CN110492956B - Error compensation multi-user detection method and device for MUSA (multiple input multiple output) system - Google Patents

Error compensation multi-user detection method and device for MUSA (multiple input multiple output) system Download PDF

Info

Publication number
CN110492956B
CN110492956B CN201910834807.8A CN201910834807A CN110492956B CN 110492956 B CN110492956 B CN 110492956B CN 201910834807 A CN201910834807 A CN 201910834807A CN 110492956 B CN110492956 B CN 110492956B
Authority
CN
China
Prior art keywords
matrix
user
transmitting end
error
noise
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201910834807.8A
Other languages
Chinese (zh)
Other versions
CN110492956A (en
Inventor
陈发堂
石贝贝
邓青
李小文
王丹
王华华
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.)
Chongqing University of Post and Telecommunications
Original Assignee
Chongqing University of Post and Telecommunications
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 Chongqing University of Post and Telecommunications filed Critical Chongqing University of Post and Telecommunications
Priority to CN201910834807.8A priority Critical patent/CN110492956B/en
Publication of CN110492956A publication Critical patent/CN110492956A/en
Application granted granted Critical
Publication of CN110492956B publication Critical patent/CN110492956B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B17/00—Monitoring; Testing
    • H04B17/30—Monitoring; Testing of propagation channels
    • H04B17/309—Measuring or estimating channel quality parameters
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B17/00—Monitoring; Testing
    • H04B17/30—Monitoring; Testing of propagation channels
    • H04B17/382—Monitoring; Testing of propagation channels for resource allocation, admission control or handover
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B17/00—Monitoring; Testing
    • H04B17/30—Monitoring; Testing of propagation channels
    • H04B17/391—Modelling the propagation channel
    • H04B17/3912—Simulation models, e.g. distribution of spectral power density or received signal strength indicator [RSSI] for a given geographic region
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L27/00—Modulated-carrier systems
    • H04L27/18—Phase-modulated carrier systems, i.e. using phase-shift keying
    • H04L27/20—Modulator circuits; Transmitter circuits
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L27/00—Modulated-carrier systems
    • H04L27/18—Phase-modulated carrier systems, i.e. using phase-shift keying
    • H04L27/22—Demodulator circuits; Receiver circuits

Abstract

The invention belongs to the technical field of mobile communication, and relates to an uplink multi-user detection technology of an MUSA (multi user agent) system; an error compensation multi-user detection method and device for MUSA system; the method comprises the steps that after user data of a transmitting end is modulated by QPSK, a complex spreading sequence is randomly selected for spreading; the receiving end adopts parallel interference elimination detection to the received signal, and calculates a weight matrix to obtain an estimated value of the user signal of the transmitting end; performing eigenvalue decomposition on the characteristic matrix, and selecting partial decomposition results to reconstruct a user signal of a transmitting end; minimizing the equivalent noise matrix by using a Lagrange multiplier method, and readjusting the estimated value of the equivalent noise matrix; and calculating the actual user signal estimation value by adopting the maximum likelihood estimation and selecting one for output. The invention estimates the error of the signal by compensating the MMSE criterion, projects the related noise to the feature vector space, searches the signal in the direction of noise enhancement, obtains better detection performance, improves the detection accuracy and the like.

