因此在实际通信系统中,需要采取互易性校准技术以确保得到的下行链路CSI的精确性。
绝大部分大规模MIMO技术相关研究都是基于TDD系统的。然而基于FDD的蜂窝网络是当今的主流,因此研究FDD大规模MIMO系统的CSI获取问题具有重要的实际应用价值。目前已有研究人员提出FDD大规模MIMO系统的信道估计技术,如有限反馈,压缩感知和信道投影等。还有一些文献在闭环训练的假设下,提出大规模MIMO的导频波束设计,以提高信道估计性能。基于此,本发明提出了一种同时利用信道的空间和时间相关性的训练序列设计方案,以解决信道估计的准确性及训练开销的问题。
发明内容
本发明提出一种FDD大规模MIMO系统的下行链路训练序列设计方法。FDD大规模MIMO系统中,配备有大规模天线阵列的基站与单天线用户终端进行通信,下行链路信道状态信息的获得是一个重要的问题。下行链路的信道估计采用基于训练序列的方法,这就要求得到准确性和时间开销方面都具有优势的训练序列。本发明采用基于卡尔曼滤波的连续信道估计,同时利用了信道的时间及空间相关特性。在每个相干时间内,用户进行卡尔曼滤波和预测,得到当前相干时间的信道状态信息估计值,同时给出下一个相干时间的信道预测。在卡尔曼滤波和预测的过程中产生的一些中间量,即一些必要的统计信息,由用户通过 上行链路将必要的统计信息反馈回基站,以便于基站进行最优的训练序列设计。通过仿真分析可以看出,一方面,本发明能够提高信道估计的准确性;另一方面,可以大大减少FDD大规模MIMO系统的下行链路训练时长,因而更多的时间用于传输有用信号,从而极大的提高了链路吞吐率与系统容量。
在一个无线通信蜂窝内,基站配置有Nt个天线,对于大规模MIMO系统,Nt可以取非常大的值,小区内用户均为配置单根天线的终端,基站在同一时频资源上与多个用户进行无线通信。针对基站和一个单天线用户进行下行链路数据传输的场景,此时可以视为一个多输入单输出(MISO)系统,将下行信道建模为一个离散时间块衰落信道且定义相干时间长度为Tc(单位为channel use,符号时间),即,在一个相干时间内信道状态信息(CSI)保持不变,一直持续到下一个相干时间。在第i个相干时间的第k个符号时间内,系统的输入输出函数表示为:
Figure PCTCN2016085000-appb-000001
其中,yi(k)为当前接收的信号,
Figure PCTCN2016085000-appb-000002
为第i个相干时间内的信道向量,
Figure PCTCN2016085000-appb-000003
为当前发送的信号,ni(k)满足标准复高斯分布(即ni(k)~CN(0,1))代表当前时刻的加性高斯白噪声(AWGN)。
对于下行链路的信道估计,可以采用基于训练序列的方法,并设计正交训练序列(训练序列长度与基站天线数量相等),然而,这会导致不可接受的时间开销。分析此MISO系统可知:一方面,由于基站天线数量较大,多个天线和用户间的信道不会是完全独立的,而是会存在空间上的相关性;另一方面,由于用户不会以极高的速度在小区内移动,则地理场景不会发生急剧的变化,因此信道在连续两个相干时间内也不是完全独立的。利用一阶马尔可夫模型,信道可以写为如下形式:
Figure PCTCN2016085000-appb-000004
式中η代表时间相关系数,gi为一组Nt维列向量代表一个复高斯过程,它的每个元素都服从标准复高斯分布,即
Figure PCTCN2016085000-appb-000005
且对于不同的相干时间gi是完全独立的。式中的矩阵Rh为信道空间相关矩阵,且可以建模为一个指数模型,即
Figure PCTCN2016085000-appb-000006
其中参数a(0≤a≤1)为信道的空间相关系数,a越大代表信道间的相关性越强。
大规模MIMO系统的帧结构如图1所示,其中,相干时间Tc分为三部分:第一部分为信道训练阶段,其时间长度记为Tt;第二部分为信道状态信息(CSI)反馈阶段;第三部分为有用信息传输阶段,其时间长度记为Td。其中为了分析的简便,CSI反馈阶段的时间长度可忽略不计。在训练阶段,用户接收到的Tt维训练信号可表示为:
