CN117792546A - Distributed MIMO signal detection method and system based on general recursive least square - Google Patents
Distributed MIMO signal detection method and system based on general recursive least square Download PDFInfo
- Publication number
- CN117792546A CN117792546A CN202410006694.3A CN202410006694A CN117792546A CN 117792546 A CN117792546 A CN 117792546A CN 202410006694 A CN202410006694 A CN 202410006694A CN 117792546 A CN117792546 A CN 117792546A
- Authority
- CN
- China
- Prior art keywords
- matrix
- channel
- vector
- distributed
- recursive
- 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.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 56
- 239000011159 matrix material Substances 0.000 claims abstract description 327
- 238000000034 method Methods 0.000 claims abstract description 80
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 75
- 238000012545 processing Methods 0.000 claims abstract description 53
- 238000013139 quantization Methods 0.000 claims abstract description 15
- 230000008569 process Effects 0.000 claims description 46
- 238000004364 calculation method Methods 0.000 claims description 23
- 238000013016 damping Methods 0.000 claims description 5
- 238000004891 communication Methods 0.000 claims description 4
- 230000014759 maintenance of location Effects 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 description 20
- 238000010586 diagram Methods 0.000 description 9
- 238000004590 computer program Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 4
- 230000009191 jumping Effects 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 238000009826 distribution Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 208000026753 anterior segment dysgenesis Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Landscapes
- Radio Transmission System (AREA)
Abstract
The invention discloses a distributed MIMO signal detection method and system based on a general recursive least square, which comprises initializing related parameters of a general recursive least square algorithm, wherein the related parameters comprise an estimated vector, a channel inverse matrix and a diagonalization parameter C 0 The method comprises the steps of carrying out a first treatment on the surface of the For front C 0 The distributed units are used for continuously updating the estimated vector and the channel inverse matrix by utilizing the last channel inverse matrix and the estimated vector, and the local channel matrix and the received signal based on a general recursive least square algorithm without antenna constraint; for post C-C 0 The distributed units simplify the general recursive least square algorithm based on the channel hardening characteristic of the MIMO system, and continuously update the estimated vector by using the last channel inverse matrix and the estimated vector and the local channel matrix and the received signal; and carrying out quantization processing on the finally obtained current estimated vector by using a central processing unit. The invention can realizeThe excellent compromise of performance, complexity and bandwidth is suitable for the practical application scene of various MIMO systems.
Description
Technical Field
The invention belongs to the technical field of MIMO signal detection, and particularly relates to a distributed MIMO signal detection method and system based on universal recursion least square.
Background
The large-scale multiple input multiple output (Multiple Input Multiple Output, MIMO) system has the advantages of large capacity, ultra-high data transmission rate, high energy efficiency and the like, and gradually becomes a key technology in B5G or even 6G wireless communication. However, most existing detection schemes are typically implemented in a centralized fashion. With the increasing number of antennas at the base station side, even with the most advanced hardware support, there is an urgent challenge in transmitting large amounts of raw data for signal processing. Meanwhile, the rapid increase in computational complexity and data storage requirements also makes it difficult for a single computing chip to meet practical requirements. To address these issues, researchers have proposed three distributed frameworks and a series of distributed detection algorithms, a distributed baseband processing, a fully distributed feed forward, and a daisy chain. Wherein the ADMM, CG and EPA algorithms under a decentralized baseband processing framework require multiple interactions of the distributed units with the central processor, resulting in high bandwidth and complexity costs. MMSE and GMP algorithms under a fully distributed feed forward framework always suffer from large performance losses. The existing RLS, SGD and ASGD algorithms of the daisy chain framework limit the number of antennas of the distributed units, and are not practical and generally poor in performance. Therefore, in most MIMO system application scenarios, the existing distributed detection algorithms cannot realize good trade-offs among performance, computational complexity and transmission bandwidth. None of these conventional distributed detection schemes is applicable when it is desired to meet both lower complexity, bandwidth and higher accuracy requirements.
Therefore, from the practical application point of view, a signal detection scheme suitable for a distributed massive MIMO system is urgently needed to achieve an excellent compromise in performance, complexity and bandwidth.
Disclosure of Invention
Aiming at the problems, the invention provides a distributed MIMO signal detection method and system based on the general recursion least square, which not only can realize excellent compromise of performance, complexity and bandwidth, but also has convergence characteristics of theoretical support, and can adapt to the actual application scenes of various MIMO systems.
In order to achieve the technical purpose and achieve the technical effect, the invention is realized by the following technical scheme:
in a first aspect, the present invention provides a distributed MIMO signal detection method based on a common recursive least squares, including:
initializing correlation parameters of a generic recursive least squares algorithm, said correlation parameters comprising an estimation vector, a channel inverse matrix and a diagonalization parameter C 0 ；
For front C 0 The distributed units are sequentially utilized to update the estimated vector and the channel inverse matrix by utilizing the last distributed unit or the initialized channel inverse matrix and the estimated vector, and the local channel matrix and the received signal based on a general recursion least square algorithm without antenna constraint;
For post C-C 0 The distributed units are sequentially utilized to simplify a general recursive least square algorithm based on the channel hardening characteristic of the MIMO system, the channel inverse matrix and the estimated vector of the last distributed unit are continuously utilized, the local channel matrix and the received signal are utilized to update the estimated vector, and C is the total number of the distributed units;
and carrying out quantization processing on the finally obtained current estimated vector by using a central processing unit to obtain a corresponding constellation point.
Optionally, the local channel matrix is obtained by performing channel estimation by using pilot signals, where the dimension of the local channel matrix is b×k, B is the number of receiving antennas that a single distributed unit has, b=n/C, C is the total number of distributed units, N is the total number of receiving antennas, and K is the total number of transmitting antennas.
Optionally, the initializing relevant parameters of the general recursive least square algorithm includes:
setting an initial value of an estimation vector to x 0 =0, dimension k×1;
setting the initial value of the channel inverse matrix to P 0 ＝I K ，I K A unit matrix with K multiplied by K dimension;
setting the initial value of the diagonalization parameter as:
wherein,representing rounding to the nearest integer.
Optionally, the general recursive least square algorithm based on the non-antenna constraint continuously uses the last distributed unit or initialized channel inverse matrix and the estimated vector, and updates the estimated vector and the channel inverse matrix by the local channel matrix and the received signal, including:
Initializing the general recursive least square algorithm with the c=1st distributed unit as the start point and the C as the end point 0 The distributed units sequentially perform channel detection tasks;
repeating the preset loop step until c=c 0 The preset circulation steps comprise: in the c-th distributed unit, the channel inverse matrix P of the last distributed unit is utilized c-1 And estimate vector x c-1 Local channel matrix H c Sequentially calculating a current process matrix R c Channel inverse matrix P c And estimate vector x c The method comprises the steps of carrying out a first treatment on the surface of the Comparison of C and C 0 If c is the size of<C 0 Then inverse matrix P of current channel c And the current estimate vector x c Pass on to the next distributed unit, recursive sequence number c=c+1;
diagonalizing the current channel inverse matrix P c Obtaining Wherein diag (diag (P) c ) For diagonalization operations, a matrix P is reserved c Diagonal elements of (a)Whereas the non-diagonal elements are set to zero, the diagonal matrix +.>Is K x K;
inverse matrix of current channelAnd the current estimate vector x c Pass on to the next distributed unit, recursive sequence number c=c+1.