Description

Error compensation multi-user detection method and device for MUSA (multiple input multiple output) system
Technical Field
The invention belongs to the technical field of mobile communication, and particularly relates to an uplink multi-user detection technology of an MUSA (multi user agent) system; in particular to an error compensation multi-user detection method and device for an MUSA system.
Background
The 3GPP R15 is the first edition of the 5G standard, and mainly focuses on supporting scenarios such as eMBB and URLLC, and the non-orthogonal multiple access is used as an alternative to a 5G new air interface uplink multiple access scheme, and when a subsequent 5G scenario is specially designed for an mtc scenario, its unique advantages will inevitably emerge, so as to meet the requirements of 5G for different service scenarios. Therefore, the non-orthogonal multiple access technology needs to be researched at present. The existing Non-Orthogonal Multiple Access technology includes a SCMA (Sparse Code Multiple Access), a MUSA (Multiple user shared Access), a NOMA (Non-Orthogonal Multiple Access), and a PDMA (Pattern division Multiple Access).
The non-orthogonal multiple access technology has obvious gain in the aspects of system uplink throughput and the number of access users. The MUSA technology proposed by Zhongxing company is a code domain, non-orthogonal multiple access scheme based on spread spectrum communication, and does not need a complex scheduling access process because the accessed user randomly selects a spreading sequence, namely, the MUSA technology is a scheduling-free mode, and the scheduling-free mode saves scheduling time and reduces signaling overhead, so that the MUSA technology has great advantages on massive connection and low time delay in three scenes with 5G requirements. The specific scheme can include that a plurality of users accessing the system at the sending end independently select a plurality of spreading sequences to spread modulation symbols of the users, and then the spread user data is sent in the same time frequency resource. The receiving end identifies and separates the data of each user through a minimum mean square error multi-user detection technology.
The traditional multi-user detection technology generally comprises a minimum mean square error-serial interference cancellation (MMSE-SIC) algorithm, a minimum mean square error-parallel interference cancellation (MMSE-PIC) algorithm and a quasi-parallel interference cancellation detection algorithm; the number of stages of detection carried out by the MMSE-SIC detection method is determined by the number of users, and each stage of detection needs to carry out a weight matrix omegaMMSEThe calculation of the weight matrix involves matrix inversion, which is highly complex with a complexity of O (M)3) When the number of users is large, the overall complexity of the method is high, so that the time delay is increased, and the method is not suitable for a 5G low-time-delay scene. Although the MMSE-PIC detection method has a small detection level and can effectively reduce the time delay, the overall detection performance is poor. Although the quasi-parallel interference elimination detection algorithm has smaller processing time delay than MMSE-SIC and better performance than MMSE-PIC, the detection performance is still poorer than MMSE-SIC.
Disclosure of Invention
Based on the problems in the prior art, the invention considers that if the performance of the MMSE-PIC detection method can be improved and the detection accuracy is improved, the requirements of low time delay can be met while the performance is ensured, and the calculation complexity can be reduced; therefore, the invention provides an error compensation multi-user detection method and device for an MUSA system.
The method for error compensation multi-user detection for the MUSA system, as shown in fig. 1, may include:
s1, after modulating the user data of the transmitting end, randomly selecting a plurality of spreading sequences to spread the respective modulation symbols, and transmitting the spreading sequences from the same time-frequency resource;
s2, the receiving end carries out parallel interference elimination detection on the received signal y through an MMSE detector, and calculates a weight matrix omegaMMSEObtaining a modulation symbol estimation value x of a user signal of a transmitting end;
s3, decomposing the eigenvalue of the eigenvalue matrix in the weight matrix to obtain the eigenvalue lambdalCorresponding feature vector vlAnd constructing equivalent noise a by using the H matrix of the feature vector and the noisel(ii) a l ═ 1, 2.., M }, where M denotes the number of users at the transmitting end;
s4, selecting the first N larger eigenvalues lambdakCorresponding feature vector vkAnd equivalent noise akCalculating the error e between each modulation symbol corresponding to the user signal of the transmitting terminal and the estimated modulation symbol estimated valuel(m)；k＝{1,2,...,N}；
S5, adopting Lagrange multiplier method to make equivalent noise matrix aHa minimization using calculated error and eigenvalue lambdakAnd a feature vector vkCalculating the estimation value of the equivalent noise matrix corresponding to each modulation symbol
Figure BDA0002191814890000021
S6, adopting maximum likelihood estimation mode to estimate the value according to the equivalent noise matrix
Figure BDA0002191814890000022
Readjusting error el(m) and using it to compensate for errors formed by the detection algorithm, thereby calculating the actual user signal estimate
Figure BDA0002191814890000031
And selects the actual user signal estimate in which the error is the smallestValue of
Figure BDA0002191814890000032
As a final output;
the H matrix represents a Hermitian matrix; m represents the number of symbols corresponding to the modulation scheme.
For example, when the modulation scheme used is QPSK modulation, m is {1,2,3,4 }.
In addition, the present invention also provides an error compensation multi-user detection apparatus for a MUSA system, the apparatus comprising:
a transmitting antenna: for transmitting user data;
the modulator: for modulating user data;
a sequence spreading module: for performing sequence spreading on the modulated data;
a receiver: for receiving the modulated spread user data;
MMSE detector: for solving a weight matrix by minimizing a minimum mean square error between the transmit vector and the estimate vector;
a characteristic decomposition module: the characteristic matrix in the weight matrix is used for carrying out characteristic value decomposition; the characteristic value lambda is resolvediCorresponding feature vector vi；
An equivalent noise construction module: the H matrix used for decomposing the eigenvector matrix according to the eigenvector decomposition unit is multiplied by the noise;