yi,t=Xi,t Hhi+ni,t,(4)其中yi,t=[yi(1),…,yi(Tt)]H表示接受的训练信号,Xi,t=[xi(1),…,xi(Tt)]为基站发送的训练序列,它是一个Nt×Tt维的矩阵,且满足能量约束tr(Xi,t HXi,t)=ρTt。ni,t=[ni(1),…,ni(Tt)]H为加性高斯白噪声,是一个Tt维列向量。
根据公式(2)描述的信道状态方程,以及公式(4)描述的训练阶段系统输入输出关系,基于卡尔曼滤波的信道估计状态方程表示如下:
Figure PCTCN2016085000-appb-000007
Figure PCTCN2016085000-appb-000008
Figure PCTCN2016085000-appb-000009
Ri+1|i=η2Ri|i+(1-η2)Rh,       (8)公式
(6)中,
Figure PCTCN2016085000-appb-000010
表示卡尔曼增益矩阵。卡尔曼滤波包含估计和预测两个部分,其中公式(5)代表第i个相干时间的信道估计过程,公式(6)中的Ri|i表示相应的估计误差协方差矩阵;公式(7)表示该想干时间内的信道预测过程,Ri+1|i为相应的预测误差协方差矩阵。卡尔曼滤波的初值设计为:
Figure PCTCN2016085000-appb-000011
以及R0|-1=Rh=E[h0h0 H]。
当信道估计在用户端完成并通过上行链路反馈给基站后,基站想用户发送有用的数据信号,并对发送的数据进行预处理。此时基站端发出的信号可以表示为:
xi,d(k)=wisi(k),     (9)
式中
Figure PCTCN2016085000-appb-000012
表示第i个相干时间的预处理矩阵,且满足||wi||=1;si(k)表示第i个相干时间的第k个符号时间内的有用信号,且满足E[|si(k)|2]=ρd。因而在有用数据传输阶段用户端的归一化平均接收信噪比为:
Figure PCTCN2016085000-appb-000013
本发明的重点在于FDD大规模MIMO系统下行链路的训练序列设计,而训练序列的设计基于最小化估计均方误差(MMSE)准则,且在设计中需要的统计信息由用户端在进行估计时计算得到,因而信道估计采用闭环训练模式。
闭环训练的模型如图2所示:在一个相干时间内的训练阶段,基站向用户发送训练序列Xi,t;用户接收到相应的训练信号后,利用卡尔曼滤波得到当前相干时间的信道估计值,同时对下一个相干时间的信道进行预测;最后将计算得到的信道预测误差协方差矩阵Ri+1|i反馈给基站,以便基站设计下一个相干时间的最佳训练序列。
分析卡尔曼滤波的状态方程可知,在第i个相干时间内,信道估计的均方误差可写为:
Figure PCTCN2016085000-appb-000014
其中,训练序列Xi,t是待定量,Ri|i-1表示上一相干时间到当前相干时间的信道预测值。
根据优化目标,同时结合训练序列的能量约束条件可以将此优化问题写为:
Figure PCTCN2016085000-appb-000015
式中Ri|i可以进行进一步的简化得到:
Figure PCTCN2016085000-appb-000016
为了解决该优化问题,将目标函数继续进行推导转化,即:
Figure PCTCN2016085000-appb-000017
其中第一步推导,即步骤(a)是由Ri|i的特征分解得出,且满足
Figure PCTCN2016085000-appb-000018
式中Λi代表Ri|i的特征值组成的对角阵,即
Figure PCTCN2016085000-appb-000019
其中特征值
Figure PCTCN2016085000-appb-000020
是按数值大小降序排列的;相应的,Ui是与l个特征值相对应的特征向量组成的矩阵。
