MULTI-USER DETECTION METHOD AND APPARATUS FOR CELL-FREE MIMO

Abstract

Embodiments of the present disclosure provide a multi-user detection method and apparatus for cell-free Multiple-Input Multiple-Output (MIMO). The method includes: receiving, by a Central Processing Unit (CPU) of a cell-free MIMO system, an MRC merged data symbol stream SMRC=HHY of K users transmitted by an Access Point (AP), and multiplying SMRC by a conjugate transpose matrix SMRCH of SMRC to obtain a K*K matrix SMRCSMRCH; and performing singular value decomposition on the matrix SMRCSMRCH to obtain a unitary matrix V and a diagonal matrix A, obtaining a diagonal matrix A according to the diagonal matrix A, and estimating sending-end data symbols of the K users by a formula {tilde over (S)}=VΛVHSMRC.

Description

TECHNICAL FIELD

Embodiments of the present disclosure relate to the field of communication, and more particularly relate to a multi-user detection method and apparatus for cell-free Multiple-Input Multiple-Output (MIMO).

BACKGROUND

When a plurality of pieces of User Equipment (UE) simultaneously transmit data to a cell-free MIMO network, a Central Processing Unit (CPU) in a cell-free MIMO system is required to perform a multi-user detection. The CPU of the cell-free MIMO system was initially designed to perform the multi-user detection based on Maximal Ratio Combining (MRC), because the MRC is the simplest and does not require increasing a fronthaul bandwidth from an Access Point (AP) in the cell-free MIMO system to the CPU. However, the MRC-based multi-user detection usually does not have optimal performance due to neglect of interference between users. Particularly, as the number of simultaneously accessing users is increased, the performance of the MRC-based multi-user detection degrades. Thus, a multi-user detection method having better performance is required to be designed for the cell-free MIMO system. However, in order to obtain a multi-user detection better than the MRC-based multi-user detection for the current cell-free MIMO system, APs are required to transmit to the CPU the channel information of users having access to the APs, so that the CPU can consider, based on the received channel information, relevance of spatial domain channels of the users, thereby performing multi-user detection that can better restrain multi-user mutual interference. Some examples of such multi-user detection with performance better than the MRC-based multi-user detection include a Zero Forcing (ZF)-based multi-user detection or a Minimum Mean Square Error (MMSE)-based multi-user detection.

Compared with the MRC-based multi-user detection, the ZF-based multi-user detection or the MMSE-based multi-user detection in the current industry has one defect: the ZF-based multi-user detection or the MMSE-based multi-user detection requires the APs to transmit to the CPU the channel information of the users having access to the APs, while the original MRC-based multi-user detection does not. The AP is required to transmit the channel information to the CPU, which undoubtedly increases the fronthaul bandwidth from the AP to the CPU, resulting in an increase in cost of the cell-free MIMO system.

SUMMARY

Embodiments of the present disclosure provide a multi-user detection method and apparatus for cell-free MIMO, which may at least solve the problems that in related technologies, during a multi-user detection, an AP is required to transmit channel information to a CPU, resulting in an increase in a fronthaul bandwidth from the AP to the CPU.

According to an embodiment of the present disclosure, a multi-user detection method for cell-free MIMO is provided and includes:

receiving, by a receiving end of a cell-free MIMO system, a data symbol stream S_MRC=H^HY of K users transmitted by an Access Point (AP), where S_MRC=H^HY denotes a K*L matrix, with each row representing L data symbols of one user, and both K and L are positive integers;

multiplying S_MRC by a conjugate transpose matrix S_MRC^H of S_MRC to obtain a K*K matrix S_MRCS_MRC^H;

performing Singular Value Decomposition (SVD) on the matrix S_MRCS_MRC^H to obtain a unitary matrix V=[v₁, v₂, . . . v_K] and K real numbers a₁, a₂. . . a_K, such that

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H=VA{V}^H,$

where V^H is a transpose matrix of the unitary matrix V , and

$A=\left[\begin{array}{cccc}{a}_1& & & \\ & a_2& & \\ & & \ddots & \\ & & & a_K\end{array}\right];$

or calculating K characteristic values a₁, a₂. . . a_K and corresponding K characteristic vectors v₁, v₂, . . . v_K of the matrix S_MRCS_MRC^H or

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H,$

such that

$\left(\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H\right)v_k=v_k{a}_k;$

obtaining a diagonal matrix where

$\Lambda =\left[\begin{array}{cccc}{\lambda }_1& & & \\ & {\lambda }_2& & \\ & & \ddots & \\ & & & {\lambda }_K\end{array}\right],$

where

${\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+{\sigma }^2}$ $\mathrm{or}{\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+c{\sigma }^2},$

c is a real number greater than 1, and σ₂ is a noise variance on received signals of the AP;

estimating, by a formula {tilde over (S)}=VΛV^HS_MRC, sending-end data symbols of the K users corresponding to the data symbol stream S_MRC.

According to an exemplary embodiment, before obtaining the diagonal matrix A, the method further includes: receiving, by the receiving end, the noise variance σ² transmitted by the AP, or determining, by the receiving-end CPU, the noise variance σ² according to an attribute of the AP.

According to an embodiment of the present disclosure, a multi-user detection method for cell-free MIMO is further provided and includes:

receiving, by a receiving end of a cell-free MIMO system, a data symbol stream S_MRC=H^HY of K users transmitted by an Access Point (AP), where S_MRC=H^HY denotes a K*L matrix, with each row representing L data symbols of one user, and both K and L are positive integers;

performing singular value decomposition on the matrix S_MRC to obtain a unitary matrix V=[v₁, v₂, . . . v_K] and K real numbers ω₁, ω₂, . . . ω_K, such that S_MRC=V(√{square root over (L)}Ω)U^H, where U^H is a transpose matrix of a unitary matrix U, and Ω is a K*L matrix:

$\Omega =\left[\begin{array}{cccccc}{\omega }_1& & & & & \\ & {\omega }_2& & & 0& \\ & & \ddots & & & \\ & & & {\omega }_K& & \end{array}\right];$

obtaining a diagonal matrix

$\Lambda =\left[\begin{array}{cccc}{\lambda }_1& & & \\ & {\lambda }_2& & \\ & & \ddots & \\ & & & {\lambda }_K\end{array}\right],$

where

${\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+{\sigma }^2}$ $\mathrm{or}{\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+c{\sigma }^2},$

c is a real number greater than 1, and σ² is a noise variance on received signals of the AP, a_k=|ω_k|²; and

estimating, by a formula {tilde over (S)}=VΛV^HS_MRC, sending-end data symbols of the K users corresponding to the data symbol stream S_MRC, where V^H is a transpose matrix of the unitary matrix V.