an error estimation module: taking the first N larger eigenvalues, the corresponding eigenvectors and the equivalent noise to calculate the error between each modulation symbol corresponding to the user signal of the transmitting end and the estimated modulation symbol estimated value;
lagrange multiplier module: the method is used for calculating the estimation value of the equivalent noise matrix corresponding to each symbol according to a Lagrange multiplier method;
an error calculation module: the system is used for calculating actual user signals according to the estimated values of all equivalent noise matrixes calculated by the Lagrange multiplier method module;
a hard decision module: for hard decision of the calculated actual user signal and selecting the error with the minimum as the output.
The invention has the beneficial effects that:
the method is based on MMSE criterion, and based on the MMSE criterion, the error of the signal estimated by MMSE detection is compensated, the related noise is projected to the characteristic vector of the matrix R, and the signal is searched in the direction of noise enhancement, so that the detection accuracy is improved. Analysis and simulation results show that the detection performance of the method is superior to that of an MMSE-SIC detection method, and the performance is greatly improved compared with that of the MMSE-PIC detection method.
Drawings
FIG. 1 is a flow chart of a system employed in the method of the present invention;
FIG. 2 is a block diagram of a MUSA system employed by the present invention;
FIG. 3 is a constellation diagram of ternary complex sequence elements;
FIG. 4 is a flow chart of spreading sequence selection;
fig. 5 is a graph of simulation result performance.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more clearly and completely apparent, the technical solutions in the embodiments of the present invention are described below with reference to the accompanying drawings, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments.
In this embodiment, taking fig. 2 as an example, assuming that the number of users at the sending end is M, each user data stream is subjected to coding and modulation processes; each user independently and randomly selects a complex spreading sequence satisfying a balance criterion and a correlation criterion, and the modulation symbols are spread by the respective spreading sequences.
In one embodiment, the encoding process employs differential encoding; the modulation process adopts a Quadrature Phase Shift Keying (QPSK) modulation method. The length of the spreading sequence is N, and then the spreading sequence is sent on the same time frequency resource, namely a multi-user shared channel, noise is passively generated after the transmission of the channel, and a receiving end separates each user data through multi-user detection.
As an implementation manner, the invention can also adopt modulation modes such as 16QAM, 64QPSK and the like.
In this embodiment, the real part and the imaginary part of the complex spreading sequence are respectively taken from a ternary set { -1,0,1}, and the constellation diagram thereof is shown in fig. 3. The user data is QPSK modulated and then independently selects spreading sequences, and the flow of spreading sequence selection can be as shown in fig. 4.
Each user independently selects an extended sequence, and initially, i is set to 1;
randomly generating an extended sequence by the ith user, judging whether the extended sequence meets a balance criterion and an autocorrelation criterion, and if so, making i equal to i + 1; otherwise, continuing to randomly generate a spreading sequence for the ith user;
when i is i +1, judging whether i is less than or equal to M, if so, randomly generating an extended sequence for the ith user; otherwise, the process of independently selecting the spreading sequences by all users is finished;
and judging whether the cross-correlation criterion is met between every two users, if so, ending the whole process, otherwise, starting from the initial state, and selecting the extension sequence independently for each user again.
In one embodiment, the process of separating each user data by the receiving end through multi-user detection may include:
step 1, a receiving end firstly carries out parallel interference elimination detection on a received signal y through an MMSE detector;
in the process, a weight matrix needs to be calculated and a modulation symbol of a transmitting end user is estimated according to the weight matrix;
the weight matrix is represented as: omegaMMSE＝(HHH+σ2I)-1HHAnd R ═ HHH+σ2I)-1；
The estimated modulation symbols are represented as:
Figure BDA0002191814890000051
wherein, H is a channel matrix of a user at a transmitting end; hHA Hermitian matrix which is a channel matrix H; sigma2Representing the variance of the noise; i denotes an identity matrix.
The received signal y is represented as:
y＝Hx+n；
which is represented by a scalar quantity
Figure BDA0002191814890000052
Is popularized to obtain, wherein SmSpreading sequence for mth user, gmChannel gain, x, for the mth usermIs the modulation symbol of the mth user.
In the vector expression, y is (y)1,y2,y3,......,yN)HH is an equivalent channel matrix including a spreading sequence and a channel gain, and is obtained by channel estimation, where x ═ x (x)1,x2,x3,......,xM)H,n＝(n1,n2,n3,......,nN)H，nkWhich represents the k-th noise signal, N is the spreading sequence length and M is the number of users.
Step 2, decomposing the characteristic value eigenvector of the matrix R to obtain R ═ VDVH(ii) a Specifically decomposing the characteristic value lambdaiCorresponding feature vector viConstructing equivalent noise a by using the H matrix of the eigenvector and the noisei(ii) a i ═ 1, 2.., M }, where M denotes the number of users at the transmitting end;
wherein, V represents that the characteristic vector matrix comprises M × M dimensional unitary matrix composed of M mutually orthogonal M × 1 dimensional column characteristic vectors, and V ═ V1,v2,......,vM](ii) a D represents that the eigenvalue matrix comprises an M multiplied by M diagonal matrix consisting of M eigenvalues, D ═ diag [ lambda ]1,λ2,......,λM]；VHA Hermitian matrix representing the eigenvector matrix, wherein an equivalent noise matrix comprising M equivalent noises is defined as a ═ VHn'，n'＝(n'1,n'2,...,n'M)H，n'iRepresenting the ith assumed noise signal, the superscript H represents the Hermitian matrix, and since a is equivalent noise, its order statistical properties are unchanged, i.e. its mean, variance and n are equal to (n)1,n2,n3,......,nN)HThe same is true.