分析可知,当公式(14)中的矩阵
Figure PCTCN2016085000-appb-000021
为一个对角阵时,可以使得估计均方误差达到最小值。记
Figure PCTCN2016085000-appb-000022
其中Γi是一个待定对角矩阵,因而最佳训练序列的结构可以由以下推导得出:
Figure PCTCN2016085000-appb-000023
式中
Figure PCTCN2016085000-appb-000024
代表Γi的Cholesky分解。
根据上文的推导,优化问题可以重新写为以下形式:
Figure PCTCN2016085000-appb-000025
式中γi,l(l=1,2,…,Nt)是对角矩阵Γi的对角线元素。为了求解这一组待定的γi,l,可采用拉格朗日乘子法,由优化问题写出的拉格朗日函数为:
Figure PCTCN2016085000-appb-000026
将拉格朗日函数L(μ,γi,l)对变量γi,l分别求偏导,且令其值为0,此时得到的γi,l的值为:
Figure PCTCN2016085000-appb-000027
从公式(18)中可以看出,最优的γi,l值是由拉格朗日乘子μ确定的。利用拉格朗日乘子法中的快速迭代算法,μ的值可以算出,相应的γi,l的值也随之确定。乘子μ的迭代由下式给出:
Figure PCTCN2016085000-appb-000028
其中δ表示迭代过程中的步长,该值决定了μ的迭代速度;此外,μ的初值为:
Figure PCTCN2016085000-appb-000029
μ0的得出是将公式(18)带入约束条件
Figure PCTCN2016085000-appb-000030
中计算得到的。
从上述的分析中可以看出,第i个相干时间的最佳训练序列Xi,t,opt是由前一个相干时间的预测误差协方差矩阵Ri|i-1的特征向量确定的,因此Xi,t,opt的各列之间是相互正交的。此外,Xi,t,opt的每一列所乘的功率因子,是由注水算法得出,具体的说是针对Ri|i-1的特征值进行注水。
附图说明
图1为大规模MIMO系统帧结构示意图;
图2为下行链路闭环信道估计概念示意图;
图3为大规模MIMO系统下行链路中,三种不同的信道估计方法对应的估计归一化均方误差以及不同的信道空间相关系数对该值的影响;
图4为大规模MIMO系统下行链路中,三种不同的信道估计方法对应的归一化平均 接收信噪比以及不同的信道空间相关系数对该值的影响;
图5为大规模MIMO系统下行链路中,本发明中提出的信道估计方法的归一化平均接收信噪比以及不同的训练序列发送信噪比和用户移动速度对该值的影响;
图6为大规模MIMO系统下行链路中,本发明中提出的信道估计方法的归一化平均接收信噪比随基站天线数的变化以及不同的训练序列长度该值的影响。
具体实施方式
以下结合具体实施例进一步阐述本发明,应理解这些实施例仅用于说明本发明而不用于限制本发明的范围,本发明的保护范围不限于下述实施例。在阅读了本发明之后,本领域技术人员对于本发明的各种等价形式的修改均落于本申请所附权利要求所限定的范围。
为了通过仿真评估对比本发明的下行链路训练方案以及训练序列设计方法与仅考虑单个相干时间的信道估计的性能,采用蒙特卡洛方法进行仿真得到前十个相干时间的信道估计均方误差以及归一化平均接收信噪比。考虑信道在空间和时间上都是相关的,其中空间相关系数a人为给定;另一方面,当用户移动速度v=3km/h时,由Jacks模型可以得到时间相关系数η=0.9881。作为对比,另外一种基于卡尔曼滤波的闭环信道训练方法(先确定训练序列集合,用户在该集合中选择最佳训练序列并将其序号反馈回基站)的信道估计均方误差以及归一化平均接收信噪比也由仿真给出。
实施例1:一种单个相干时间内的信道训练场景