According to an exemplary embodiment, before obtaining the diagonal matrix A, the method further includes: receiving, by the receiving end, the noise variance σ² transmitted by the AP, or determining, by the receiving-end CPU, the noise variance σ² according to an attribute of the AP.

According to an embodiment of the present disclosure, a multi-user detection apparatus for cell-free MIMO is provided, is located in a CPU of a cell-free MIMO system, and includes:

a receiving module, configured to receive a data symbol stream S_MRC=H^HY of K users transmitted by an Access Point (AP), where S_MRC=H^HY denotes a K*L matrix, with each row representing L data symbols of one user, and both K and L are positive integers;

a first acquiring module, configured to multiply S_MRC by a conjugate transpose matrix S_MRC^H of S_MRC to obtain a K*K matrix S_MRCS_MRC^H;

a decomposition module, configured to perform singular value decomposition on the matrix S_MRCS_MRC^H to obtain a unitary matrix V=[v₁, v₂, . . . v_K] and K real numbers a₁ a₂. . . a_K, such that

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H=VA{V}^H,$

where V^H is a transpose matrix of the unitarymatrix V, and

$A=\left[\begin{array}{cccc}{a}_1& & & \\ & a_2& & \\ & & \ddots & \\ & & & a_K\end{array}\right];$

or calculate K characteristic values a₁ a₂ . . . a_K and corresponding K characteristic vectors v₁, v₂, . . . v_K of the matrix S_MRCS_MRC^H or

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H,$

such that

$\left(\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H\right)v_k=v_k{a}_k;$

a second acquiring module, configured to obtain a diagonal matrix

$\Lambda =\left[\begin{array}{cccc}{\lambda }_1& & & \\ & {\lambda }_2& & \\ & & \ddots & \\ & & & {\lambda }_K\end{array}\right],$

where

${\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+{\sigma }^2}\mathrm{or}{\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+c{\sigma }^2},$

c is a real number greater than 1, and σ² is a noise variance on received signals of the AP; and

an estimation module, configured to estimate, by a formula {tilde over (S)}=VΛV^HS_MRC, sending-end data symbols of the K users corresponding to the data symbol stream S_MRC.

According to an exemplary embodiment, the receiving module is further configured to receive the noise variance σ² transmitted by the AP, or determine the noise variance σ² according to an attribute of the AP.

According to an embodiment of the present disclosure, a multi-user detection apparatus for cell-free MIMO is further provided, is located in a CPU of a cell-free MIMO system, and includes:

a receiving module, configured to receive a data symbol stream S_MRC=H^HY of K users transmitted by an Access Point (AP), where S_MRC=H^HY denotes a K*L matrix, with each row representing L data symbols of one user, and both K and L are positive integers;

a decomposition module, configured to perform singular value decomposition on the matrix S_MRC to obtain a unitary matrix V=[v₁, v₂, . . . v_K] and K real numbers ω₁,ω₂, . . . ω_K, such that S_MRC=V (√{square root over (L)}Ω)U^H, where U^H is a transpose matrix of a unitary matrix U, and Ω is a K*L matrix:

$\Omega =\left[\begin{array}{cccccc}{\omega }_1& & & & & \\ & {\omega }_2& & & 0& \\ & & \ddots & & & \\ & & & {\omega }_K& & \end{array}\right];$

an acquiring module, configured to obtain a diagonal matrix

$\Lambda =\left[\begin{array}{cccc}{\lambda }_1& & & \\ & {\lambda }_2& & \\ & & \ddots & \\ & & & {\lambda }_K\end{array}\right],$

where

${\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+{\sigma }^2}\mathrm{or}{\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+c{\sigma }^2},$

c is a real number greater than 1, and σ² is a noise variance on received signals of the AP, a_k=|ω_k|²; and

an estimation module, configured to estimate, by a formula {tilde over (S)}=VΛV^HS_MRC, sending-end data symbols of the K users corresponding to the data symbol stream S_MRC, where V^H is a transpose matrix of the unitary matrix V.

According to an exemplary embodiment, the receiving module is further configured to receive the noise variance σ² transmitted by the AP, or determine the noise variance σ² according to an attribute of the AP.

According to another embodiment of the present disclosure, a cell-free MIMO system is further provided and includes the multi-user detection apparatus according to any of the above embodiments.

According to another embodiment of the present disclosure, a computer-readable storage medium is further provided and stores computer programs, where the computer programs, when being executed by a processor, causes the processor to implement the operations in any above method embodiment.

According to another embodiment of the present disclosure, an electronic device is further provided and includes a memory and a processor, the memory stores computer programs, and the processor is configured to run the computer programs to implement the operations in any above method embodiment.

In the above embodiments of the present disclosure, the CPU does not need to acquire channel information of every user, and can estimate the data symbols only according to the MRC merged data symbols transmitted by the AP, such that the CPU can realize an optimal MMSE-based multi-user detection without increasing a fronthaul bandwidth from the AP to the CPU, thereby realizing optimal uplink multi-user transmission of the cell-free MIMO system.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic diagram of data symbol transmission of a cell-free MIMO system according to related technologies;

FIG. 2 is a flowchart of a multi-user detection method for cell-free MIMO according to an embodiment of the present disclosure;

FIG. 3 is a flowchart of a multi-user detection method for cell-free MIMO according to another embodiment of the present disclosure;

FIG. 4 is a module structure diagram of a multi-user detection apparatus for cell-free MIMO according to an embodiment of the present disclosure;

FIG. 5 is a module structure diagram of a multi-user detection apparatus for cell-free MIMO according to another embodiment of the present disclosure;

FIG. 6 is a schematic structural diagram of a cell-free MIMO system according to an embodiment of the present disclosure;

FIG. 7 is a flowchart of a multi-user detection method for cell-free MIMO according to an embodiment 1 of the present disclosure; and

FIG. 8 is a flowchart of a multi-user detection method for cell-free MIMO according to an embodiment 2 of the present disclosure.

DETAILED DESCRIPTION

Embodiments of the present disclosure are described in detail in reference to drawings and in combination with embodiments as below.