Step 3, selecting larger N (N is less than M) eigenvalues lambda from the M eigenvaluesk(k ═ {1,2,.. multidata, N }), and the eigenvectors v corresponding to these N eigenvalueskNew matrix V and matrix D are formed, and the relation is that the actual user sending signal s has an error with the estimated value x
Figure BDA0002191814890000061
Thus, the new matrix V is an M N dimensional matrix and the new matrix D is an N order matrix, and estimates the actual user transmitted signals, denoted as
Figure BDA0002191814890000062
At this time, since the equivalent noise matrix a is not calculated, it is necessary to estimate a next.
Step 4, utilizing Lagrange multiplier method to make equivalent noise matrix aHa is minimized, and the estimation value of the equivalent noise matrix corresponding to each symbol is calculated through constraint conditions
Figure BDA0002191814890000063
Estimation value of equivalent noise matrix corresponding to each modulation symbol
Figure BDA0002191814890000064
The calculation formula of (2) is as follows:
Figure BDA0002191814890000065
wherein, C ═ C1,c2,...,cM]，
Figure BDA0002191814890000066
(·)↑Represents Moore-Penrose pseudo-inverse; e ═ E1 *(m),e2 *(m),...,eM *(m)]；
Figure BDA0002191814890000067
bl(m) denotes the mth symbol that the ith transmitting end user may send,
Figure BDA0002191814890000068
Figure BDA0002191814890000069
indicating the modulation symbol estimated value sent by the ith transmitting terminal user; (v)k)lRepresenting a feature vector vkThe l element of (1); the superscript H denotes the Hermitian matrix.
In the process, the error between the transmission symbol and the modulation symbol corresponding to each symbol needs to be calculated
Figure BDA0002191814890000071
Because the QPSK modulation scheme is adopted in this embodiment, and each transmission symbol has four symbols, then
Figure BDA0002191814890000072
bl(m) denotes the mth symbol that the ith transmitting user may send.
In a preferred embodiment, an estimate of the equivalent noise matrix for each modulation symbol is determined
Figure BDA0002191814890000073
The calculation formula of (2) further comprises each modulation symbol b corresponding to the user signal of the transmitting endl(m) minimizing the cost function f [ a ] using Lagrange multiplier method with the error between the estimated modulation symbol estimated value x and each as constraint condition]Selecting two transmitting end users l1And l2For the estimated value
Figure BDA0002191814890000074
Make an estimationCounting; the process of minimizing the cost function by the Lagrange multiplier method comprises the following steps:
Figure BDA0002191814890000075
s.t.
constraint 1:
Figure BDA0002191814890000076
constraint 2:
Figure BDA0002191814890000077
constraint 3:
Figure BDA0002191814890000078
wherein, f [ a ]]Representing an estimated value
Figure BDA0002191814890000079
The cost function of (2); gamma ray1And gamma2All represent lagrangian coefficients; e.g. of the typel1(m) denotes the l-th in the actual user signal1Error corresponding to the mth symbol possibly transmitted by a transmitting end user, el2(m) denotes the l-th in the actual user signal2The error corresponding to the mth symbol possibly sent by each transmitting end user represents a conjugate symbol;
Figure BDA00021918148900000710
Figure BDA00021918148900000711
Figure BDA00021918148900000712
subscriptl1And subscriptsl2Represents two different transmitting end users belonging to {1, 2.., M }; (v)1)l1Representing a feature vector v1L. 11An element; (v)1)l2Representing a feature vector v1L. 12And (4) each element.
In addition, constraint 1 is generated by pairing a*Obtaining a partial derivative:
Figure BDA0002191814890000081
it will be appreciated that the lagrangian multiplier procedure described above uses only two modulation symbols b of each type corresponding to the user signal at the transmitting endl(m) constraints on the error between the estimated modulation symbol estimates x, i.e. constraints corresponding to two transmitting end users are selected from C:
Figure BDA0002191814890000082
and
Figure BDA0002191814890000083
(constraint 2 and constraint 3); accordingly, E ═ El1 *(m),el2 *(m)]H(ii) a If in order to make the estimation value
Figure BDA0002191814890000084
The estimation result is more accurate, and more constraints can be selected from M transmitting end users, for example, the constraints are also increased:
Figure BDA0002191814890000085
l3represents the transmitting end users belonging to {1, 2.., M }; at this time, E is [ E ]l1 *(m),el2 *(m),el3 *(m)]H(ii) a Of course, the transmitting end user l at this time1、l2And l3Is also selected after calculating the SINR sequence; specifically, the following procedures can be referred to:
transmitting end user l1And l2The selection process comprises the steps of calculating and sorting the signal to interference plus noise ratio (SINR) of each transmitting end user, and selecting two users with the minimum SINR as transmitting end usersl1And transmitting end usersl2(ii) a Wherein the SINR isThe calculation formula is expressed as:
Figure BDA0002191814890000086
wherein ω isi,MMSEDenoted is the ith row, h, of the weight matrixiRepresenting the ith column, σ, of the equivalent channel matrix H2Is the noise variance, | | | · | | represents the Frobenius norm.
Step 5, adopting a maximum likelihood estimation mode to estimate the value according to the equivalent noise matrix
Figure BDA0002191814890000087
Calculating an actual user signal estimate
Figure BDA0002191814890000088
And selects the actual user signal estimate in which the error is the smallest
Figure BDA0002191814890000089
As the final output.
Since the values of b (m) are different from each other, E ═ E is calculated in all cases1 *(m),e2 *(m)]H，
Figure BDA00021918148900000810
And calculate
Figure BDA00021918148900000811
Obtain the conditions of
Figure BDA00021918148900000812
(·)↑Representing the Moore-Penrose pseudo-inverse, the final output is calculated from the following equation:
Figure BDA00021918148900000813
wherein argmin represents
Figure BDA0002191814890000091
Corresponding independent variable when obtaining minimum value
Figure BDA0002191814890000092
Q (-) denotes a hard decision; and | l | · | | represents the Frobenius norm.
In one embodiment, the present invention also provides an error-compensated multi-user detection apparatus for a MUSA system, the apparatus comprising:
a transmitting antenna: for transmitting user data;
the modulator: for modulating user data;
a sequence spreading module: for performing sequence spreading on the modulated data;
a receiver: for receiving the modulated spread user data;
MMSE detector: for solving a weight matrix by minimizing a minimum mean square error between the transmit vector and the estimate vector;
a characteristic decomposition module: the characteristic matrix in the weight matrix is used for carrying out characteristic value decomposition; the characteristic value lambda is resolvediCorresponding feature vector vi；