大规模MIMO下行链路训练中,仅考虑基站和用户信道在空间上的相关性而忽略其时间上的相关性,简称该方法为single-shot训练。在每个相干时间内进行信道估计时,采用传统的MMSE信道估计方法;同时,最佳训练序列的设计也满足最小化均方误差准则。基站天线数Nt设置为64,训练序列的发送信噪比ρ=20dB,训练序列时间长度Tt=2,用户移动速度v=3km/h,在信道空间相关系数a=0.6和0.9时,相应的single-shot训练的均方误差以及归一化接收信噪比如图3和图4所示。由图可知,由于没有采用卡尔曼,不同的相干时间内的信道估计不会彼此影响,因此,信道估计的均方误差以及相应的归一化接收信噪比不会随时间变化,表现在图中是一条水平线。另外,当相关系数a增大时,估计的均方误差减小,对应的归一化接受信噪比增大,这说明,信道在空间上越相关,在相同条件下,估计的准确性越高。
实施例2:一种预设训练序列集合的基于卡尔曼滤波的闭环信道估计场景
作为对比,另外一种基于卡尔曼滤波的闭环信道训练方法的信道估计均方误差以及归一化平均接收信噪比也由仿真给出。为区别于本发明提出的训练模式及训练序列设计方法,称其为有备选集合的闭环信道估计方法。该方法与本发明的不同之处在于,基站根据某种准则,预先准备好一个训练序列结合
Figure PCTCN2016085000-appb-000031
其中包含2B个备选的训练序列,B为反馈长度。该方法采用基于卡尔曼滤波的信道估计方法,在每个相干时间内用户进行信道的估计和预测,并基于最小化估计均方误差或最大化接收信噪比的准则在集合
Figure PCTCN2016085000-appb-000032
中选取最佳的训练序列,并将B(bit)序号反馈回基站。基站天线数Nt设置为64,训练序列的发送信噪比ρ=20dB,训练序列时间长度Tt=2,用户移动速度v=3km/h,信道空间相关系数取a=0.6和 0.9。从图3和图4中看出,相比于single-shot训练,该方法的估计准确性更高,且随着卡尔曼滤波的收敛,估计准确性进一步的提高。这是由于该方法考虑了信道在时间上的相关性,同时,用户采用卡尔曼滤波的方法,不仅考虑当前相干时间接收到的训练序列,还考虑了之前的接收值,因此,随着迭代次数增加,卡尔曼滤波的准确性也随之增加。
实施例3:一种训练序列实时设计的基于卡尔曼滤波的闭环信道估计场景
与例2中方法不同,本发明提出的训练序列设计是实时的,即每个相干时间内,都要设计出下一相干时间的最佳训练序列。训练序列的设计由基站来完成,用户负责将训练序列设计所必要的信息(如信道预测误差协方差矩阵Ri|i-1)反馈回基站,因此这也是一个闭环的设计方法。基站天线数Nt设置为64,训练序列的发送信噪比ρ=20dB,训练序列时间长度Tt=2,信道空间相关系数取a=0.6和0.9。图3和图4是三个例子中的估计方法性能对比,从图中看出本例的方法比例2的设计方法准确性更高,这是由于训练序列设计时进行了功率分配,对信噪比大的信道分配了更高的功率。
图5描述了本发明的信道估计方法在不同的信噪比以及不同的用户移动速率时归一化平均接收信噪比的值。此处设定空间相关系数取a=0.6,当v=10km/h时,η=0.8721,其他条件相同。从图中可以看出,当用户移动速度增大时,时间相关系数η减小,即时间维度上的相关性减弱,此时信道估计的准确性也随之降低。当训练序列的功率提高(发送信噪比变大)时,信道估计的准确性也会相应的有所提高,但表现在图中并不明显。这是由于本例的信道估计方法存在一种称为饱和效应的局限,即接收信噪比的上界仅受空间相关系数a影响,a确定后上界也随之确定,因此无论发送信噪比如何增大,归一化接收信噪比也不会出现极其明显的增大。
图6描述了本发明的信道估计方法在第九个相干时间内的归一化接受信噪比随基站天线数的变化曲线。训练序列的发送信噪比设定为ρ=20dB,用户移动速度v=3km/h,信道空间相关系数取a=0.9。由于基站天线数Nt的变化会对信道预测误差协方差矩阵Ri|i-1的特征值产生影响,从而影响到对训练序列功率的分配,因此归一化平均接收信噪比会随着Nt的增大而增大。同时,当训练序列时间长度Tt增大时,信道估计的准确性也会随之提高。