It needs to be explained that terms such as “first” and “second” of the specification and the claims of the present disclosure and the above drawings are used to distinguish similar objects but are unnecessarily used to describe specific sequences or precedence orders.

FIG. 1 is a schematic diagram of data symbol transmission of a cell-free MIMO system according to related technologies. As shown in FIG. 1, the cell-free MIMO system includes many access points (APs), each of which is denoted by one circle in the figure, and these access points are usually deployed in a distributed manner, and are connected to a Central Processing Unit (CPU) through a certain connection manner (topology). For example, N access points shown in FIG. 1(a) are connected to the CPU in a chained or strip-like connection manner. One part of N access points shown in FIG. 1(b) are connected to the CPU in a chained or strip-like connection manner, and the other part of the N access points are connected to the CPU in another chained or strip-like connection manner.

Wireless signals of the cell-free MIMO system are transmitted and received by the AP. Uplink multi-user transmission is illustrated in FIG. 1(a). Assuming that K pieces of UE transmit data to the AP (i.e., perform uplink data transmission), and L symbols transmitted by each user pass through the same wireless channel (channels through which the symbols of the different users pass are independent of each other), then L data symbols received by the AP_m are denoted by:

$y_m=\sum _{k=1}^K{h}_{\mathrm{mk}}{s}_k+n_m,$

where y_m=[Y_m,1, y_m,2, . . . y_m,L] denotes a vector constituted by the L data symbols received by the AP_m, and is a row vector having a length of L; S_k=[S_k,1, S_k,2, S_k,L] denotes a row vector constituted by L data symbols transmitted by the UE_k; and h_mk is a scalar denoting a wireless channel from the UE_k to the AP_m, and n_m denotes Additive White Gaussian Noise (AWGN) on the AP_m, and is a row vector having a length of L.

Further, signals received by the M APs may be written in the form of a matrix:

$Y=\sum _{k=1}^K{h}_k{s}_k+N=\mathrm{HS}+N,$

where

$Y=\left[\begin{array}{c}{y}_1\\ y_2\\ ⋮\\ y_M\end{array}\right]$

where denotes an M*L matrix having M rows, and the m^th row is denoted by y_m;

$S=\left[\begin{array}{c}{s}_1\\ s_2\\ ⋮\\ s_K\end{array}\right]$

denotes an L*L matrix, and the m^th row is denoted by S_m; and

h_k=[h_1,k, h_2,k, . . . h_M,k]^T denotes a spatial domain channel vector from a user k to the M APs, which is a column vector having a length of M, i.e., the column vector has M rows. H=[h₁, h₂, . . . h_K] denotes an M*K matrix. The AWGN noise N is an M*L matrix.

The MRC-based multi-user detection process of the cell-free MIMO system is as follows.

The AP_m estimates the wireless channel h_mk from the UE_k to the AP_m according to a reference signal from the UE_k, weights the received data symbol y_m by using the conjugate h_mk^H of h_mk to obtain a data symbol h_mk^Hy_m, related to the user k, adds the data symbol h_mk^Hy_m, and a data symbol

$\sum _{j=1}^{m-1}{h}_{\mathrm{jk}}^H{y}_j$

related to the user k and transmitted by the previous AP (i.e., AP_m−1) to obtain

$\sum _{j=1}^m{h}_{\mathrm{jk}}^H{y}_j=h_{\mathrm{mk}}^H{y}_m+\sum _{j=1}^{m-1}{h}_{\mathrm{jk}}^H{y}_j,$

and then transmits the accumulated signal to a next AP (i.e., AP_m+1). In a similar way, a data symbol transmitted by the final AP, i.e., the M^th AP to the CPU and related to the user k is

$\sum _{j=1}^M{h}_{\mathrm{jk}}^{*}{y}_j.$

The signal

$\sum _{j=1}^M{h}_{\mathrm{jk}}^H{y}_j$

related to the user k in the CPU may also be denoted by h_k^HY, where h_k^H is the conjugate transpose of the vector h_k. Further,

$h_k^{H}Y=h_k^H\sum _{j=1}^K{h}_j{s}_j+h_k^{H}N={h_k}^2{s}_k+\sum _{j=1,j\ne k}^K{h}_k^H{h}_j{s}_j+h_k^{H}N.$

It is apparent that h_k^HY obtained in the CPU is the maximal ratio combining on the symbols of the user k. If the symbol s_k of the user k in h_k^HY is normalized, normalized MRC of the symbol of the user k may be obtained, and denoted by

$s_{k,\mathrm{MRC}}=\frac{1}{{h_k}^2}{h}_k^{H}Y=s_k+\frac{1}{{h_k}^2}\sum _{j=1,j\ne k}^K{h}_k^H{h}_j{s}_j+\frac{1}{{h_k}^2}{h}_k^{H}N.$

In the above-mentioned MRC receiving method, each AP only needs to transmit the K data symbol streams to the next AP without transmitting other information. For example, the AP_m only needs to transmit K data symbol streams

$\sum _{j=1}^m{h}_{\mathrm{jk}}^H{y}_j$

with a length of K to the AP_m+1, where k=1. . . K.

However, according to the above formula

$s_{k,\mathrm{MRC}}=s_k+\frac{1}{{h_k}^2}\sum _{j=1,j\ne k}^K{h}_k^H{h}_j{s}_j+\frac{1}{{h_k}^2}{h}_k^{H}N,$

the MRC does not consider interference between the users, such that there will be a significant user interference item

$\frac{1}{{h_k}^2}\sum _{j=1,j\ne k}^K{h}_k^H{h}_j{s}_j$

on the data symbol of the user k, and as a result, the performance is suboptimal.

A related MRC data symbol, received by the CPU, of the user k is h_k^HY, and MRC data symbols of the K users are combined to be written as S_MRC=H^HY.

The cell-free MIMO multi-user detection in the related art also adopts the ZF-based or MMSE-based multi-user detection with superior performance, but the APs are required to transmit to the CPU the channel information of users having access to the APs, so that the CPU can perform, according to relevance of user spatial domain channels, the multi-user detection restraining multi-user mutual interference, thereby achieving performance better than that of the MRC-based multi-user detection. Some examples of such multi-user detection with performance better than the MRC-based multi-user detection include a Zero Forcing (ZF)-based multi-user detection or a Minimum Mean Square Error (MMSE)-based multi-user detection.