An equivalent noise construction module: the H matrix used for decomposing the eigenvector matrix according to the eigenvector decomposition unit is multiplied by the noise;
an error estimation module: taking the first N larger eigenvalues, the corresponding eigenvectors and the equivalent noise to calculate the error between each modulation symbol corresponding to the user signal of the transmitting end and the estimated modulation symbol estimated value;
lagrange multiplier module: the method is used for calculating the estimation value of the equivalent noise matrix corresponding to each symbol according to a Lagrange multiplier method;
an error calculation module: the system is used for calculating actual user signals according to the estimated values of all equivalent noise matrixes calculated by the Lagrange multiplier method module;
a hard decision module: for hard decision of the calculated actual user signal and selecting the error with the minimum as the output.
Wherein the feature decomposition module comprises a feature value decomposition unit and a feature vector decomposition unit; the characteristic value decomposition unit is used for decomposing an M multiplied by M diagonal matrix consisting of M characteristic values; the eigenvector decomposition unit is used for decomposing M × M unitary matrixes M composed of M orthogonal M × 1 dimensional column eigenvectors to represent the number of users at the transmitting end.
The device transmits user data to a receiver through a transmitting antenna, modulates the data transmitted by the transmitting antenna through a modulator before transmitting the data to the receiver, utilizes a sequence expansion module to expand the modulated data, receives the user data after transmission through a multi-user shared channel, solves a weight matrix through an MMSE (minimum mean square error) detector, and utilizes a characteristic decomposition module to decompose a characteristic matrix in the weight matrix so as to decompose the characteristic matrix and a characteristic value; multiplying the H matrix of the decomposed eigenvector matrix by noise through an equivalent noise construction module to construct an equivalent noise matrix; selecting a result after partial decomposition and an equivalent noise matrix through an error estimation module to estimate the error between each modulation symbol corresponding to the user signal of the transmitting end and the estimated modulation symbol estimation value; thereby reconstructing the user signal of the transmitting end; and finally, calculating an estimated value of the equivalent noise through an error calculation module according to the estimated value of the equivalent noise, thereby calculating a corresponding actual user signal estimated value, and selecting one from each actual user signal estimated value through a hard decision module as an output through a hard decision mode by the hard decision module.
In this embodiment, with reference to specific data, Matlab is used to perform simulation and comparative analysis on the conventional MMSE-SIC detection method, MMSE-PIC detection method, and BER error performance in AWGN channel according to the present invention, where simulation parameters are set as shown in table 1, and performance simulation results are shown in fig. 5. According to simulation results, the performance of the error compensation multi-user parallel improved detection method is superior to that of the MMSE-SIC detection method when the SNR is larger than 10dB, and compared with the MMSE-PIC detection performance, the performance of the error compensation multi-user parallel improved detection method is greatly improved. With signal to noise ratioThe more accurate the improvement algorithm is, the more. The complexity of matrix inversion in the MMSE-SIC detection method is O (M)3) The complexity of the whole algorithm is proportional to O (M)4+M3N), and in the improved method, the complexity of the eigenvalue eigenvector decomposition of the matrix is O (M)3) In each case
Figure BDA0002191814890000101
The calculation of (a) involves MN multiplications, and the calculation of a-in each case involves 8+8N multiplications, so the complexity of the overall improved algorithm is proportional to O (M)3). For a small number of users, the improved algorithm has little advantage in complexity, but when the number of users is increased, the complexity of the improved method is greatly reduced, and the complexity of the improved method calculation is reduced by one order of magnitude compared with the complexity of the MMSE-SIC method, which is very beneficial to 5G massive connection and low-delay scenes.
Table 1 simulation parameter settings
Figure BDA0002191814890000111
The invention can solve the problems of the existing alternative multi-user detection method, the MMSE-SIC detection algorithm relates to matrix inversion, the algorithm complexity is high, and when the number of users is large, the detection operation amount is large, so that the processing time delay is large; the MMSE-PIC detection algorithm has few detection levels, and the detection performance is poor although the time delay is small; and quasi-parallel interference elimination detection algorithm, although the processing time delay is smaller than MMSE-SIC, the performance is better than MMSE-PIC, but the detection performance is still worse than MMSE-SIC. The invention is an improvement of MMSE-PIC detection method, which only carries out primary detection and has poor performance, and estimates the error of the signal by compensating MMSE criterion, projects the related noise to the characteristic vector space, and searches the signal in the direction of noise enhancement, thereby obtaining better detection performance, improving the detection accuracy and reducing the bit error rate. The invention is applied to multi-user detection of the MUSA system, improves the detection energy efficiency of the system and realizes green communication.
Those skilled in the art will appreciate that all or part of the steps in the methods of the above embodiments may be implemented by associated hardware instructed by a program, which may be stored in a computer-readable storage medium, and the storage medium may include: ROM, RAM, magnetic or optical disks, and the like.
The above-mentioned embodiments, which further illustrate the objects, technical solutions and advantages of the present invention, should be understood that the above-mentioned embodiments are only preferred embodiments of the present invention, and should not be construed as limiting the present invention, and any modifications, equivalents, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (9)