Claims (7)

  1. 一种FDD大规模MIMO系统下行链路训练序列设计方法,其特征在于:
    (1)在一个小区内,基站配备由Nt根天线组成的大规模天线阵列,小区内用户均为配置单根天线的终端,基站在同一时频资源上与多个用户进行无线通信,其中,Nt可以取较大的值,如64、128或256;
    (2)信道在空间上的相关性可以由一个空间相关矩阵Rh表示,且Rh=E[hihi H],同时考虑信道在时间上的相关性,将信道建模为离散一阶马尔可夫模型;
    (3)下行链路训练阶段,用户端的接收信号为yi,t=Xi,t Hhi+ni,t,其中Xi,t是基站发送得训练序列,且满足能量约束:tr(Xi,t HXi,t)=ρTt,hi为当前相干时间的信道矩阵,ni,t为加性高斯白噪声;
    (4)用户根据接收到的训练信号,利用卡尔曼滤波方法进行信道估计和预测,得到估计值
    Figure PCTCN2016085000-appb-100001
    预测值
    Figure PCTCN2016085000-appb-100002
    以及相应的误差协方差矩阵Ri|i和Ri+1|i,根据基站设计最佳训练序列的需求,用户将Ri+1|i反馈回基站;
    (5)基站根据最小化信道估计均方误差的准则设计最优训练序列Xi,t.opt,优化目标函数可写为:
    Figure PCTCN2016085000-appb-100003
    Figure PCTCN2016085000-appb-100004
    其中约束条件是根据训练序列的能量约束写出的;
    (6)基站遵循最优训练序列的设计准则及能量约束,在设计过程中,对用户反馈回的信息Ri+1|i进行处理,同时,在保证总能量ρTt不变的前提下,对每个符号时间的功率进行调整,得到最佳的训练序列,用于下一个相干时间的信道估计。
  2. 根据权利要求1所述的一种FDD大规模MIMO系统下行链路训练序列设计方法,其特征在于:将下行链路信道建模为离散时间块衰落模型,即信道是时变的,但在系统的相干时间Tc内,信道保持不变;一个相干时间内,一部分时间Tt用于信道估计,一部分时间Td用于反馈必要的信息,剩余的时间进行有用信号的传输。
  3. 根据权利要求1所述的一种FDD大规模MIMO系统下行链路训练序列设计方法,其特征在于:对于信道的空间相关性,利用信道矩阵h的自协方差矩阵Rh来表示:
    Figure PCTCN2016085000-appb-100005
    其中a表示空间相关系数;定义时间相关系数为η,结合信道的空间和时间相关特性,将信道建模为一个离散一阶马尔可夫模型:
    Figure PCTCN2016085000-appb-100006
  4. 根据权利要求1所述的一种FDD大规模MIMO系统下行链路训练序列设计方法,其特征在于:利用卡尔曼滤波方法给出每个相干时间内信道的估计值及预测值,其中卡尔曼滤 波的状态方程为:
    Figure PCTCN2016085000-appb-100007
    Figure PCTCN2016085000-appb-100008
    Figure PCTCN2016085000-appb-100009
    Figure PCTCN2016085000-appb-100010
    Ri+1|i=η2Ri|i+(1-η2)Rh,
    其中Ki表示卡尔曼增益矩阵。
  5. 根据权利要求1所述的一种FDD大规模MIMO系统下行链路训练序列设计方法,其特征在于:基站设计最优训练序列需利用卡尔曼滤波过程中计算得出的信道预测的误差协方差矩阵Ri+1|i,用户通过上行链路将一些必要的统计量反馈回基站,即闭环的估计方法。
  6. 根据权利要求1所述的一种FDD大规模MIMO系统下行链路训练序列设计方法,其特征在于:基站将Ri+1|i进行特征分解,选取前Tt个最大特征值对应的特征向量,并根据Xi,t满足的能量约束条件,对每个符号时间的训练序列进行功率调整,得到最优训练序列。
  7. 根据权利要求1所述的一种FDD大规模MIMO系统下行链路训练序列设计方法,其特征在于:设计出的最佳训练序列的时间长度不需要与发送天线个数相同,而是远小于发送天线数,即Tt<<Nt
PCT/CN2016/085000 2015-07-01 2016-06-06 一种fdd大规模mimo系统下行链路训练序列设计方法 Ceased WO2017000752A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510376921.2A CN105049383A (zh) 2015-07-01 2015-07-01 一种fdd大规模mimo系统下行链路训练序列设计方法
CN201510376921.2 2015-07-01