Specifically, the CPU receives the MRC merged data symbols of the K users transmitted by the AP:

S _MRC =H ^H Y=H ^H(HS+N)=H^HHS+H^HN.

S_MRC=H^HY denotes a K*L matrix, with each row representing an MRC merged symbol stream of one user. If the AP transmits H to the CPU at the same time, after obtaining the H matrix, the CPU may adopt an MMSE criterion to estimate S, namely S_MMSE=(H^H+σ²I)⁻H^HY=(H^H+σ²I)⁻¹ S_MRC, and in other words, MMSE estimation is adopted for estimating S according to Y after the CPU obtains H. Herein, σ² denotes a variance of AWGN on received signals of respective APs. In scenarios for the MMSE estimation, it is assumed that the CPU knows σ².

It is apparent that compared with the MRC-based multi-user detection, the ZF-based or MMSE-based multi-user detection method in the related art has one defect: the APs are required to transmit to the CPU the channel information of the users having access to the APs, but in the original MRC-based multi-user detection, the channel information of the users is not needed, which undoubtedly increases the fronthaul bandwidth from the APs to the CPU, resulting in an increase in cost of the cell-free MIMO system. The embodiments of the present disclosure provide a brand new method through which the AP does not need to transmit to the CPU the channel information of the users having access to the AP, that is, there is no need to increase the fronthaul bandwidth from the AP to the CPU, and in other words, the CPU can realize the optimal MMSE-based multi-user detection by adopting original MRC fronthaul.

Accordingly, an embodiment of the present disclosure provides a brand new MMSE-based multi-user detection method. In the embodiment, the CPU can realize the optimal MMSE-based multi-user detection without increasing the fronthaul bandwidth from the AP to the CPU, thereby realizing optimal uplink multi-user transmission of the cell-free MIMO system.

In the embodiment of the present disclosure, under the situation that the original MRC fronthaul is adopted and the AP does not need to transmit the channel information of the users, the CPU cannot directly obtain the channel information of the users, for example, the CPU does not know H. Without the channel information, the CPU cannot estimate, according to a conventional MMSE method, the data symbol S of each user. In the embodiment of the present disclosure, the CPU only utilizes an MRC merged data symbol S_MRC=H^HY=H^H(HS+N)=H^HHS+H^HN transmitted by the AP to obtain a matrix H^HH+σ²I, and then obtains MMSE estimation on S, i.e., S_MMSE=(H^H+σ²I)⁻¹H^HY=(H^HH+σ²I)⁻¹S_MRC according to H^HH+σ²I and S_MRC.

In the embodiment of the present disclosure, it is assumed that the CPU knows the noise variance σ² on the AP. For example, the CPU may be enabled to obtain the noise variance σ² by two following manners. The first manner is to make the AP transmit Γ² or information related to σ² to the CPU. The second manner is to make the CPU determine σ² according to an attribute of the AP without requiring the AP to transmit σ² or information related to σ².

The embodiment provides a multi-user detection method for cell-free MIMO. FIG. 2 is a flowchart according to an embodiment of the present disclosure. As shown in FIG. 2, the process includes following operations S202 to S210.

At S202, a CPU of a cell-free MIMO system receives a data symbol stream S_MRC=H^HY of K users transmitted by an AP, where S_MRC=H^HY denotes a K*L matrix, with each row representing L data symbols of one user, and both K and L are positive integers.

At S204, S_MRC is multiplied by a conjugate transpose matrix S_MRC^H of S_MRC to obtain a K*K matrix S_MRCS_MRC^H.

At S206, singular value decomposition is performed on the matrix S_MRCS_MRC^H to obtain a unitary matrix V=[v₁, v₂, . . . v_K] and K real numbers a₁ a₂ . . . a_K, such that

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H=VA{V}^H,$

where V^H is a transpose matrix of the unitary matrix V, and

$A=\left[\begin{array}{cccc}{a}_1& & & \\ & a_2& & \\ & & \ddots & \\ & & & a_K\end{array}\right].$

As an alternative of this operation, K characteristic values a₁ a₂ . . . a_K and corresponding K characteristic vectors v₁, v₂, . . . v_K of the matrix S_MRCS_MRC^H or

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H$

may be calculated, such that

$\left(\frac{1}{L}{S}_{MRC}{S}_{MRC}^H\right)v_k=v_k{a}_k,$

and the K*K unitary matrix V=[v₁, v₂, . . . v_K] is constituted by the K characteristic vectors v_k.

At S208, a diagonal matrix

$\Lambda =\left[\begin{array}{cccc}{\lambda }_1& & & \\ & {\lambda }_2& & \\ & & \ddots & \\ & & & {\lambda }_K\end{array}\right]$

is obtained, where

${\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+{\sigma }^2}$ $\mathrm{or}{\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+c{\sigma }^2},$

c is a real number greater than 1, and σ² is a noise variance on received signals of the AP.

At S210, sending-end data symbols of the K users corresponding to the data symbol stream S_MRC are estimated by a formula {tilde over (S)}=VΛV^HS_MRC.

The embodiment further provides a multi-user detection method for cell-free MIMO. FIG. 3 is a flowchart according to an embodiment of the present disclosure. As shown in FIG. 3, the process includes following operations S302 to S308.

At S302, a receiving end of a cell-free MIMO system receives a data symbol stream S_MRC=H^HY of K users transmitted by an Access Point (AP), where S_MRC=H^HY denotes a K*L matrix, with each row representing L data symbols of one user, and both K and L are positive integers.

At S304, singular value decomposition is performed on the matrix S_MRC to obtain a unitary matrix V=[v₁, v₂, . . . v_K] and K real numbers ω₁, ω₂, . . . ω_K, such that S_MRC=V(√{square root over (L)}Ω)U^H, where U^H is a transpose matrix of a unitary matrix U, and Ω is a K*L matrix:

$\Omega =\left[\begin{array}{cccccc}{\omega }_1& & & & & \\ & {\omega }_2& & & 0& \\ & & \ddots & & & \\ & & & {\omega }_K& & \end{array}\right].$

At S306, a diagonal matrix

$\Lambda =\left[\begin{array}{cccc}{\lambda }_1& & & \\ & {\lambda }_2& & \\ & & \ddots & \\ & & & {\lambda }_K\end{array}\right]$

is obtained, where

${\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+{\sigma }^2}$ $\mathrm{or}{\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+c{\sigma }^2},$

c is a real number greater than 1, and σ² is a noise variance on received signals of the AP, a_k=|ω_k|².