Optionally, the c-th distributed unit uses the channel inverse matrix P of the last distributed unit c-1 And estimate vector x c-1 Local channel matrix H c-1 Sequentially calculating a current process matrix R c Channel inverse matrix P c And estimate vector x c The method specifically comprises the following steps:
channel inverse matrix P based on last distributed unit c-1 And a local channel matrix H c For process matrix R c Updating:
wherein I is B Is a B x B dimension identity matrix, matrix R c Is B x B;
channel inverse matrix P based on last distributed unit c-1 Local channel matrix H c And a current process matrix R c Inverse matrix P of current channel c Updating:
wherein matrix P c The dimension is K multiplied by K;
based on the current channel inverse matrix P c Local channel matrix H c And the estimated vector x of the last distributed unit c-1 Current estimation vector x c Updating:
wherein the vector x is estimated c Is K x 1.
Optionally, the channel hardening characteristic based on the MIMO system simplifies the general recursive least square algorithm, continuously uses the channel inverse matrix and the estimated vector of the last distributed unit, and updates the estimated vector by the local channel matrix and the received signal, specifically:
initializing the start of the generic recursive least squares algorithm to be c=c 0 +1 distributed units, wherein the end point is the C-th distributed unit, and channel detection tasks are sequentially carried out;
repeating a preset circulation step until c=c, wherein the preset circulation step comprises using the channel inverse matrix of the last distributed unit in the C-th distributed unit And estimate vector x c-1 And a local channel matrix H c Sequentially calculating the current process matrix +.>Channel inverse matrix->And estimate vector x c The method comprises the steps of carrying out a first treatment on the surface of the For the current estimated vector x c And the estimated vector x of the last distributed unit c-1 And (3) performing linear weighted sum, setting a damping factor as beta, and specifically, the method comprises the following steps of: x is x c ＝βx c-1 +(1-β)x c The method comprises the steps of carrying out a first treatment on the surface of the Comparing the sizes of C and C, if C<C, inverse matrix of current channel +.>And the current estimate vector x c Pass on to the next distributed unit, recursive sequence number c=c+1;
estimating the current directionQuantity x C To the central processing unit.
Optionally, the c-th distributed unit uses the channel inverse matrix of the last distributed unitAnd estimate vector x c-1 Local channel matrix H c Sequentially calculating the current process matrix +.>Channel inverse matrix->And estimate vector x c The method specifically comprises the following steps:
channel inverse matrix based on last distributed unitAnd a local channel matrix H c Process matrix->Updating:
wherein I is B Is an identity matrix of dimension B x B,is a diagonal matrix of dimension K x K, calculating +.>In the process of (1), only diagonal elements are calculated and reserved, and non-diagonal elements are set to be zero;
channel inverse matrix based on last distributed unitLocal channel matrixH c And the current process matrix- >Inverse matrix of current channel->Updating;
wherein,is a diagonal matrix of dimension K x K, < >>Is a diagonal matrix of dimension B x B, calculating +.>In the process of (1), only diagonal elements are calculated and reserved, and non-diagonal elements are set to be zero;
based on the inverse matrix of the current channelLocal channel matrix H c And the last estimated vector x c-1 Current estimation vector x c Updating:
wherein the vector x is estimated c Is K x 1 due to the dimensions ofIs a diagonal matrix, and the corresponding matrix multiplication only requires calculation and retention of diagonal elements.
Optionally, the quantization processing is performed on the finally obtained current estimated vector by using a central processing unit, and the adopted calculation formula is as follows:
wherein,output signal representing central processing unit, +.>Representing rounding to nearest the current estimate vector x finally obtained C Is a constellation point of (a).
In a second aspect, the present invention provides a distributed MIMO signal detection system based on a common recursive least squares, comprising: a central processing unit and a number of distributed units in communication with the central processing unit;
in the first distributed unit, the correlation parameters of the generic recursive least squares algorithm are initialized, including the estimated vector, the channel inverse matrix and the diagonalization parameters C 0 ；
For front C 0 The distributed units are sequentially based on a general recursive least square algorithm without antenna constraint, and continuously update the estimated vector and the channel inverse matrix by using the last distributed unit or the initialized channel inverse matrix and the estimated vector, and the local channel matrix and the received signal;
for post C-C 0 The distributed units simplify the general recursive least square algorithm based on the channel hardening characteristic of the MIMO system in sequence, and continuously update the estimated vector by using the channel inverse matrix and the estimated vector of the last distributed unit and the local channel matrix and the received signal;
the central processing unit is used for carrying out quantization processing on the finally obtained current estimated vector to obtain a corresponding constellation point.
Optionally, each distributed unit includes a radio frequency processing module, a channel estimation module, and a signal detection module;
the radio frequency processing module processes the received uplink signals to obtain pilot signals and local receiving signals;
the channel estimation module estimates a local channel matrix by utilizing pilot signals and sends the local channel matrix to the signal detection module;
for front C 0 The signal detection module is based on a general recursive least square algorithm without antenna constraint, and continuously updates the estimated vector and the channel inverse matrix by using the last distributed unit or the initialized channel inverse matrix and the estimated vector, and the local channel matrix and the received signal;
For post C-C 0 A distributed unit, the signal detection module simplifies the general recursive least square algorithm based on the channel hardening characteristic of the MIMO system, continuously uses the channel inverse matrix and the estimated vector of the last distributed unit, and updates the estimated vector by the local channel matrix and the received signal, and finally calculates the estimated vector x obtained by the C distributed unit C To the central processing unit.
Compared with the prior art, the invention has the beneficial effects that:
the method provided by the invention has no constraint on the quantity of the distributed unit antennas, and the complexity and the data bandwidth are obviously reduced by utilizing the channel hardening characteristic of a large-scale MIMO system, so that a flexible distributed detection scheme is realized, and excellent compromise among the performance, the complexity and the bandwidth is realized.