Publications (1)

Publication Number Publication Date
WO2017000752A1 true WO2017000752A1 (zh) 2017-01-05

Family

ID=54455597

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/085000 Ceased WO2017000752A1 (zh) 2015-07-01 2016-06-06 一种fdd大规模mimo系统下行链路训练序列设计方法

Country Status (2)

Country Link
CN (1) CN105049383A (zh)
WO (1) WO2017000752A1 (zh)

Cited By (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108199987A (zh) * 2017-12-21 2018-06-22 中国人民解放军空军预警学院 一种fdd大规模mimo系统中的下行链路信道估计方法
US10530610B2 (en) * 2018-01-03 2020-01-07 Postech Academy-Industry Foundation Method and apparatus for estimating channel in multiple-input multiple-output communication systems exploiting temporal correlations
CN110740005A (zh) * 2019-10-25 2020-01-31 网络通信与安全紫金山实验室 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
CN111435926A (zh) * 2019-01-11 2020-07-21 中国移动通信有限公司研究院 一种mimo系统信道预测方法、装置、介质和设备
CN113595666A (zh) * 2021-07-13 2021-11-02 华中科技大学 大规模mimo系统基于时频联合相关性的信道预测方法和系统
CN113709070A (zh) * 2021-10-29 2021-11-26 南通先进通信技术研究院有限公司 智能反射面辅助通信系统中降低训练开销的模式切换方法
CN114143149A (zh) * 2021-11-26 2022-03-04 西北大学 一种基于神经网络的LoRa信道估计方法及系统
CN114143146A (zh) * 2021-10-28 2022-03-04 西安交通大学 一种基于图信号方法的ofdm系统信道估计系统和方法
CN114338303A (zh) * 2021-12-24 2022-04-12 杭州电子科技大学 大规模MIMO系统中基于多维Hankel矩阵的信道估计方法及系统
CN114598338A (zh) * 2022-03-16 2022-06-07 北京航空航天大学 低复杂度通信系统上行链路的多用户接入方法及存储介质
CN114665930A (zh) * 2022-03-16 2022-06-24 南京邮电大学 去蜂窝大规模mimo系统的下行链路盲信道估计方法
CN114844538A (zh) * 2022-04-29 2022-08-02 东南大学 一种基于宽学习的毫米波mimo用户增量协作波束选择方法
CN116016052A (zh) * 2023-01-04 2023-04-25 西南交通大学 一种用于毫米波大规模mimo系统的信道估计方法
CN116112043A (zh) * 2022-12-30 2023-05-12 东莞理工学院 基于ris辅助基站与用户之间的近场通信方法及系统
CN116471150A (zh) * 2023-04-23 2023-07-21 中电科普天科技股份有限公司 一种mimo系统信道估计方法、装置、设备及介质
CN116707687A (zh) * 2023-06-27 2023-09-05 南京盛航海运股份有限公司 一种去蜂窝大规模mimo系统的信道预测方法
CN119172205A (zh) * 2024-08-05 2024-12-20 南京大学 适用于FDD Massive MIMO系统的自适应导频发送方法
CN119439095A (zh) * 2024-10-11 2025-02-14 上海交通大学 基于多视角天线分布设计和卷积处理的雷达原域目标识别系统

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105049383A (zh) * 2015-07-01 2015-11-11 东南大学 一种fdd大规模mimo系统下行链路训练序列设计方法
CN106028462B (zh) * 2016-05-23 2019-04-23 西安电子科技大学 大规模mimo协作式用户调度方法
CN110383719B (zh) * 2017-03-16 2021-08-13 华为技术有限公司 下行链路通道状态信息估计的硬件损伤参数确定
CN107070823A (zh) * 2017-05-12 2017-08-18 重庆大学 基于卡尔曼滤波的参数模型信道估计方法
CN108712197B (zh) * 2018-05-14 2020-06-19 西安交通大学 一种fdd大规模mimo系统中基于移动速度的反馈量选择方法
CN111666247B (zh) * 2020-05-26 2022-02-08 中国人民解放军国防科技大学 一种支持链路物理层串行信号的分级训练方法及装置
CN113364501B (zh) * 2021-06-04 2022-07-05 大连海事大学 一种莱斯信道下基于低精度adc去蜂窝大规模mimo系统的功率控制方法
CN119814503B (zh) * 2025-02-27 2026-01-30 西安电子科技大学 一种ris辅助系统的分段信道估计方法、系统、设备及介质

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102916735A (zh) * 2012-10-17 2013-02-06 东南大学 利用大规模天线阵列的无线通信方法
WO2014154293A1 (en) * 2013-03-28 2014-10-02 Nokia Solutions And Networks Oy Channel estimation in wireless communications
CN105049383A (zh) * 2015-07-01 2015-11-11 东南大学 一种fdd大规模mimo系统下行链路训练序列设计方法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101867402B (zh) * 2010-05-04 2014-01-29 西安交通大学 一种自适应天线选择的mimo系统及其应用方法
US9106280B2 (en) * 2013-04-15 2015-08-11 Broadcom Corporation Pilot design for massive MIMO communication
US20150163073A1 (en) * 2013-12-11 2015-06-11 Broadcom Corporation Massive mimo channel estimation
CN103856422B (zh) * 2014-03-17 2017-10-17 清华大学 一种多载波Large‑Scale MIMO系统的发射信号配置及信道估计的方法和设备