At S308, sending-end data symbols of the K users corresponding to the data symbol stream S_MRC are estimated by a formula {tilde over (S)}=VΛV^HS_MRC, and V^H is a transpose matrix of the unitary matrix V.

Based on the description of the above implementations, those having ordinary skill in the art can clearly know that the method according to the above embodiments may be implemented by means of software and necessary universal hardware platforms and also may be implemented through hardware, but the former is the better implementation under many situations. Based on the understanding, the technical solutions of the present disclosure essentially or parts making contribution to the related art may be embodied in the form of a software product, and the computer software product is stored in a storage medium (e.g., a Read Only Memory (ROM)/Random Access Memory (RAM), a magnetic disk and a light disk) and includes a plurality of instructions used to enable a terminal device (e.g., a mobile phone, a computer, a server or a network device) to execute the method according to the embodiments of the present disclosure.

The embodiment further provides a multi-user detection apparatus for cell-free MIMO, and the apparatus is used for implementing the above embodiments and exemplary implementations which have been described and are not repeated. The term “module” used as below may realize combination of software and/or hardware with preset functions. Apparatuses described by the following embodiments are preferably realized by the software, but it is possible and conceivable for realizing the apparatuses through the hardware or combination of the software and the hardware.

FIG. 4 is a structural block diagram of a multi-user detection apparatus for cell-free MIMO according to an embodiment of the present disclosure. The apparatus may be located in a CPU of a cell-free MIMO system, and as shown in FIG. 4, the apparatus includes:

a receiving module 10, configured to receive a data symbol stream S_MRC=H^HY of K users transmitted by an AP, where S_MRC=H^HY denotes a K*L matrix, with each row representing L data symbols of one user, and both K and L are positive integers;

a first acquiring module 20, configured to multiply S_MRC by a conjugate transpose matrix S_MRC^H of S_MRC to obtain a K*K matrix S_MRCS_MRC^H;

a decomposition module 30, configured to perform singular value decomposition on the matrix S_MRCS_MRC^H to obtain a unitary matrix V=[v₁, v₂, . . . v_K] and K real numbers a₁ a₂ . . . a_K, such that

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H=VA{V}^H,$

where V^H is a transpose matrix of the unitary matrix V, and

$A=\left[\begin{array}{cccc}{a}_1& & & \\ & a_2& & \\ & & \ddots & \\ & & & a_K\end{array}\right];$

a second acquiring module 40, configured to obtain a diagonal matrix

$\Lambda =\left[\begin{array}{cccc}{\lambda }_1& & & \\ & {\lambda }_2& & \\ & & \ddots & \\ & & & {\lambda }_K\end{array}\right],$

where

${\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+{\sigma }^2}$ $\mathrm{or}{\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+c{\sigma }^2},$

c is a real number greater than 1, and σ² is a noise variance on received signals of the AP; and

an estimation module 50, configured to estimate, by a formula {tilde over (S)}=VΛV^HS_MRC, sending-end data symbols of the K users corresponding to the data symbol stream S_MRC.

In the embodiment, the decomposition module 30 may alternatively be configured to calculate K characteristic values a₁ a₂ . . . a_K and corresponding K characteristic vectors v₁, v₂, . . . , v_K of the matrix S_MRCS_MRC^H or

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H,$

such that

$\left(\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H\right)v_k=v_k{a}_k,$

and constitute the K*K unitary matrix V=[v₁, v₂, . . . v_K] by the K characteristic vectors v_k.

FIG. 5 is a structural block diagram of a multi-user detection apparatus for cell-free MIMO according to another embodiment of the present disclosure. The apparatus is located in a CPU of a cell-free MIMO system, and as shown in FIG. 5, the apparatus includes:

a receiving module 60, configured to receive a data symbol stream S_MRC=H^HY of K users transmitted by an AP, where S_MRC=H^HY denotes a K*L matrix, with each row representing L data symbols of one user, and both K and L are positive integers;

a decomposition module 70, configured to perform singular value decomposition on the matrix S_MRC to obtain a unitary matrix V=[v₁, v₂, . . . v_K,] and K real numbers ω₁, ω₂, . . . ω_K, such that S_MRC=V(√{square root over (L)}Ω)U^H, where U^H is a transpose matrix of a unitary matrix U, and Ω is a K*L matrix:

$\Omega =\left[\begin{array}{cccccc}{\omega }_1& & & & & \\ & {\omega }_2& & & 0& \\ & & \ddots & & & \\ & & & {\omega }_K& & \end{array}\right];$

an acquiring module 80, configured to obtain a diagonal matrix

$\Lambda =\left[\begin{array}{cccc}{\lambda }_1& & & \\ & {\lambda }_2& & \\ & & \ddots & \\ & & & {\lambda }_K\end{array}\right],$

where

${\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+{\sigma }^2}\mathrm{or}{\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+c{\sigma }^2},$

c is a real number greater than 1, and σ² is a noise variance on received signals of the AP, a_k=|ω_k|²; and

an estimation module 90, configured to estimate, by a formula {tilde over (S)}=VΛV^HS_MRC, sending-end data symbols of the K users corresponding to the data symbol stream S_MRC, where V^H is a transpose matrix of the unitary matrix V.

It needs to be explained that the above modules may be realized through the software or hardware, and for the latter, the modules may be realized by modes including but not limited to a following mode that the above modules are all located in the same processor; or the above modules are respectively located in different processors in the form of any combination.

FIG. 6 illustrates a cell-free MIMO system according to an embodiment of the present disclosure. As shown in FIG. 6, the system includes the multi-user detection apparatus according to the above embodiment.

To facilitate understanding of the technology provided by the present disclosure, the embodiments are described in detail in combination with specific scenarios.

Embodiment 1

The embodiment provides a multi-user detection apparatus for cell-free MIMO. In the embodiment, during cell-free MIMO uplink transmission, after receiving a user data symbol stream S_MRC=H^HY transmitted by an AP (the matrix S_MRC has K rows, each row has L symbols representing MRC of L data symbols of one user), a CPU obtains a matrix H^HH+σ²I based on S_MRC, and then obtains MMSE estimation of a data symbol S of each user according to H^HH+σ²I and S_MRC. As shown in FIG. 7, the method in the embodiment includes following operations S702 to S708.