The complexity of the general recursive least square algorithm provided by the invention mainly depends on the complexity of an iterative formula because the general recursive least square algorithm belongs to the iterative algorithm. The calculation complexity is counted by the required complex multiplication times, and the complexity required for calculating the inverse matrix of the K multiplied by K dimension matrix is considered to be 0.5K 3 . The complexity of a single iteration depends mainly on the complexity of the following 3 computations: (1) Calculating a process matrix R c Or (b)(2) Calculating channel inverse matrixP c Or->(3) Calculating an estimated vector x c . Specifically, the computational complexity of the general recursive least square algorithm proposed by the present invention is composed of two parts. For the first stage, i.e. front C 0 In the iterations, a matrix R is calculated c 、P c Sum vector x c The required complexity is K 2 B+KB 2 +0.5B 3 、K 2 B+KB 2 And K 2 B+2kb. For the second stage, i.e. post C-C 0 In the next iteration, the matrix is calculated>Sum vector x c Only the diagonal elements need to be calculated and reserved for matrix multiplication, and matrix inversion also only needs to calculate the reciprocal of the diagonal elements, so that the required calculation amount is greatly reduced, and the calculation complexity is respectively 2KB+B, 2KB and 3KB. Thus, the total computational complexity required for the iterations of the present invention is (3K 2 B+2KB 2 +2KB+0.5B 3 )C 0 +(7KB+B)(C-C 0 ). When there are far more receive antennas than transmit antennas, e.g., n=256, k=16 or 32, the total number of distributed units C > C 0 A significant reduction in computational complexity will result compared to conventional recursive least squares algorithms.
The general recursive least square algorithm provided by the invention belongs to an iterative algorithm and belongs to a distributed algorithm, so that the bandwidth on an interconnection link needs to be considered mainly. The data transmission bandwidth is determined by the average complex value transmitted on each link, which can represent the actual overhead of the hardware interface. The transmission bandwidth is mainly dependent on the size of the following 2 items of data: (1) Transmission channel inverse matrix P c Or (b)(2) Transmission estimation vector x c . Specifically, the data transmission bandwidth of the universal recursive least square algorithm provided by the invention is composed of two parts. For the first stage, i.e. front C 0 In several iterations, the message is transmittedChannel inverse matrix P c And estimate vector x c The required bandwidth is K 2 +K. For the second stage, i.e. post C-C 0 In the next iteration, the diagonal matrix is transmitted>Only the diagonal elements thereof need to be transmitted in practice, so the required bandwidth is reduced to K. In addition, transfer x c Is K. The data transmission bandwidth required for the inventive iteration is therefore ((K) 2 +K)C 0 +2K(C-C 0 ) and/C. In a massive MIMO scenario, e.g., n=256, k=16 or 32, the total number of distributed units C > C 0 The required data transmission bandwidth will be greatly reduced compared to conventional recursive least squares algorithms.
In many distributed detection schemes, in order to ensure convergence, the application scenario is limited in practical use, for example, the number of distributed unit antennas is required to satisfy the condition B > K. The general recursive least square algorithm provided by the invention is characterized in that the parameter C 0 When the following conditions are met, convergence to a centralized MMSE detection result can be ensured, and more various network environments can be adapted:
the invention is applicable to solving problems of all equation sets shaped as y=hx+n. The invention is mainly applied to the field of uplink distributed signal detection. If properly adapted, should also be applicable to the calculation of the downlink beamforming problem.
Drawings
For a clearer description of an embodiment of the invention or of the solutions of the prior art, the drawings that are needed in the embodiment will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art, in which:
fig. 1 is a schematic diagram of a scenario of a massive MIMO system of the present invention based on a common recursive least squares distributed MIMO signal detection method;
FIG. 2 is a diagram of a daisy chain distributed system architecture of a distributed MIMO signal detection method based on a common recursive least squares in accordance with the present invention;
FIG. 3 is a flow chart of a method for detecting distributed MIMO signals based on the general recursive least squares according to the present invention;
FIG. 4 is a diagram showing a comparison of the convergence of a distributed MIMO signal detection method based on the general recursive least squares in accordance with the present invention;
FIG. 5 is a graph showing the comparison of error rates between different distributed algorithms in a distributed MIMO signal detection method based on the general recursive least square;
fig. 6 is a comparison chart of bit error rates under different parameter choices of a distributed MIMO signal detection method based on a common recursive least square according to the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In addition, if there is a description of "first", "second", etc. in the embodiments of the present invention, the description of "first", "second", etc. is for descriptive purposes only and is not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In addition, the technical solutions of the embodiments may be combined with each other, but it is necessary to base that the technical solutions can be realized by those skilled in the art, and when the technical solutions are contradictory or cannot be realized, the combination of the technical solutions should be considered to be absent and not within the scope of protection claimed in the present invention.
Example 1
As shown in fig. 1, the complex MIMO system includes K transmitting antennas and N receiving antennas, and the relationship between a signal y received by the receiving antennas and a signal x transmitted by the transmitting antennas is shown as follows:
y＝Hx+n
wherein H represents a channel matrix, and the dimension is NxK; n represents Gaussian noise interference, the dimension is Nx1, the obeying mean is 0, and the variance isComplex gaussian distribution of (a); x represents a transmission signal transmitted by a user side, the dimension is Kx1, each element of the transmission signal belongs to a constellation point set which is modulated by M-QAM +.>The decoding candidate point set is +.>y represents a received signal at the base station side, and the dimension is n×1.
As shown in fig. 2, the base station side is connected in series using a daisy chain distribution framework. The N receiving antennas at the base station side are uniformly divided into C distributed units, wherein each distributed unit is configured with b=n/C antennas and one independent hardware computing unit. The uplink signals received by the B receiving antennas of the distributed units are processed by the radio frequency processing module to obtain pilot signals and local receiving signals y c The channel estimation module estimates the local channel matrix H by using the pilot frequency c Local channel matrix H c Is B x K. The distributed unit then uses the local channel matrix H c Local received signal y c (dimension B x 1) and transfer from last distributed unit to estimate vector x c And channel inverse matrix P c-1 The method for detecting the signal by adopting the universal recursive least square is used for detecting the signal. Finally, the final estimated vector of the C-th distributed unit is transferred to the central processing unitAnd the device completes the channel decoding task through simple quantization operation.
The traditional distributed recursive least square algorithm calculates the complete channel inverse matrix to detect the signal, and calculates in the c-th distributed unit:
in the formula, h c And y c A local channel matrix and a local received signal in the 1 xk dimension, respectively.
However, this conventional recursive least squares method limits the number of antennas of the distributed unit, i.e., b=1. If the base station side has hundreds or thousands of receiving antennas, the conventional recursive least square algorithm needs to be equipped with the same number of hardware processing units, which brings about huge expenditure and limits the application range. On the other hand, the channel inverse matrix P c Requiring the computation of multiple high-dimensional matrix multiplications, which can result in high complexity costs. Furthermore, the estimated vector x is updated c When P of the last distributed unit is needed c-1 And x c-1 This requires the transmission of a matrix of K x K dimensions and a vector of K x 1 dimensions, resulting in a large amount of data bandwidth. As the number of transmit antennas on the user side increases, the computational complexity and transmission bandwidth required increases rapidly, and is not affordable by the existing hardware computational power and data transmission capabilities.