1. An error-compensated multi-user detection method for a MUSA system, the method comprising:
s1, after modulating the user data of the transmitting end, randomly selecting a plurality of spreading sequences to spread the respective modulation symbols, and transmitting the spreading sequences from the same time-frequency resource;
s2, the receiving end carries out parallel interference elimination detection on the received signal y through an MMSE detector, and calculates a weight matrix omegaMMSEObtaining the modulation symbol estimation value of the user signal at the transmitting end
Figure FDA0003013096900000011
S3, decomposing the eigenvalue of the eigenvalue matrix in the weight matrix to obtain the eigenvalue lambdalCorresponding feature vector vlAnd constructing equivalent noise a by Hermitian matrix of the feature vector and the noisel(ii) a l ═ 1, 2.., M }, where M denotes the number of users at the transmitting end;
s4, selecting the first N larger eigenvalues lambdakCorresponding feature vector vkAnd equivalent noise akCalculating the error e between each modulation symbol corresponding to the user signal of the transmitting terminal and the estimated modulation symbol estimated valuel(m)；k＝{1,2,...,N}；
S5, adopting Lagrange multiplier method to make equivalent noise matrix aHa minimization using calculated error and eigenvalue lambdakAnd a feature vector vkCalculating the estimation value of the equivalent noise matrix corresponding to each modulation symbol
Figure FDA0003013096900000012
S6, adopting maximum likelihood estimation mode to estimate the value according to the equivalent noise matrix
Figure FDA0003013096900000013
Readjusting error el(m) and using it to compensate for errors formed by the detection algorithm, thereby calculating the actual user signal estimate
Figure FDA0003013096900000014
And selects the actual user signal estimate in which the error is the smallest
Figure FDA0003013096900000015
As a final output;
superscript H denotes the Hermitian matrix; m represents the number of symbols corresponding to the modulation scheme.
2. The method of claim 1, wherein the eigenvalue decomposition of the eigen matrix in the weight matrix in step S3 includes that the weight matrix is ωMMSE＝(HHH+σ2I)-1HHLet the feature matrix R be (H)HH+σ2I)-1Carrying out eigenvalue eigenvector decomposition on the characteristic matrix R to obtain R ═ VDVH(ii) a H is a channel matrix of a user at a transmitting end; hHA Hermitian matrix which is a channel matrix H; sigma2Representing the variance of the noise; i represents an identity matrix; v represents that the characteristic vector matrix comprises an M multiplied by M dimensional unitary matrix formed by M orthogonal M multiplied by 1 dimensional column characteristic vectors; d represents that the eigenvalue matrix comprises MM multiplied by M diagonal arrays formed by characteristic values; vHA Hermitian matrix representing the feature vector matrix, wherein an equivalent noise matrix including M-dimensional equivalent noise is defined as a-VHn'，n'＝(n'1,n'2,...,n'l,...,n'M)H，n'lRepresenting the noise signal carried by the hypothetical ith transmitting end user, and the superscript H representing the Hermitian matrix.
3. The method of claim 1, wherein each modulation symbol corresponding to the user signal at the transmitting end is associated with an estimated modulation symbol value
Figure FDA0003013096900000021
The error between is expressed as
Figure FDA0003013096900000022
el(m) when the mth symbol sent by the ith transmitting end user is represented, the modulation symbol estimated value correspondingly estimated is related to the mth symbol
Figure FDA0003013096900000023
The error between; bl(m) represents the mth symbol that the ith transmitting end user may send;
Figure FDA0003013096900000024
indicating the modulation symbol estimated value sent by the ith transmitting terminal user; (v)k)lRepresenting a feature vector vkThe ith element of (1).
4. The method of claim 1 wherein the equivalent noise matrix is estimated for each modulation symbol
Figure FDA0003013096900000025
The calculation formula of (2) is as follows:
Figure FDA0003013096900000026
wherein, C ═ C1,c2,...,cl,...,cM]，
Figure FDA0003013096900000027
(·)↑Represents Moore-Penrose pseudo-inverse; e ═ E1 *(m),e2 *(m),...,el *(m),...,eM *(m)]H(ii) a Denotes a conjugate symbol;
Figure FDA0003013096900000028
bl(m) denotes the mth symbol that the ith transmitting end user may send,
Figure FDA0003013096900000029
Figure FDA00030130969000000210
indicating the modulation symbol estimated value sent by the ith transmitting terminal user; (v)k)lRepresenting a feature vector vkThe l element of (1); the superscript H denotes the Hermitian matrix.
5. The method of claim 4 wherein the equivalent noise matrix is estimated for each modulation symbol
Figure FDA00030130969000000211
The calculation formula of (2) further comprises each modulation symbol b corresponding to the user signal of the transmitting endl(m) separately comparing the estimated modulation symbol estimates
Figure FDA00030130969000000212
The error between them is used as constraint condition, and the Lagrange is usedRidge multiplier method minimization cost function f [ a ]]Selecting two transmitting end users l1And l2For the estimated value