At S702, a data symbol matrix S_MRC is multiplied by a conjugate transpose matrix S_MRC^H of S_MRC to obtain a K*K matrix S_MRCS_MRC^H.

At S704, SVD is performed on a square matrix S_MRCS_MRC^H or

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H$

or on a matrix cS_MRCS_MRC ^H formed by multiplying the matrix S_MRCS_MRC^Hby any constant, so as to obtain a K*K unitary matrix V and K real numbers a₁ a₂ . . . a_K, such that S_MRCS_MRC^H=V(L·A)V ^H, namely

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H={\mathrm{VAV}}^H,$

where c is a constant, A is a K*K diagonal matrix with the K real numbers a₁ a₂ . . . a_K as diagonal elements, namely

$A=\left[\begin{array}{cccc}{a}_1& & & \\ & a_2& & \\ & & \ddots & \\ & & & a_K\end{array}\right],$

and off-diagonal elements of the diagonal matrix are all 0.

In the embodiment, alternatively, K characteristic values a₁ a₂ . . . a_K and corresponding K characteristic vectors v₁, v₂, . . . , v_K of the square matrix S_MRCS_MRC^H or

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H$

may be calculated, such that

$\left(\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^H\right)v_k=v_k{a}_k$

which is satisfied for k=1, 2. . . K. The K characteristic vectors v_k constitute a K*K unitary matrix V=[v₁, v₂, . . . v_K].

At S706, a new diagonal matrix

$\Lambda =\left[\begin{array}{cccc}{\lambda }_1& & & \\ & {\lambda }_2& & \\ & & \ddots & \\ & & & {\lambda }_K\end{array}\right]$

is calculated, where

${\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+{\sigma }^2},\mathrm{or}{\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+c{\sigma }^2},$

and c is a real number greater than 1.

At S708, {tilde over (S)}=VΛV^HS_MRC is calculated, such that corresponding sending-end data symbols of the K users are estimated.

Embodiment 2

The embodiment provides a multi-user detection apparatus for cell-free MIMO. In the embodiment, during cell-free MIMO uplink transmission, after receiving a user data symbol stream S_MRC=H^HY transmitted by an AP (the matrix S_MRC has K rows, each row has L symbols representing MRC of L data symbols of one user), a CPU obtains a matrix H^HH+σ²I based on S_MRC, and then obtains MMSE estimation of a data symbol S of each user according to H^HH+σ²I and S_MRC. As shown in FIG. 8, the flow in the embodiment includes following operations S802 to S806.

At S802, SVD is performed on a K*L data symbol matrix S_MRC or

$\frac{1}{\sqrt{L}}{S}_{\mathrm{MRC}}$

to obtain a K*K unitary matrix V and K real numbers ω₁, ω₂, . . . ω_K, such that S_MRC=V(√{square root over (L)}Ω)U^H or

$\frac{1}{\sqrt{L}}{S}_{\mathrm{MRC}}=V\Omega U^H,$

where Ωis a K*L matrix generated by the K real numbers ω₁, ω₂, . . . ω_K, U^H is a K*K transpose matrix which may be uniquely determined through the SVD:

$\Omega =\left[\begin{array}{cccccc}{\omega }_1& & & & & \\ & {\omega }_2& & & 0& \\ & & \ddots & & & \\ & & & {\omega }_K& & \end{array}\right].$

At S804, a new diagonal matrix

$\Lambda =\left[\begin{array}{cccc}{\lambda }_1& & & \\ & {\lambda }_2& & \\ & & \ddots & \\ & & & {\lambda }_K\end{array}\right]$

is generated, where

${\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+{\sigma }^2}\mathrm{or}{\lambda }_k=\frac{2}{\sqrt{{\sigma }^4+4a_k}+c{\sigma }^2},$

and c is a real number greater than 1, a_k=|ω_k|².

At S806, {tilde over (S)}=VΛV^HS_MRC is calculated, such that corresponding sending-end data symbols of K users are estimated.

To facilitate understanding of the above embodiments of the present disclosure, principles on which the embodiments of the present disclosure are based will be described in detail below.

When K users have access to M APs at the same time, each user sends L data symbols, and a symbol matrix received by the CPU is:

S _MRC =H ^H Y=H ^H(HS+N)=H^HHS+H^HN,

where

$Y=\left[\begin{array}{c}{y}_0\\ y_1\\ ⋮\\ y_{M-1}\end{array}\right]$

denotes an M*L matrix having M rows, and the m^th row is y_m, which denotes the L data symbols received by the m^th AP.

$S=\left[\begin{array}{c}{s}_0\\ s_1\\ ⋮\\ s_{K-1}\end{array}\right]$

denotes a K*L matrix, the k^th row is S_k, which denotes L data symbols sent by k^th UE.

H=[h₀, h₁, . . . h_K−1] denotes an M*K matrix, where the k^th column.

h_k=[h_0,k, h_1,k, . . . h_M−1,k]^T denotes a spatial domain channel vector from the user k to the M APs, which is a column vector having a length of M, i.e., the column vector has M rows. The AWGN noise N is an M*L matrix.

Thus, the symbol matrix S_MRC received by the CPU is a K*L matrix.

If H or H^HH is given, MMSE estimation may be performed on the data symbol S of each user according to S_MRC=H^HY, that is, S_MMSE=(H^H+σ²I)⁻¹H^HY=(H^HH +σ²I)⁻¹S_MRC.

I denotes a K*K unit matrix. Assuming that modulation symbols transmitted by the users all have normalized energy, the variance of the AWGN on each AP is σ².

But the CPU does not know H and H^HH, and thus the above MMSE cannot be implemented. However, it can be seen that after obtaining the data symbol S_MRC=H^HY, the CPU can further obtain H^HH+σ²I required by the MMSE as long as obtaining H^HH from the data symbol S_MRC, thereby implementing the MMSE.

How to obtain H^HH+σ²I from the data symbol S_MRC is described below.