For this purpose, in the embodiment of the present invention, it is proposed that: the traditional recursive least square algorithm in the uplink large-scale MIMO distributed detection scene is expanded, the limit of the number of the antennas of the distributed units is eliminated, and the distributed units can be flexibly selected according to actual requirements. Meanwhile, the invention utilizes the channel hardening characteristic to perform diagonal approximation on the matrix, greatly simplifies the operations of high-dimensional matrix multiplication and matrix inversion required by calculation, and reduces the required calculation complexity and transmission bandwidth, thereby realizing excellent compromise in calculation precision and cost.
Specifically, in an embodiment of the present invention, a distributed MIMO signal detection method based on universal recursive least squares is provided, as shown in fig. 3, including the following steps:
initializing correlation parameters of a generic recursive least squares algorithm, said correlation parameters comprising an estimation vector, a channel inverse matrix and a diagonalization parameter C 0 ；
For front C 0 The distributed units are utilized to continuously update the estimated vector and the channel inverse matrix by utilizing the last distributed unit or the initialized channel inverse matrix and the estimated vector, and the local channel matrix and the received signal sequentially based on a general recursive least square algorithm without antenna constraint;
for post C-C 0 The distributed units are utilized to simplify a general recursive least square algorithm based on the channel hardening characteristic of the MIMO system in sequence, the channel inverse matrix and the estimated vector of the last distributed unit are continuously utilized, the local channel matrix and the received signal are utilized to update the estimated vector, and C is the total number of the distributed units;
and carrying out quantization processing on the finally obtained current estimated vector by using a central processing unit to obtain a corresponding constellation point.
In a specific implementation manner of the embodiment of the present invention, the local channel matrix is obtained by performing channel estimation using a pilot signal, where the dimension of the local channel matrix is b×k, B is the number of receiving antennas that a single distributed unit has, b=n/C, C is the total number of distributed units, and N is the total number of receiving antennas.
In a specific implementation manner of the embodiment of the present invention, as shown in fig. 3, the initializing relevant parameters of the general recursive least square algorithm includes:
Setting an initial value of an estimation vector to x 0 =0, dimension k×1;
setting the initial value of the channel inverse matrix to P 0 ＝I K ，I K In units of dimension K x KA matrix;
setting the initial value of the diagonalization parameter as:
wherein,representing rounding to the nearest integer.
In a specific implementation manner of the embodiment of the present invention, the general recursive least square algorithm based on no antenna constraint continuously uses the channel inverse matrix and the estimated vector of the last distributed unit or initialization, and updates the estimated vector and the channel inverse matrix by using the local channel matrix and the received signal, including:
initializing the general recursive least square algorithm with the c=1st distributed unit as the start point and the C as the end point 0 The distributed units sequentially perform channel detection tasks;
repeating the preset loop step until c=c 0 The preset circulation steps comprise: in the c-th distributed unit, the channel inverse matrix P of the last distributed unit is utilized c-1 And estimate vector x c-1 Local channel matrix H c Sequentially calculating a current process matrix R c Channel inverse matrix P c And estimate vector x c The method comprises the steps of carrying out a first treatment on the surface of the Comparison of C and C 0 If c is the size of<C 0 Then inverse matrix P of current channel c And the current estimate vector x c Pass on to the next distributed unit, recursive sequence number c=c+1;
diagonalizing the current channel inverse matrix P c Obtaining Wherein diag (diag (P) c ) For diagonalizing operations, reserve momentsArray P c Is set to zero, the diagonal matrix +.>Is K x K;
inverse matrix of current channelAnd the current estimate vector x c Pass on to the next distributed unit, recursive sequence number c=c+1.
In a specific implementation manner of the embodiment of the present invention, as shown in fig. 3, the c-th distributed unit uses the last channel inverse matrix P c-1 And estimate vector x c-1 Local channel matrix H c-1 Sequentially calculating a current process matrix R c Channel inverse matrix P c And estimate vector x c The method specifically comprises the following steps:
channel inverse matrix P based on last distributed unit c-1 And a local channel matrix H c For process matrix R c Updating:
wherein I is B Is a B x B dimension identity matrix, matrix R c Is B x B;
channel inverse matrix P based on last distributed unit c-1 Local channel matrix H c And a current process matrix R c Inverse matrix P of current channel c Updating:
wherein matrix P c The dimension is K multiplied by K;
based on the current channel inverse matrix P c Local channel matrix H c And the estimated vector x of the last distributed unit c-1 For a pair ofCurrent estimation vector x c Updating:
wherein the vector x is estimated c Is K x 1.
In a specific implementation manner of the embodiment of the present invention, as shown in fig. 3, the channel hardening characteristic based on the MIMO system simplifies the general recursive least square algorithm, continuously uses the channel inverse matrix and the estimated vector of the last distributed unit, and updates the estimated vector by the local channel matrix and the received signal, specifically:
initializing the start of the generic recursive least squares algorithm to be c=c 0 +1 distributed units, wherein the end point is the C-th distributed unit, and channel detection tasks are sequentially carried out;
repeating a preset circulation step until c=c, wherein the preset circulation step comprises using the channel inverse matrix of the last distributed unit in the C-th distributed unitAnd estimate vector x c-1 And a local channel matrix H c Sequentially calculating the current process matrix +.>Channel inverse matrix->And estimate vector x c The method comprises the steps of carrying out a first treatment on the surface of the For the current estimated vector x c And the estimated vector x of the last distributed unit c-1 And (3) performing linear weighted sum, setting a damping factor as beta, and specifically, the method comprises the following steps of: x is x c ＝βx c-1 +(1-β)x c The method comprises the steps of carrying out a first treatment on the surface of the Comparing the sizes of C and C, if C <C, inverse matrix of current channel +.>And the current estimated directionQuantity x c Pass on to the next distributed unit, recursive sequence number c=c+1; in a specific implementation, the β=0.1.
The current estimation vector x C To the central processing unit.
In a specific implementation manner of the embodiment of the present invention, as shown in fig. 3, the c-th distributed unit uses the last channel inverse matrixAnd estimate vector x c-1 Local channel matrix H c Sequentially calculating the current process matrix +.>Channel inverse matrix->And estimate vector x c The method specifically comprises the following steps:
based on the last channel inverse matrixAnd a local channel matrix H c Process matrix->Updating:
wherein I is B Is an identity matrix of dimension B x B,is a diagonal matrix of dimension K x K, calculating +.>In the process of (1), only diagonal elements are calculated and reserved, and non-diagonal elements are set to be zero;
channel inverse matrix based on last distributed unitLocal channel matrix H c And the current process matrix->Inverse matrix of current channel->Updating;
wherein,is a diagonal matrix of dimension K x K, < >>Is a diagonal matrix of dimension B x B, calculating +.>In the process of (1), only diagonal elements are calculated and reserved, and non-diagonal elements are set to be zero;
based on the inverse matrix of the current channel Local channel matrix H c And the last estimated vector x c-1 Current estimation vector x c Updating:
wherein the vector x is estimated c Is K x 1 due to the dimensions ofIs a diagonal matrix, and the corresponding matrix multiplication only requires calculation and retention of diagonal elements.