Figure FDA0003013096900000031
Carrying out estimation; the process of minimizing the cost function by the Lagrange multiplier method comprises the following steps:
Figure FDA00030130969000000311
s.t.
Figure FDA0003013096900000032
Figure FDA0003013096900000033
Figure FDA0003013096900000034
wherein, f [ a ]]Representing an estimated value
Figure FDA0003013096900000035
The cost function of (2); gamma ray1And gamma2All represent lagrangian coefficients; e.g. of the typel1(m) denotes the l-th in the actual user signal1Error corresponding to the mth symbol possibly transmitted by a transmitting end user, el2(m) denotes the l-th in the actual user signal2The error corresponding to the mth symbol possibly sent by each transmitting end user represents a conjugate symbol;
Figure FDA0003013096900000036
Figure FDA0003013096900000037
Figure FDA0003013096900000038
subscriptl1And subscriptsl2Represents two different transmitting end users belonging to {1, 2.., M }; (v)1)l1Representing a feature vector v1L. 11An element; (v)1)l2Representing a feature vector v1L. 12And (4) each element.
6. The method of claim 5 wherein the transmitting end user/, is used in the MUSA system1And l2The selection process comprises the steps of calculating and sorting the signal to interference plus noise ratio (SINR) of each transmitting end user, and selecting two users with the minimum SINR as transmitting end users l1And transmitting end user l2(ii) a The calculation formula of the SINR is shown as follows:
Figure FDA0003013096900000039
wherein, ω isi,MMSEDenoted is the ith row, h, of the weight matrixiRepresenting the ith column, σ, of the equivalent channel matrix H2Is the noise variance, | | | · | | represents the Frobenius norm.
7. The method of claim 1 wherein the final output of the actual user signal estimate of step S6 is the actual user signal estimate
Figure FDA00030130969000000310
Expressed as:
Figure FDA0003013096900000041
wherein the content of the first and second substances,
Figure FDA0003013096900000042
representing the estimated calculation; argmin stands for
Figure FDA0003013096900000043
Corresponding independent variable when obtaining minimum value
Figure FDA0003013096900000044
Q (-) denotes a hard decision; i | · | | represents the Frobenius norm; h is the channel matrix of the user at the transmitting end.
8. An error-compensated multi-user detection apparatus for a MUSA system, the apparatus comprising:
a transmitting antenna: for transmitting user data;
the modulator: for modulating user data;
a sequence spreading module: for performing sequence spreading on the modulated data;
a receiver: for receiving the modulated spread user data;
MMSE detector: for solving a weight matrix by minimizing a minimum mean square error between the transmit vector and the estimate vector;
a characteristic decomposition module: the characteristic matrix in the weight matrix is used for carrying out characteristic value decomposition; the characteristic value lambda is resolvediCorresponding feature vector vi；
An equivalent noise construction module: the Hermitian matrix of the eigenvector matrix decomposed according to the eigenvector decomposition unit is multiplied by noise;
an error estimation module: taking the first N larger eigenvalues, the corresponding eigenvectors and the equivalent noise to calculate the error between each modulation symbol corresponding to the user signal of the transmitting end and the estimated modulation symbol estimated value;
lagrange multiplier module: the method is used for calculating the estimation value of the equivalent noise matrix corresponding to each symbol according to a Lagrange multiplier method;
an error calculation module: the system is used for calculating actual user signals according to the estimated values of all equivalent noise matrixes calculated by the Lagrange multiplier method module;
a hard decision module: for hard decision of the calculated actual user signal and selecting the error with the minimum as the output.
9. The apparatus of claim 8, wherein the eigen decomposition module comprises an eigenvalue decomposition unit and an eigenvector decomposition unit; the characteristic value decomposition unit is used for decomposing an M multiplied by M diagonal matrix consisting of M characteristic values; the eigenvector decomposition unit is used for decomposing an M multiplied by M unitary matrix formed by M orthogonal M multiplied by 1 column eigenvectors; m represents the number of users at the transmitting end.
CN201910834807.8A 2019-09-05 2019-09-05 Error compensation multi-user detection method and device for MUSA (multiple input multiple output) system Active CN110492956B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910834807.8A CN110492956B (en) 2019-09-05 2019-09-05 Error compensation multi-user detection method and device for MUSA (multiple input multiple output) system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910834807.8A CN110492956B (en) 2019-09-05 2019-09-05 Error compensation multi-user detection method and device for MUSA (multiple input multiple output) system