• • 1) Due to S_MRC=H^HY=H^HHS+H^HN, a correlation matrix of S_MRC is:

C _MRC =S _MRC S _MRC ^H=(h^HHS+H^HN)(H^HHS+H^HN)^H

Because the AWGN on the different APs is independent while the data symbols of the different users are independent as well, along with increasing of the symbol number L,

$\frac{{\mathrm{SS}}^H}{L}$

will approach the K*K unit matrix while

$\frac{N{N}^H}{L}$

will approach a product of the K*K unit matrix and σ². Thus, as long as the symbol number L is large enough, (H^HHH^HH+σ²H^H) may be replaced with

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^{*}.$

• • 2) Further, because H^HH is a complex symmetric matrix, suppose H^HH=VXV^H,

$X=\left[\begin{array}{cccc}{x}_1& & & \\ & x_2& & \\ & & \ddots & \\ & & & x_K\end{array}\right]$

is a diagonal matrix, off-diagonal elements are 0, and elements x_k on the diagonal line are non-negative real numbers not less than 0, that is, x_k≥0. V is a unitary matrix, and V^H is a conjugate symmetric matrix of V.

How to obtain V and X from (H^HHH^HH+σ²H^HH) is described below.

H^HH=VXV^H is substituted into (H^HHH^HH+σ²H^HH) to obtain:

$\left(H^{H}H{H}^{H}H+{\sigma }^2{H}^{H}H\right)=V\left[\begin{array}{cccc}{x}_1^2+{\sigma }^2{x}_1& 0& \dots & 0\\ 0& x_2^2+{\sigma }^2{x}_2& \dots & 0\\ \dots & \dots & \dots & \dots \\ 0& 0& \dots & x_K^2+{\sigma }^2{x}_K\end{array}\right]V^H$

Further, SVD is performed on

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^{*},$

namely, (H^HHH^HH+σ²H^HH) to obtain:

H^HHH^HH+σ²H^HH=VΛV^H, where

$A=\left[\begin{array}{cccc}{a}_1& & & \\ & a_2& & \\ & & \ddots & \\ & & & a_K\end{array}\right].$

Because singular values of one matrix are unique, suppose that the singular values are arranged in order, then x_k²+σ²x_k=a_k may be obtained, where k=1, . . . , K. Each of the K relational expressions is essentially a quadratic equation in one unknown, and may have two solutions. However, due to x_k≥0, the negative solution is meaningless, and accordingly,

$x_k=\frac{\sqrt{{\sigma }^4+4a_k}-{\sigma }^2}{2}$

can be obtained, where k=1, . . . , K.

Thus, V and X in H^HH=VXV^H both can be obtained by the SVD on

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^{*}.$

Further, H^HH+σ²I=V(X+σ²I)V^H can be obtained, suppose Λ=(X+σ²I)⁻¹, namely,

${\lambda }_k=\frac{1}{x_k+{\sigma }^2}=\frac{1}{\frac{\sqrt{{\sigma }^4+4a_k}-{\sigma }^2}{2}+{\sigma }^2}=\frac{2}{\sqrt{{\sigma }^4+4a_k}+{\sigma }^2},$

thus, MMSE estimation may be performed on the data symbol S of each user according to S_MRC=H^HY that is, S_MMSE=(H^HH+σ²I)⁻¹H^HY=(H^HH+σ²I)⁻¹S_MRC=V(X+σ²I)⁻¹V^HS_MRC=VΛV^HS_MRC.

Another method for obtaining H^HH+σ²I from the data symbol S_MRC is described below.

• • 1) Due to S_MRC=H^HY=H^HHS+H^HN, a correlation matrix of S_MRC is:

C _MRC =S _MRC S _MRC ^H=(H^HHS+H^HN)(H^HHS+H^HN)^H.

Because the AWGN on the different APs is independent while the data symbols of the different users are independent as well, along with increasing of the symbol number L,

$\frac{{\mathrm{SS}}^H}{L}$

will approach the K*K unit matrix while

$\frac{{\mathrm{NN}}^H}{L}$

will approach a product of the K*K unit matrix and σ². Thus, as long as the symbol number L is large enough, (H^HHH^HH+σ²H^HH) may be replaced with

$\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^{*}.$

• • 2) Further, according to

$H^H{\mathrm{HH}}^{H}H+{\sigma }^2{H}^{H}H=\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^{*},H^H{\mathrm{HH}}^{H}H+{\sigma }^2{H}^{H}H+\frac{1}{4}{\sigma }^2{I}_{K\times K}=\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^{*}+\frac{1}{4}{\sigma }^2{I}_{K\times K},$

that is,

$\left(H^{H}H+\frac{1}{2}{\sigma }^2{I}_{K\times K}\right)\left(H^{H}H+\frac{1}{2}{\sigma }^2{I}_{K\times K}\right)=\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^{*}+\frac{1}{4}{\sigma }^2{I}_{K\times K}$

thus,

$H^{H}H+\frac{1}{2}{\sigma }^2{I}_{K\times K}=\sqrt{\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^{*}+\frac{1}{4}{\sigma }^2{I}_{K\times K}},$

accordingly,

$H^{H}H+{\sigma }^2{I}_{K\times K}=\sqrt{\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^{*}+\frac{1}{4}{\sigma }^2{I}_{K\times K}}+\frac{1}{2}{\sigma }^2{I}_{K\times K}$

can be further obtained,

and thus, MMSE estimation may be performed on the data symbol S of each user according to S_MRC=H^HY that is,

$S_{\mathrm{MMSE}}={\left(H^{H}H+{\sigma }^{2}I\right)}^{-1}{H}^{H}Y={\left(H^{H}H+{\sigma }^{2}I\right)}^{-1}{S}_{\mathrm{MRC}}={\left(\sqrt{\frac{1}{L}{S}_{\mathrm{MRC}}{S}_{\mathrm{MRC}}^{*}+\frac{1}{4}{\sigma }^2{I}_{K\times K}}+\frac{1}{2}{\sigma }^2{I}_{K\times K}\right)}^{-1}{S}_{\mathrm{MRC}}$

An embodiment of the present disclosure further provides a computer-readable storage medium storing computer programs, where the computer programs, when being executed by a processor, causes the processor to implement the operations in any above method embodiment.

In an exemplary embodiment, the above computer-readable storage medium may include but not limited to: a U disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a mobile hard disk, a magnetic disk or a light disk or other media capable of storing the computer programs.

An embodiment of the present disclosure further provides an electronic device including a memory and a processor, where the memory stores computer programs, and the processor is configured to run the computer programs to implement the operations in above any method embodiment.

In an exemplary embodiment, the above electronic device may further include a transmission device and an input and output device, where the transmission device is connected to the above processor, and the input and output device is connected to the above processor.