In a specific implementation manner of the embodiment of the present invention, the quantization processing is performed on the finally obtained current estimated vector by using a central processing unit, and the adopted calculation formula is as follows:
wherein,output signal representing central processing unit, +.>Representing rounding to nearest the current estimate vector x finally obtained C Is a constellation point of (a).
The following describes a general recursive least squares-based distributed MIMO signal detection method in an embodiment of the present invention in detail with reference to a specific implementation.
The specific implementation of the distributed MIMO signal detection method according to the embodiment of the present invention is described below by taking a distributed massive MIMO system with a transmitting antenna k=4, a receiving antenna n=32, a distributed unit number c=16, and a modulation scheme of 16-QAM as an example.
Step 1: for a distributed MIMO system comprising K transmitting antennas and N receiving antennas, C distributed units perform channel estimation by using pilot frequency to obtain a local channel matrix H c The dimension is B x K. For a distributed massive MIMO system with transmit antennas k=4 and receive antennas n=32, c=16 distributed units are each assigned to b=n/c=2 receive antennas, the local channel matrix H of the C-th distributed unit c And a local received signal y c C=1, 2,3, …, can be expressed as:
wherein H is c Is 2 x 4, y c Is 2 x 1.h is a i,j Representing the channel response between the jth transmit antenna and the ith receive antenna, y i Representing the received signal on the ith receive antenna.
Step 2: a generic recursive least squares algorithm is initialized. Initializing an estimated vector x 0 Channel inverse matrix P 0 And diagonalization parameter C 0 . The specific operation steps are as follows:
step 2.1 sets the initial value of the estimated vector to a zero vector with dimensions 4 x 1.
Step 2.2, the initial value of the channel inverse matrix is set as a unit diagonal matrix, and the dimension is 4×4.
Step 2.3 setting the initial value of the diagonalization parameter to
Wherein,rounded to the nearest integer less than or equal to a.
Step 3: based on a recursive least square algorithm mechanism of the distributed unit without antenna constraint, the estimation vector is updated continuously by using a local channel matrix and a received signal, and the specific operation is as follows:
Step 3.1 initializing the universal recursive least squares algorithm with the starting point being the c=1st distributed unit and the ending point being the C 0 =4 distributed units, and signal detection tasks are performed sequentially.
Step 3.2 at the c-th distributed unit, the last channel inverse matrix P is used c-1 And estimate vector x c-1 Local channel matrix H c Sequentially calculating a current process matrix R c Channel inverse matrix P c And estimate vector x c . The specific calculation steps are as follows:
step 3.2.1: channel inverse matrix P based on last distributed unit c-1 And a local channel matrix H c For process matrix R c Updating:
wherein I is 2 Is a unit matrix with 2X 2 dimension, and the matrix R is calculated c The dimension is 2×2.
Step 3.2.2: channel inverse matrix P based on last distributed unit c-1 Local channel matrix H c And a current process matrix R c Inverse matrix P of current channel c Updating:
wherein the calculated matrix P c The dimension is 4×4.
Step 3.2.3: based on the current channel inverse matrix P c Local channel matrix H c And the estimated vector x of the last distributed unit c-1 Current estimation vector x c Updating:
wherein the calculated vector x c The dimension is 4×1.
Step 3.3 comparing C and C 0 Size of=4, if c <And 4, jumping to the step 3.4. If c=4, step 3.5 is skipped.
Step 3.4 willChannel inverse matrix P of current 4 x 4 dimensions c And the current 4 x 1 dimension of the estimated vector x c Pass on to the next distributed unit, recursion sequence number c=c+1, jump step 3.2 continues the iteration.
Step 3.5 diagonalizing the current channel inverse matrix P c Order-making
Wherein diag (diag (P) c ) For diagonalization operations, only the matrix Pc is reserved A kind of electronic device Diagonal elements, but not diagonal elements, are zeroed out, specifically:
wherein p is i,j Representation matrix P c The ith row and jth column element of (c).
Step 3.6 inverse matrix of current channelDiagonal vector of 4 x 1 dimensions and estimated vector x of current 4 x 1 dimensions c Pass on to the next distributed unit, recursion sequence number c=c+1, execute step 4.
Step 4: the general recursive least square algorithm is simplified based on the channel hardening characteristic of the large-scale MIMO system, and the estimation vector is updated by continuously utilizing the local channel matrix and the received signal, and the specific operation is as follows:
step 4.1 initializing the starting point of the generic recursive least squares algorithm to be c=c 0 +1=5 distributed units, and the endpoint is the c=16 th distributed unit, and channel detection tasks are performed sequentially.
Step 4.2 at the c-th distributed unit, the channel inverse matrix of the last distributed unit is used And estimate vector x c-1 Local channel matrix H c Sequentially calculating the current process matrix +.>Channel inverse matrix->And estimate vector x c . The specific calculation steps are as follows:
step 4.2.1: channel inverse matrix based on last distributed unitAnd a local channel matrix H c Process matrix->Updating:
wherein I is 2 Is an identity matrix of 2 x 2 dimensions,is a diagonal matrix of dimensions 4 x 4. Calculate->Only the diagonal elements need to be calculated and reserved, and the non-diagonal elements are set to zero.
Step 4.2.2: channel inverse matrix based on last distributed unitLocal channel matrix H c And the current process matrix->Inverse matrix of current channel->Updating:
wherein,is a diagonal matrix of dimension K x K, < >>Is a diagonal matrix of dimension 2 x 2. Calculate->Only the diagonal elements need to be calculated and reserved, and the non-diagonal elements are set to zero.
Step 4.2.3: based on the inverse matrix of the current channelLocal channel matrix H c And the estimated vector x of the last distributed unit c-1 Current estimation vector x c Updating:
wherein the calculated vector x c The dimension is Kx1.
Step 4.3 current estimate vector x c And the estimated vector x of the last distributed unit c-1 Linear weighted sum is performed, and a damping factor is set to be beta=0.1, and the specific operation is as follows:
x c ＝βx c-1 +(1-β)x c
Step 4.4 compares the magnitudes of C and c=16, and if C <16, jumps to step 4.5. If c=16, step 4.6 is skipped.
Step 4.5 inverse matrix of current channelAnd current estimationMetering vector x c Pass on to the next distributed unit, recursion sequence number c=c+1, jump step 4.2 continues the iteration. />
Step 4.6 the current estimate vector x C And (5) transmitting the data to a central processing unit, and executing step 5.
Still taking a distributed massive MIMO system with transmit antenna k=4, receive antenna n=32, distributed unit number c=16 and each configured with b= 2 receive antennas as an example, the specific operation procedure of step 4 is explained:
step 1, initializing an iterative recursion process according to step 4.1, and setting the starting point of the general recursion least square algorithm as c=c 0 +1=5 distributed units, and the endpoint is set as the c=16 distributed unit, and signal detection tasks are sequentially performed.
Step 2, calculating a matrix in the c=5 th distributed unit according to step 4.2.1
First, a matrix is calculatedAnd->Is denoted as matrix a. In the calculation of A, due to +.>Is a diagonal matrix, so the computation is simplified:
wherein h is i,j And p i,j Respectively represent matrix H 5 Andthe ith row and jth column element of (c). The dimension of matrix a is 4 x 2.
Then Calculating a local channel matrix H 5 The product of the matrix A and the matrix B is denoted as matrix B. In the process of calculating B, only diagonal elements are needed to be calculated and reserved, and non-diagonal elements are set to be zero, so that the calculation is simplified:
wherein a is i,j And b i,j Representing the j-th column element of the i-th row of matrices a and B, respectively. The dimension of matrix B is 2 x 2.
Finally, the inverse matrix of the diagonal matrix B is calculatedIn the calculation process, only the reciprocal of the diagonal element needs to be solved, and matrix inversion operation is greatly simplified:
calculating to obtain diagonal matrixIts dimension is 2×2.
Step 3, calculating a matrix in the c=5 th distributed unit according to step 4.2.2
First, calculateAnd->The product of (a), matrix A and +.>Is denoted as matrix D. In the calculation of D, due to +.>Is a diagonal matrix, so the computation is simplified:
wherein a is i,j And r i,j Representing matrix A and matrix B, respectivelyThe ith row and jth column element of (c). The dimension of matrix D is 2 x 4.
Then, calculate D andthe product of matrix D and matrix AH is denoted as matrix E. In the process of calculating E, only diagonal elements are needed to be calculated and reserved, and non-diagonal elements are set to be zero, so that the calculation is simplified: />
Wherein d is i,j And e i,j The j-th column elements of the i-th row of matrices D and E, respectively. The dimension of the matrix E is 4 x 4.
Finally, calculate the matrixDifference from E due to->And E are both diagonal matrices, so only the difference of diagonal elements need be calculated:
calculating to obtain diagonal matrixIts dimension is 4×4.
Step 4, updating the estimated vector x according to step 4.2.3 5 。
First, a matrix is calculatedAnd->Is denoted as matrix F. In the calculation of F, due to +.>Is a diagonal matrix, so the calculation is simplified
Wherein h is i,j And q i,j Respectively represent matrix H 5 Andthe ith row and jth column element of (c). The dimension of the matrix F is 4 x 2.
Finally, based on the inverse matrix of the current channelLocal channel matrix H 5 And the last estimated vector x 4 Current estimation vector x 5 Updating:
x 5 ＝x 4 +F(y 5 -H 5 x 4 )
wherein the calculated vector x 5 The dimension is 4×1. In this process, only matrix and vector multiplication exists, and the required computational complexity is low.
Step 5, according to step 4.3, the current estimated vector x 5 And the estimation result x of the last distributed unit 4 Performing linear weighted sum and setting damping factorFor β=0.1, the specific operation is:
x 5 ＝βx 4 +(1-β)x 5
step 6, comparing the sizes of C and c=16 according to step 4.4, finding C <16, and jumping to step 4.5.
Step 7, according to step 4.5, matrix is formedDiagonal element and vector x of (2) 5 To the next distributed unit. The recursive sequence number c=5+1=6 and the jump to step 4.2 continues the iteration.
Step 8, the iterative process is repeatedly performed according to steps 4.2 to 4.5.
Step 9, comparing the sizes of C and c=16 according to step 4.4, finding that c=16, and jumping to step 4.6.
Step 10, according to step 4.6, the current estimated vector x of dimension 4 x 1 C And (5) transmitting the data to a central processing unit, and executing step 5.
Step 5: the CPU pair estimates the vector x C And carrying out constellation point quantization to obtain a final quantization detection result. The specific quantization operation can be expressed as:
wherein,rounded to nearest x C Is a constellation point of (a). I.e. compare x C Each element x of (2) k And decoding candidate Point set->Euclidean distance between constellation points of (a) and x k The constellation point with the minimum Euclidean distance is x k Is a result of the quantization of (2).
The present scheme may be replaced by a variety of distributed detection algorithms such as gaussian message passing, newton's iterations, recursive least squares, etc. if only from the completion of the multi-antenna distributed signal detection task. However, these algorithms either have very high computational complexity or data bandwidth cost, or limit the number of distributed unit antennas, which cannot be flexibly implemented. The scheme provided by the invention has the best comprehensive performance by comprehensively considering the comparison of detection performance, calculation complexity and transmission bandwidth.
For ease of understanding herein, a comparison of fig. 5-6 is provided, wherein GRLS is the proposed method of universal recursive least squares based distributed MIMO detection, where N is the number of antennas on the side of the massive MIMO base station, K is the total number of antennas for the user, C is the number of distributed units, B is the number of antennas therein, I and T are the number of inner and outer loops respectively,is the constellation size. The actual complexity and bandwidth size in the case of n=256, k=16 and 32 are also compared, wherein the GRLS algorithm sets c=64, b=4, the remaining algorithm sets c=8, b=32, i=t=3, < >>For comparison, the results of the test performance comparison are shown in FIGS. 5-6. As shown in table 1, in the system of n=256, k=16, C is set up 0 =7 (denoted as C in fig. 5-6) min ) The overall complexity of GRLS is 51972. Compared with 213248 complex multiplications required by the traditional RLS algorithm, the GRLS brings about quite significant reduction of computational complexity. Likewise, the transmission bandwidth required for GRLS is 58. Compared with 272 complex numbers required to be transferred by the traditional RLS algorithm, the GRLS greatly reduces the required data transmission bandwidth. In a system with n=256 and k=32, GRLS still has excellent performance with low computational complexity and transmission bandwidth. In addition, by increasing C 0 Further improving GRLS performance. C is also shown in FIGS. 5-6 0 Taking 2 XC min And 3 XC min The complexity and transmission bandwidth of GRLS.
TABLE 1
Complexity and bandwidth comparison for distributed algorithms
Example 2
The embodiment of the invention provides a distributed MIMO signal detection device based on general recursive least square, as shown in fig. 2, comprising: a central processing unit and a number of distributed units in communication with the central processing unit;
in the first distributed unit, the correlation parameters of the generic recursive least squares algorithm are initialized, including the estimated vector, the channel inverse matrix and the diagonalization parameters C 0 ；
For front C 0 The distributed units are sequentially based on a general recursive least square algorithm without antenna constraint, and continuously update the estimated vector and the channel inverse matrix by using the last distributed unit or the initialized channel inverse matrix and the estimated vector, and the local channel matrix and the received signal;
for C 0 -C distributed units, each of which in turn simplifies the generic recursive least squares algorithm based on the channel hardening characteristics of the MIMO system, continuously updating the estimated vector with the channel inverse matrix and the estimated vector of the last distributed unit, and the local channel matrix and the received signal;
The central processing unit is used for carrying out quantization processing on the finally obtained current estimated vector to obtain a corresponding constellation point.
In a specific implementation manner of the embodiment of the present invention, each distributed unit includes a radio frequency processing module, a channel estimation module and a signal detection module;
the radio frequency processing module processes the received uplink signals to obtain pilot signals and local receiving signals;
the channel estimation module estimates a local channel matrix by utilizing pilot signals and sends the local channel matrix to the signal detection module;
for front C 0 The signal detection module is based on a general recursive least square algorithm without antenna constraint, and continuously updates the estimated vector and the channel inverse matrix by using the last distributed unit or the initialized channel inverse matrix and the estimated vector, and the local channel matrix and the received signal;
for post C-C 0 A distributed unit, the signal detection module simplifies the general recursive least square algorithm based on the channel hardening characteristic of the MIMO system, continuously uses the channel inverse matrix and the estimated vector of the last distributed unit, and updates the estimated vector by the local channel matrix and the received signal, and finally calculates the estimated vector x obtained by the C distributed unit C To the central processing unit.
The remainder was the same as in example 1.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The embodiments of the present invention have been described above with reference to the accompanying drawings, but the present invention is not limited to the above-described embodiments, which are merely illustrative and not restrictive, and many forms may be made by those having ordinary skill in the art without departing from the spirit of the present invention and the scope of the claims, which are all within the protection of the present invention.
The foregoing has shown and described the basic principles and main features of the present invention and the advantages of the present invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, and that the above embodiments and descriptions are merely illustrative of the principles of the present invention, and various changes and modifications may be made without departing from the spirit and scope of the invention, which is defined in the appended claims. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (10)
1. A distributed MIMO signal detection method based on a common recursive least squares, comprising:
initializing correlation parameters of a generic recursive least squares algorithm, said correlation parameters comprising an estimation vector, a channel inverse matrix and a diagonalization parameter C 0 ；
For front C 0 The distributed units are sequentially utilized to update the estimated vector and the channel inverse matrix by utilizing the last distributed unit or the initialized channel inverse matrix and the estimated vector, and the local channel matrix and the received signal based on a general recursion least square algorithm without antenna constraint;
for post C-C 0 The distributed units are sequentially utilized to simplify a general recursive least square algorithm based on the channel hardening characteristic of the MIMO system, the channel inverse matrix and the estimated vector of the last distributed unit are continuously utilized, the local channel matrix and the received signal are utilized to update the estimated vector, and C is the total number of the distributed units;
And carrying out quantization processing on the finally obtained current estimated vector by using a central processing unit to obtain a corresponding constellation point.
2. The method for detecting distributed MIMO signals based on the general recursive least square according to claim 1, wherein the local channel matrix is obtained by channel estimation using pilot signals, the dimension of the local channel matrix is b×k, B is the number of receiving antennas that a single distributed unit has, b=n/C, C is the total number of distributed units, N is the total number of receiving antennas, and K is the total number of transmitting antennas.
3. The method for detecting distributed MIMO signals based on the universal recursive least squares according to claim 2, wherein the initializing the correlation parameters of the universal recursive least squares algorithm comprises:
setting an initial value of an estimation vector to x 0 =0, dimension k×1;
setting the initial value of the channel inverse matrix to P 0 ＝I K ，I K A unit matrix with K multiplied by K dimension;
setting the initial value of the diagonalization parameter as:
wherein,representing rounding to the nearest integer.
4. A method for detecting a distributed MIMO signal based on the common recursive least squares method as claimed in claim 3, wherein the common recursive least squares algorithm based on the non-antenna constraint continuously updates the estimated vector and the channel inverse matrix with the last distributed element or initialized channel inverse matrix and the local channel matrix and the received signal, comprising:
Initializing the general recursive least square algorithm with the c=1st distributed unit as the start point and the C as the end point 0 The distributed units sequentially perform channel detection tasks;
repeating the preset loop step until c=c 0 The preset circulation steps comprise: in the c-th distributed unit, the channel inverse matrix P of the last distributed unit is utilized c-1 And estimate vector x c-1 Local channel matrix H c Sequentially calculating a current process matrix R c Channel inverse matrix P c And estimate vector x c The method comprises the steps of carrying out a first treatment on the surface of the Comparison of C and C 0 If c is the size of<C 0 Then inverse matrix P of current channel c And the current estimate vector x c Pass on to the next distributed unit, recursive sequence number c=c+1;
diagonalizing the current channel inverse matrix P c ObtainingWherein diag (diag (P) c ) For diagonalization operations, a matrix P is reserved c Is defined by the diagonal elements of the (c) in the drawing,while the non-diagonal elements are set to zero, the diagonal matrix +.>Is K x K; inverse matrix of current channel->And the current estimate vector x c Pass on to the next distributed unit, recursive sequence number c=c+1.
5. The method for detecting distributed MIMO signals based on the universal recursive least squares according to claim 4, wherein said at the c-th distributed unit uses the inverse channel matrix P of the last distributed unit c-1 And estimate vector x c-1 Local channel matrix H c-1 Sequentially calculating a current process matrix R c Channel inverse matrix P c And estimate vector x c The method specifically comprises the following steps:
channel inverse matrix P based on last distributed unit c-1 And a local channel matrix H c For process matrix R c Updating:
wherein I is B Is a B x B dimension identity matrix, matrix R c Is B x B;
channel inverse matrix P based on last distributed unit c-1 Local channel matrix H c And a current process matrix R c Inverse matrix P of current channel c Updating:
wherein matrix P c The dimension is K multiplied by K;
based on the current channel inverse matrix P c Local channel matrix H c And the estimated vector x of the last distributed unit c-1 Current estimation vector x c Updating:
wherein the vector x is estimated c Is K x 1.
6. A distributed MIMO signal detection method based on the general recursive least squares according to claim 3, wherein: the channel hardening characteristic based on the MIMO system simplifies the general recursive least square algorithm, continuously uses the channel inverse matrix and the estimated vector of the last distributed unit, and updates the estimated vector by the local channel matrix and the received signal, and specifically comprises the following steps:
Initializing the start of the generic recursive least squares algorithm to be c=c 0 +1 distributed units, wherein the end point is the C-th distributed unit, and channel detection tasks are sequentially carried out;
repeating a preset circulation step until c=c, wherein the preset circulation step comprises using the channel inverse matrix of the last distributed unit in the C-th distributed unitAnd estimate vector x c-1 And a local channel matrix H c Sequentially calculating the current process matrix +.>Channel inverse matrix->And estimate vector x c The method comprises the steps of carrying out a first treatment on the surface of the For the current estimated vector x c And the estimated vector x of the last distributed unit c-1 And (3) performing linear weighted sum, setting a damping factor as beta, and specifically, the method comprises the following steps of: x is x c ＝βx c-1 +(1-β)x c The method comprises the steps of carrying out a first treatment on the surface of the Comparing the sizes of C and C, if C<C, inverse matrix of current channel +.>And the current estimate vector x c Pass on to the next distributed unit, recursive sequence number c=c+1;
the current estimation vector x C To the central processing unit.
7. The method for detecting distributed MIMO signals based on the universal recursive least squares according to claim 6, wherein said at the c-th distributed unit uses the inverse channel matrix of the last distributed unitAnd estimate vector x c-1 Local channel matrix H c Sequentially calculating the current process matrix +. >Channel inverse matrix->And estimate vector x c The method specifically comprises the following steps:
channel inverse matrix based on last distributed unitAnd a local channel matrix H c Process matrix->Updating:
wherein I is B Is of dimension B X BThe identity matrix is used as a matrix of units,is a diagonal matrix of dimension K x K, calculating +.>In the process of (1), only diagonal elements are calculated and reserved, and non-diagonal elements are set to be zero;
channel inverse matrix based on last distributed unitLocal channel matrix H c And the current process matrix->Inverse matrix of current channel->Updating;
wherein,is a diagonal matrix of dimension K x K, < >>Is a diagonal matrix of dimension B x B, calculating +.>In the process of (1), only diagonal elements are calculated and reserved, and non-diagonal elements are set to be zero;
based on the inverse matrix of the current channelLocal channel matrix H c And the last estimated vector x c-1 Current estimation vector x c Updating:
wherein the vector x is estimated c Is K x 1 due to the dimensions ofIs a diagonal matrix, and the corresponding matrix multiplication only requires calculation and retention of diagonal elements.
8. The method for detecting distributed MIMO signals based on the general recursive least square according to claim 1, wherein the quantization processing is performed on the finally obtained current estimated vector by using a central processing unit, and the calculation formula is:
Wherein,output signal representing central processing unit, +.>Representing rounding to nearest the current estimate vector x finally obtained C Is a constellation point of (a).
9. A distributed MIMO signal detection system based on a common recursive least squares, comprising: a central processing unit and a number of distributed units in communication with the central processing unit;
in the first distributed unit, a generic recursive least squares is initializedRelated parameters of the algorithm including the estimated vector, the channel inverse matrix and the diagonalization parameter C 0 ；
For front C 0 The distributed units are sequentially based on a general recursive least square algorithm without antenna constraint, and continuously update the estimated vector and the channel inverse matrix by using the last distributed unit or the initialized channel inverse matrix and the estimated vector, and the local channel matrix and the received signal;
for post C-C 0 The distributed units simplify the general recursive least square algorithm based on the channel hardening characteristic of the MIMO system in sequence, and continuously update the estimated vector by using the channel inverse matrix and the estimated vector of the last distributed unit and the local channel matrix and the received signal;
The central processing unit is used for carrying out quantization processing on the finally obtained current estimated vector to obtain a corresponding constellation point.
10. The system of claim 9, wherein each of the distributed units comprises a radio frequency processing module, a channel estimation module, and a signal detection module;
the radio frequency processing module processes the received uplink signals to obtain pilot signals and local receiving signals;
the channel estimation module estimates a local channel matrix by utilizing pilot signals and sends the local channel matrix to the signal detection module;
for front C 0 The signal detection module is based on a general recursive least square algorithm without antenna constraint, and continuously updates the estimated vector and the channel inverse matrix by using the last distributed unit or the initialized channel inverse matrix and the estimated vector, and the local channel matrix and the received signal;
for post C-C 0 A distributed unit, wherein the signal detection module simplifies the general recursive least square algorithm based on the channel hardening characteristics of the MIMO system, and continuously uses the last distributed unitChannel inverse matrix and estimated vector, local channel matrix and received signal update estimated vector, and estimated vector x obtained by calculating it by C-th distributed unit C To the central processing unit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410006694.3A CN117792546A (en) | 2024-01-03 | 2024-01-03 | Distributed MIMO signal detection method and system based on general recursive least square |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410006694.3A CN117792546A (en) | 2024-01-03 | 2024-01-03 | Distributed MIMO signal detection method and system based on general recursive least square |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117792546A true CN117792546A (en) | 2024-03-29 |
Family
ID=90383402
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410006694.3A Pending CN117792546A (en) | 2024-01-03 | 2024-01-03 | Distributed MIMO signal detection method and system based on general recursive least square |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117792546A (en) |
-
2024
- 2024-01-03 CN CN202410006694.3A patent/CN117792546A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108650056B (en) | Mixed iteration detection method in uplink of large-scale MIMO system | |
| CN101499840B (en) | Iteration detection method for MIMO system | |
| CN110417515B (en) | Large-scale MIMO signal detection method based on discrete iterative estimation | |
| CN113541747B (en) | Large-scale MIMO detection method, device and storage medium | |
| CN113114313A (en) | Method, system and storage medium for detecting pilot auxiliary signal of MIMO-NOMA system | |
| CN111162828B (en) | Low-complexity signal detection method of large-scale MIMO system | |
| CN114204971B (en) | Iterative aggregate beam forming design and user equipment selection method | |
| CN108809383A (en) | A kind of associated detecting method for massive MIMO up-link signals | |
| CN113630160B (en) | Large-scale MIMO detection method, device, equipment and storage medium | |
| CN117792546A (en) | Distributed MIMO signal detection method and system based on general recursive least square | |
| CN111555785B (en) | Signal precoding method and device and electronic equipment | |
| CN115967421A (en) | Method and system for joint parameter estimation and signal reconstruction under distributed antenna system | |
| CN110190908B (en) | Constellation diagram design method for minimizing bit error rate of incoherent large-scale SIMO system in ISI channel | |
| CN107888537B (en) | Signal detection method for improving system complexity in large-scale antenna system | |
| CN110784274B (en) | Method for determining receiver algorithm for large-scale multiple-input multiple-output | |
| Seidel et al. | A novel approach for combining local estimates for fully decentralized feedforward massive MIMO equalization: The multistep fusion | |
| CN111769975A (en) | MIMO system signal detection method and system | |
| CN118054819B (en) | Distributed large-scale MIMO low-complexity precoding method | |
| CN115865155B (en) | MAB-based two-stage precoding method under multi-scattering cluster channel | |
| CN113328771B (en) | Large-scale MIMO signal detection method based on conjugate gradient algorithm | |
| CN117793758B (en) | Multi-variable joint optimization method in RIS-assisted NOMA system | |
| WO2023169799A1 (en) | Precoding for beamforming | |
| CN114124640A (en) | Signal compensation processing method and device | |
| CN117118538A (en) | High-performance low-complexity wireless antenna system signal detection method and device | |
| CN117118486A (en) | Associated channel-oriented mMIMO detection preprocessing and parameter pre-training 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 |