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102916735A (zh) * 2012-10-17 2013-02-06 东南大学 利用大规模天线阵列的无线通信方法
WO2014154293A1 (en) * 2013-03-28 2014-10-02 Nokia Solutions And Networks Oy Channel estimation in wireless communications
CN105049383A (zh) * 2015-07-01 2015-11-11 东南大学 一种fdd大规模mimo系统下行链路训练序列设计方法

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
NOH SONG ET AL.: "Pilot Beam Pattern Design for Channel Estimation in Massive MIMO Systems", IEEE JOURNAL OF SELECTED TOPICS IN SIGNAL PROCESSING, vol. 8, no. 5, 30 May 2014 (2014-05-30), XP011558887, ISSN: 1932-4553 *
WANG DONG ET AL.: "Design of Downlink Training Sequences for FDD Massive MIMO systems", IEEE INTERNATIONAL CONFERENCE ON COMMUNICATIONS (ICC, 12 June 2015 (2015-06-12), pages 4570 - 4575, XP055342520, ISSN: 1550-3607 *

Cited By (30)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108199987A (zh) * 2017-12-21 2018-06-22 中国人民解放军空军预警学院 一种fdd大规模mimo系统中的下行链路信道估计方法
CN108199987B (zh) * 2017-12-21 2020-08-18 中国人民解放军空军预警学院 一种fdd大规模mimo系统中的下行链路信道估计方法
US10530610B2 (en) * 2018-01-03 2020-01-07 Postech Academy-Industry Foundation Method and apparatus for estimating channel in multiple-input multiple-output communication systems exploiting temporal correlations
CN111435926A (zh) * 2019-01-11 2020-07-21 中国移动通信有限公司研究院 一种mimo系统信道预测方法、装置、介质和设备
CN110740005A (zh) * 2019-10-25 2020-01-31 网络通信与安全紫金山实验室 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
CN110740005B (zh) * 2019-10-25 2021-12-24 网络通信与安全紫金山实验室 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
CN113595666A (zh) * 2021-07-13 2021-11-02 华中科技大学 大规模mimo系统基于时频联合相关性的信道预测方法和系统
CN113595666B (zh) * 2021-07-13 2022-05-20 华中科技大学 大规模mimo系统基于时频联合相关性的信道预测方法和系统
CN114143146B (zh) * 2021-10-28 2023-08-15 西安交通大学 一种基于图信号方法的ofdm系统信道估计系统和方法
CN114143146A (zh) * 2021-10-28 2022-03-04 西安交通大学 一种基于图信号方法的ofdm系统信道估计系统和方法
CN113709070A (zh) * 2021-10-29 2021-11-26 南通先进通信技术研究院有限公司 智能反射面辅助通信系统中降低训练开销的模式切换方法
US11848726B2 (en) 2021-10-29 2023-12-19 Nantong Research Inst For Adv Comm Tech Co., Ltd. Mode switching method for reducing training overheads in reconfigurable intelligent surface (RIS)-assisted communication system
CN114143149A (zh) * 2021-11-26 2022-03-04 西北大学 一种基于神经网络的LoRa信道估计方法及系统
CN114143149B (zh) * 2021-11-26 2023-03-10 西北大学 一种基于神经网络的LoRa信道估计方法及系统
CN114338303B (zh) * 2021-12-24 2024-02-13 杭州电子科技大学 大规模MIMO系统中基于多维Hankel矩阵的信道估计方法及系统
CN114338303A (zh) * 2021-12-24 2022-04-12 杭州电子科技大学 大规模MIMO系统中基于多维Hankel矩阵的信道估计方法及系统
CN114665930A (zh) * 2022-03-16 2022-06-24 南京邮电大学 去蜂窝大规模mimo系统的下行链路盲信道估计方法
CN114598338A (zh) * 2022-03-16 2022-06-07 北京航空航天大学 低复杂度通信系统上行链路的多用户接入方法及存储介质
CN114665930B (zh) * 2022-03-16 2023-07-18 南京邮电大学 去蜂窝大规模mimo系统的下行链路盲信道估计方法
CN114844538A (zh) * 2022-04-29 2022-08-02 东南大学 一种基于宽学习的毫米波mimo用户增量协作波束选择方法
CN116112043A (zh) * 2022-12-30 2023-05-12 东莞理工学院 基于ris辅助基站与用户之间的近场通信方法及系统
CN116112043B (zh) * 2022-12-30 2024-04-30 东莞理工学院 基于ris辅助基站与用户之间的近场通信方法及系统
CN116016052A (zh) * 2023-01-04 2023-04-25 西南交通大学 一种用于毫米波大规模mimo系统的信道估计方法
CN116016052B (zh) * 2023-01-04 2024-05-07 西南交通大学 一种用于毫米波大规模mimo系统的信道估计方法
CN116471150A (zh) * 2023-04-23 2023-07-21 中电科普天科技股份有限公司 一种mimo系统信道估计方法、装置、设备及介质
CN116707687A (zh) * 2023-06-27 2023-09-05 南京盛航海运股份有限公司 一种去蜂窝大规模mimo系统的信道预测方法
CN116707687B (zh) * 2023-06-27 2024-05-10 南京盛航海运股份有限公司 一种去蜂窝大规模mimo系统的信道预测方法
CN119172205A (zh) * 2024-08-05 2024-12-20 南京大学 适用于FDD Massive MIMO系统的自适应导频发送方法
CN119172205B (zh) * 2024-08-05 2025-11-07 南京大学 适用于FDD Massive MIMO系统的自适应导频发送方法
CN119439095A (zh) * 2024-10-11 2025-02-14 上海交通大学 基于多视角天线分布设计和卷积处理的雷达原域目标识别系统

Also Published As

Publication number Publication date
CN105049383A (zh) 2015-11-11

Similar Documents

Publication Publication Date Title
WO2017000752A1 (zh) 一种fdd大规模mimo系统下行链路训练序列设计方法
US8848822B2 (en) Feedback method for interference alignment in wireless network
CN101873159B (zh) 一种多输入多输出下行传输控制方法及装置
CN108234101B (zh) 能效最大化导频信号设计方法及大规模多天线系统
Blum MIMO with limited feedback of channel state information
CN105338609B (zh) 多天线系统高能效动态功率分配方法
CN106506052B (zh) 一种在大规模mimo系统中基于天线选择提高通信能效的方法
CN101110623B (zh) 一种有限反馈多入多出多用户系统的部分用户反馈方法
CN105246142B (zh) 基于能效最优的单用户大规模天线中继系统功率分配方法
CN101159462A (zh) 一种多天线多小区系统中的有限反馈预编码干扰抑制方法
CN105049100B (zh) 一种多小区mimo系统双层预编码方法
CN110708716B (zh) 一种基于多天线能量捕获的短包通信传输性能分析方法
KR20110084818A (ko) 다중 셀 다중 사용자 기반의 프리코딩 방법 및 통신 장치
CN107171704B (zh) 一种大规模mimo系统的上行功率控制方法及装置
CN103595455B (zh) 基于用户满意度的lte‑a非码本波束赋形方法
CN104954054B (zh) 一种解决c-ran架构下多小区系统边缘用户干扰的消除方法
Hammarwall et al. Utilizing the spatial information provided by channel norm feedback in SDMA systems
CN102447500B (zh) 一种天线控制装置、便携式电子设备及天线控制方法
CN118661392A (zh) 用于优化波束成形的发射器和接收器之间的通信方法、对应的发射器、接收器和计算机程序
CN113258985A (zh) 单站多星mimo上注系统能效优化方法
CN107579762B (zh) 一种基于量化和统计信道信息的多小区协作预编码方法
CN102742175A (zh) 一种mimo模式配置方法及通信设备
Li et al. Outage probability analysis for two-antennas MISO-NOMA downlink with statistical CSI
CN110445519B (zh) 基于信干噪比约束的抗组间干扰方法及装置
KR101768362B1 (ko) 대규모 안테나를 이용하는 다중 사용자 무선 통신 시스템에서 채널 정보 추정 방법

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: 16817116

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16817116

Country of ref document: EP

Kind code of ref document: A1