Specific examples in the embodiment may refer to the examples described in the above embodiments and exemplary implementations, so that no detail is repeated in the embodiment.

Obviously, those having ordinary skill in the art should understand that the modules or operations in the present disclosure may be implemented through a universal computing device, may be centralized in a single computing device or distributed in a network formed by multiple computing devices, and may be implemented by executable program code of the computing device, such that the modules or operations may be stored in a storage apparatus to be executed by the computing device; and the shown or described operations may be executed in sequence different from the sequence herein under some situations, or the modules or operations may be made into various integrated circuit modules, or more of the modules or operations may be made into single integrated circuit modules to be implemented. Thus, the present disclosure is not limited to any specific hardware and software combination.

The above contents are merely exemplary embodiments of the present disclosure, and are not intended to limit the present disclosure, and for those having ordinary skill in the art, the present disclosure may have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the principle of the present disclosure shall fall within the scope of protection of the present disclosure.

Claims

1. A multi-user detection method for cell-free Multiple-Input Multiple-Output (MIMO), comprising: receiving, by a Central Processing Unit (CPU) of a cell-free MIMO system, an MRC merged data symbol stream SMRC=HHY of K users transmitted by an Access Point (AP), wherein SMRC=HHY denotes a K*L matrix, with each row representing L data symbols of one user, and both K and L are positive integers;multiplying SMRC by a conjugate transpose matrix SMRC H of SMRC to obtain a K*K matrix SMRCSMRCH;performing singular value decomposition on the matrix SMRCSMRCH or

2. The method according to claim 1, wherein before obtaining the diagonal matrix Λ, the method further comprises: receiving, by the CPU, the noise variance σ2 transmitted by the AP, or determining, by the CPU, the noise variance σ2 according to an attribute of the AP.

3. A multi-user detection method for cell-free Multiple-Input Multiple-Output (MIMO), comprising: receiving, by a Central Processing Unit (CPU) of a cell-free MIMO system, an MRC merged data symbol stream SMRC=HHY of K users transmitted by an Access Point (AP), wherein SMRC=HHY denotes a K*L matrix, with each row representing L data symbols of one user, and both K and L are positive integers;performing singular value decomposition on the matrix SMRC or to obtain a unitary matrix V=[v1, v2, . . . vK] and K real numbers ω1, ω2, . . . ωK, such that SMRC=V(√{square root over (L)}Ω)UH or, wherein UH is a transpose matrix of a unitary matrix U, and Ω is a K*L matrix:

4. The method according to claim 3, wherein before obtaining the diagonal matrix Λ, the method further comprises: receiving, by the CPU, the noise variance σ2 transmitted by the AP, or determining, by the CPU, the noise variance σ2 according to an attribute of the AP.

5. A multi-user detection apparatus for cell-free Multiple-Input Multiple-Output (MIMO), located in a Central Processing Unit (CPU) of a cell-free MIMO system, and comprising a memory storing instructions and a processor in communication with the memory, wherein the processor is configured to execute the instructions to: a data symbol stream SMRC=HHY of K users transmitted by an Access Point (AP), wherein SMRC=HHY denotes a K*L matrix, with each row representing L data symbols of one user, and both K and L are positive integers;multiply SMRC by a conjugate transpose matrix SMRCH of SMRC to obtain a K*K matrix SMRCSMRCH;perform singular value decomposition on the matrix SMRCSMRCH or

6. The apparatus according to claim 5, wherein the processor is further configured to execute the instructions to receive the noise variance σ2 transmitted by the AP, or determine the noise variance Γ2 according to an attribute of the AP.

7. A multi-user detection apparatus for cell-free Multiple-Input Multiple-Output (MIMO), located in a Central Processing Unit (CPU) of a cell-free MIMO system, and comprising a memory storing instructions and a processor in communication with the memory, wherein the processor is configured to execute the instructions to implement the operations of the method according to claim 3

8. (canceled)

9. A cell-free Multiple-Input Multiple-Output (MIMO) system, comprising the multi-user detection apparatus according to claim 5.

10. A non-transitory computer-readable storage medium, storing computer programs, wherein the computer programs, when being executed by a processor, causes the processor to implement the operations of the method according to claim 1.

11. (canceled)

12. The method according to claim 1, wherein in a case of calculating K characteristic values a1 a2 . . . aK and corresponding K characteristic vectors v1, v2, . . . vK of the matrix SMRCSMRCH or

13. The method according to claim 1, wherein off-diagonal elements of the diagonal matrix

14. The method according to claim 1, wherein estimating, by a formula {tilde over (S)}=VΛVHSMRC, sending-end data symbols of the K users corresponding to the data symbol stream SMRC comprises: obtaining MMSE estimation of sending-end data symbols of the K users according to HHH+σ2I and SMRC.

15. The method according to claim 3, wherein estimating, by a formula {tilde over (S)}=VΛVHSMRC, sending-end data symbols of the K users corresponding to the data symbol stream SMRC comprises: obtaining MMSE estimation of sending-end data symbols of the K users according to HHH+σ2I and SMRC.

16. The apparatus according to claim 5, wherein in a case of calculating K characteristic values a1 a2 . . . aK and corresponding K characteristic vectors v1, v2, VK of the matrix SMRCSMRCH or

17. The apparatus according to claim 5, wherein off-diagonal elements of the diagonal matrix

18. The apparatus according to claim 5, wherein the processor is configured to execute the instructions to: obtain MMSE estimation of sending-end data symbols of the K users according to HHH+σ2I and SMRC.

19. A multi-user detection apparatus for cell-free Multiple-Input Multiple-Output (MIMO), located in a Central Processing Unit (CPU) of a cell-free MIMO system, and comprising a memory storing instructions and a processor in communication with the memory, wherein the processor is configured to execute the instructions to implement the operations of the method according to claim 4.

20. A multi-user detection apparatus for cell-free Multiple-Input Multiple-Output (MIMO), located in a Central Processing Unit (CPU) of a cell-free MIMO system, and comprising a memory storing instructions and a processor in communication with the memory, wherein the processor is configured to execute the instructions to implement the operations of the method according to claim 15.

21. A cell-free Multiple-Input Multiple-Output (MIMO) system, comprising the multi-user detection apparatus according to claim 7.

22. A non-transitory computer-readable storage medium, storing computer programs, wherein the computer programs, when being executed by a processor, causes the processor to implement the operations of the method according to claim 3.