Publications (2)

Publication Number Publication Date
CN110492956A CN110492956A (en) 2019-11-22
CN110492956B true CN110492956B (en) 2021-07-06

Family

ID=68556589

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910834807.8A Active CN110492956B (en) 2019-09-05 2019-09-05 Error compensation multi-user detection method and device for MUSA (multiple input multiple output) system

Country Status (1)

Country Link
CN (1) CN110492956B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114665934B (en) * 2022-03-18 2023-05-12 五邑大学 Signal detection method, system, equipment and medium based on decomposition factor matrix

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102891815A (en) * 2012-09-19 2013-01-23 北京航空航天大学 Estimation method for low-complexity channel in time division duplex (TDD) multi-base station cooperative system
CN106059968A (en) * 2016-05-27 2016-10-26 重庆邮电大学 MUSA system multistage portion parallel interference elimination multi-user detection method
CN107809299A (en) * 2017-10-23 2018-03-16 哈尔滨工业大学 After the first string of multiple users share access technology up-link and multi-user test method

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10868602B2 (en) * 2018-01-19 2020-12-15 Lenovo (Singapore) Pte. Ltd. Channel compression matrix parameters

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102891815A (en) * 2012-09-19 2013-01-23 北京航空航天大学 Estimation method for low-complexity channel in time division duplex (TDD) multi-base station cooperative system
CN106059968A (en) * 2016-05-27 2016-10-26 重庆邮电大学 MUSA system multistage portion parallel interference elimination multi-user detection method
CN107809299A (en) * 2017-10-23 2018-03-16 哈尔滨工业大学 After the first string of multiple users share access technology up-link and multi-user test method

Also Published As

Publication number Publication date
CN110492956A (en) 2019-11-22

Similar Documents

Publication Publication Date Title
CN107483088B (en) Large-scale MIMO robust precoding transmission method
US6987819B2 (en) Method and device for multiple input/multiple output transmit and receive weights for equal-rate data streams
KR101381130B1 (en) Multi-user precoding and scheduling method and base station for implementing the method
US9077516B2 (en) Reduced complexity frequency domain equalization of multiple input multiple output channels
US8077788B2 (en) Soft demapping for MIMO decoding
CN107947840B (en) Time reversal anti-interference method based on millimeter wave large-scale MIMO
KR100918717B1 (en) Sequence estimating method and device in mimo ofdm communication system
CN111698182A (en) Time-frequency blocking sparse channel estimation method based on compressed sensing
JP4966190B2 (en) Method and apparatus for transmitting a signal in a multi-antenna system, signal and method for estimating a corresponding transmission channel
WO2010029765A1 (en) Wireless transmitter and precoding method
KR20070122511A (en) Method and apparatus for singular value decomposition of a channel matrix
CN110166401B (en) Phase noise suppression method of large-scale MIMO orthogonal frequency division multiplexing system
Jeya et al. Optimized semiblind sparse channel estimation algorithm for MU-MIMO OFDM system
WO2015112883A1 (en) System and method for early termination in iterative null-space directed singular value decomposition for mimo
US9031123B2 (en) Communication system and method using subspace interference cancellation
JP2010502140A (en) Equalization structure and equalization method
CN114285444A (en) Power optimization method for large-scale de-cellular MIMO system
WO2017132984A1 (en) Method and apparatus of topological pilot decontamination for massive mimo systems
CN108566227B (en) Multi-user detection method
CN110492956B (en) Error compensation multi-user detection method and device for MUSA (multiple input multiple output) system
CN108600125B (en) Channel estimation method based on iteration
CN114124168A (en) MIMO-NOMA system signal detection method and system based on deep learning
Raja et al. Improved spectral efficiency in massive MIMO ultra-dense networks through optimal pilot-based vector perturbation precoding
Banister et al. Feedback assisted stochastic gradient adaptation of multiantenna transmission
JP5288622B2 (en) Wireless communication apparatus, wireless communication system, and communication method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant