CN114900214A - Low-complexity linear precoding algorithm based on OFDM system - Google Patents
Low-complexity linear precoding algorithm based on OFDM system Download PDFInfo
- Publication number
- CN114900214A CN114900214A CN202210488206.8A CN202210488206A CN114900214A CN 114900214 A CN114900214 A CN 114900214A CN 202210488206 A CN202210488206 A CN 202210488206A CN 114900214 A CN114900214 A CN 114900214A
- Authority
- CN
- China
- Prior art keywords
- channel
- matrix
- ofdm system
- layer
- precoding
- 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.)
- Granted
Links
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 57
- 238000000034 method Methods 0.000 claims abstract description 62
- 239000011159 matrix material Substances 0.000 claims abstract description 45
- 239000000969 carrier Substances 0.000 claims abstract description 9
- 239000013598 vector Substances 0.000 claims description 29
- 230000005540 biological transmission Effects 0.000 claims description 16
- 238000011176 pooling Methods 0.000 claims description 16
- 230000004913 activation Effects 0.000 claims description 13
- 238000001994 activation Methods 0.000 claims description 13
- 238000013528 artificial neural network Methods 0.000 claims description 12
- 238000013527 convolutional neural network Methods 0.000 claims description 8
- 230000008878 coupling Effects 0.000 claims description 8
- 238000010168 coupling process Methods 0.000 claims description 8
- 238000005859 coupling reaction Methods 0.000 claims description 8
- 230000004044 response Effects 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 238000013459 approach Methods 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 14
- 238000005562 fading Methods 0.000 abstract description 7
- 230000008569 process Effects 0.000 abstract description 6
- 238000010801 machine learning Methods 0.000 description 21
- 238000010586 diagram Methods 0.000 description 17
- 230000000875 corresponding effect Effects 0.000 description 8
- 238000010295 mobile communication Methods 0.000 description 8
- 238000004891 communication Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 238000003780 insertion Methods 0.000 description 3
- 230000037431 insertion Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000003213 activating effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000002939 conjugate gradient method Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000010363 phase shift Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000003313 weakening effect Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B7/00—Radio transmission systems, i.e. using radiation field
- H04B7/02—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas
- H04B7/04—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas
- H04B7/0413—MIMO systems
- H04B7/0456—Selection of precoding matrices or codebooks, e.g. using matrices antenna weighting
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/0202—Channel estimation
- H04L25/024—Channel estimation channel estimation algorithms
- H04L25/0254—Channel estimation channel estimation algorithms using neural network algorithms
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L27/00—Modulated-carrier systems
- H04L27/26—Systems using multi-frequency codes
- H04L27/2601—Multicarrier modulation systems
- H04L27/2647—Arrangements specific to the receiver only
- H04L27/2655—Synchronisation arrangements
- H04L27/2689—Link with other circuits, i.e. special connections between synchronisation arrangements and other circuits for achieving synchronisation
- H04L27/2695—Link with other circuits, i.e. special connections between synchronisation arrangements and other circuits for achieving synchronisation with channel estimation, e.g. determination of delay spread, derivative or peak tracking
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Power Engineering (AREA)
- Radio Transmission System (AREA)
Abstract
The invention discloses a linear precoding method based on an OFDM system, which is characterized in that the OFDM system is adopted to carry out orthogonal frequency division multiplexing on a channel, the channel is converted into a plurality of narrow-band sub-channels, parameters such as Lagrange operators and the like of each sub-channel in the precoding matrix solving process are considered to be slowly changed in the time and frequency directions, only sub-carriers in partial time and frequency directions are selected to insert pilot frequency, and necessary parameters are respectively obtained. And estimating channel parameters of other frequencies by an interpolation algorithm by using the acquired parameters of the known frequency points, thereby directly calculating all precoding matrixes. The linear precoding algorithm adopts an interpolation method to estimate the channel parameters of the unknown subcarriers, thereby reducing the complexity of the algorithm and improving the operation efficiency and the practicability. The OFDM technology adopted by the linear precoding algorithm has great advantages in resisting frequency selective fading or narrow-band interference.
Description
Technical Field
The invention belongs to the technical field of large-scale MIMO communication, and particularly relates to a low-complexity linear precoding algorithm based on an OFDM system.
Background
With the development of mobile communication technology, mobile intelligent terminals have gradually become an indispensable part of people's daily production life. In addition to human-centric communications, machine-centric communications are also becoming increasingly important. These newly accessed devices will place a great deal of new service requirements on the mobile communication system, such as communication cost, complexity, energy consumption, data rate, mobility, delay, reliability, etc. In order to meet the development requirements of the future human society and deal with the appearance of novel services and application scenes, the fifth generation mobile communication system in 2020 and later comes.
MIMO, which uses a plurality of transmitting antennas and receiving antennas at a transmitting end and a receiving end, respectively, transmits and receives signals through the plurality of antennas of the transmitting end and the receiving end, thereby improving communication quality. The multi-antenna multi-transmission multi-reception mobile communication system can fully utilize space resources, realizes multi-transmission and multi-reception through a plurality of antennas, can improve the system channel capacity by times under the condition of not increasing frequency spectrum resources and antenna transmitting power, shows obvious advantages, and is regarded as the core technology of next generation mobile communication.
Because the improvement effect on the capacity of the mobile communication system is remarkable, a large-scale multiple-input multiple-output technology, namely, a large-scale MIMO technology, in which a large-scale antenna array is configured on the base station side is one of key technologies of a new generation of mobile communication network, and is also a research hotspot in the field of mobile communication in recent years.
The research on the large-scale MIMO technology with moderate complexity not only has important theoretical significance, but also has important application value in meeting the business requirements of low communication cost, low complexity, low energy consumption, high data rate, low time delay, high reliability and the like proposed by 5G and the later 5G.
Since massive MIMO has a plurality of sub-channels and requires no mutual interference between the sub-channels, an Orthogonal Frequency Division Multiplexing (OFDM) system is generally used. The parallel transmission of high-speed serial data is realized through frequency division multiplexing, so that the large-scale MIMO has better multi-path weakening resistance, and multi-user access can be supported.
The main idea of OFDM is to divide the channel into several orthogonal sub-channels, convert the high-speed data signal into parallel low-speed sub-data streams, and modulate them to be transmitted on each sub-channel. The orthogonal signals can be separated by using correlation techniques at the receiving end, which can reduce mutual interference between the sub-channels. The signal bandwidth on each subchannel is smaller than the associated bandwidth of the channel, so that flat fading can be seen on each subchannel, thereby eliminating inter-symbol interference, and since the bandwidth of each subchannel is only a small fraction of the original channel bandwidth, channel equalization becomes relatively easy.
As a research focus of the current massive MIMO technology, the precoding technology can fully exploit spatial degrees of freedom and minimize intra-cell and inter-cell interference. If different phase shifts are used in different frequency bands of the system bandwidth to achieve small-scale fading resistance, the technique is defined as a precoding technique.
Typical precoding techniques mostly assume that perfect channel state information, i.e., CSI, can be obtained. However, in a mobile environment, due to processing delay caused by processes of pilot transmission, channel estimation, precoding matrix calculation, and the like, the estimated CSI has a large deviation from a channel in actual downlink transmission. Currently, the study of the transmission of massive MIMO under imperfect CSI conditions is still in the preliminary stage. In order to improve transmission performance in a mobile environment, improvements can be made from two dimensions: designing and improving a precoding method on the premise of non-perfection of CSI (channel state information), thereby improving the system performance; the capability of acquiring the channel information of the base station side is improved, and the accuracy of the CSI is improved by using a channel prediction method.
Whether nonlinear precoding or linear precoding, obtaining accurate channel state information is a precondition for realizing high-performance precoding. Aiming at the phenomenon that large-scale MIMO channel information is not perfect in a mobile scene, the channel information of a plurality of known detection periods can be utilized to predict future channel information, so that the pre-coding performance is enhanced. In a static channel environment, more accurate channel state information can be obtained by using a channel sounding reference signal and a common channel estimation method, and channel prediction brings certain gain to system performance no matter for a narrow-band system or a broadband system. The actual wireless channel is continuously changed, and especially in a mobile scene, the speed of the channel change is positively correlated with the relative speed of the mobile device and the base station. Therefore, it is necessary to design a low-complexity linear precoding algorithm suitable for the imperfect channel state information condition.
Disclosure of Invention
The invention aims to provide a low-complexity linear precoding algorithm based on an OFDM system so as to solve the technical problems in the background technology.
In order to solve the technical problems, the specific technical scheme of the invention is as follows:
a low-complexity linear precoding algorithm based on an OFDM system comprises the following steps:
step 1, inserting pilot frequency into an OFDM system from time and frequency dimensions respectively to obtain a channel matrix of the pilot frequency signal;
step 2, utilizing a convolution neural network to calculate Lagrange operators and user power vector parameters of the pilot channel;
step 3, estimating Lagrange operator parameters of the rest subcarriers by using an interpolation method through Lagrange operators of the known pilot channel and user power vector parameters;
and 4, calculating a precoding matrix by using Lagrange operator parameters of all subcarriers.
Further, in step 1, known pilot symbols are inserted into the transmitting end, and the receiving end estimates the channel characteristics by using the pilot symbols; consider a single-cell multi-user massive MIMO downlink transmission link, where the number of base station side antennas is M t The number of users is K, and each user is provided with M k A root antenna; setting downlink data sent to the kth user asPrecoding matrix for user k isThe corresponding frequency domain channel matrix isReceived noise of n k Noise power of σ 2 (ii) a The received signal y of user k k Is shown as
If the transmitted signals of K users are jointly expressed asThe received signals of all users are represented asSystem transfer model is rewritten as
y=HPx+n
Extracting received pilot signal y from y (k) P (k) Estimating the channel frequency response { H } of the sub-channel with the inserted pilot frequency P (k)}。
Further, in step 2, in order to obtain instantaneous correlation coefficients required for calculating lagrangian operators in different scenes, the low-complexity linear precoding algorithm based on the OFDM system according to claim 2 is characterized in that, in step 2, it can be found through simulation of subsequent experiments that the value of the optimal instantaneous correlation coefficient β is mainly related to the user moving speed and the uplink sounding period, and on the premise that different users are assumed to have the same optimal instantaneous correlation coefficient β, we select the user moving speed and the uplink sounding period as inputs, and find the optimal instantaneous correlation coefficient β through a table look-up method.
Further, the Lagrangian in the step 2 is obtained through a convolution neural network structure; and constructing a convolutional neural network with a precoding direction matrix, an instantaneous channel, a channel coupling array and a signal-to-noise ratio as input and an optimal Lagrange multiplier as output to obtain the optimal Lagrange multiplier. Compared with the traditional algorithm, the method has lower computational complexity.
Further, step 4, a conjugate gradient method is used, the minimum generalized eigenvalue is approached through an iteration method, and the optimal precoding direction matrix is an eigenvector corresponding to the minimum generalized eigenvalue.
Further, the convolutional neural network is divided into an input layer, a convolutional layer, a pooling layer, an activation layer, a full-link layer and an output layer; the input layer inputs a data structure to be detected; considering that the data structures of instantaneous channel information and a channel coupling matrix are different, firstly, the channel information needs to be normalized, and normalized data samples respectively enter a neural network from three input layers; then, carrying out repeated convolution, pooling and activation on the three paths of input information; the convolution layer performs convolution operation on each input sample; the pooling layer reduces the number of the features extracted from the convolutional layer, and redundant features are removed through pooling; the function of the activation layer is to introduce a nonlinear factor, and after each layer is pooled, the activation layer is activated by using a Relu (·) function; after completing convolution, pooling and activation for multiple times, three-way data m with the same data structure 1 、m 2 And m 3 Respectively inputting the full-connection layers, and mapping the extracted features into predicted values; meanwhile, the signal-to-noise ratio is also used as a node to participate in full connection, and the training result is adjusted; the result of the prediction finally output from the output layer is a K × 1 vector μ.
Obtaining Lagrange operator mu with instantaneous correlation coefficient of 1 by using instantaneous channel information 1 (ii) a Obtaining Lagrange operator mu with instantaneous correlation coefficient of 0 by utilizing statistical information 0 (ii) a Using a weighted model
Obtaining the corresponding Lagrange operator of the current channelWhere η represents the correlation factor.
Further, the user power vector in step 2 is obtained by using a weighted model method; firstly, solving a neural network similar to a Lagrange operator mu to solve a power vector when an instantaneous correlation coefficient is zeroCombining the obtained power vectorsUsing a weighted model
Obtaining a power vector corresponding to a current channelWhere ξ represents the correlation factor.
Further, step 3 is based on the channel being slowly varied, so the parameters μ,Is slowly changed, so that the analysis of the rest sub-carriers one by one is not needed, and only the parameters mu of the sub-channel with the inserted pilot frequency,Estimating the parameters mu, mu of the rest sub-carriers by using an interpolation method,And further obtain the channel responses H of the remaining subcarriers.
Further, after the lagrangian operator mu is obtained in the step 4, the direction information of the precoding matrix is obtained by using a method for solving the generalized eigenvalue, and the obtained user power vector is combinedThe parameters quickly get a complete precoding matrix.
The low-complexity linear precoding algorithm based on the OFDM system has the following advantages:
1. the linear precoding algorithm adopts an interpolation method to estimate the channel parameters of the unknown subcarriers, thereby reducing the complexity of the algorithm and improving the operation efficiency and the practicability.
2. The invention adopts the method of inserting the pilot frequency through the OFDM technology under the assumption of slow fading of the channel, and is an optimal solution.
3. The OFDM technology adopted by the linear precoding algorithm has great advantages in resisting frequency selective fading or narrow-band interference. In this multi-carrier system, only a small fraction of the carriers are interfered with, with high accuracy.
Drawings
FIG. 1 is a diagram of a massive MIMO system model according to the present invention;
FIG. 2(a) is a block pilot insertion diagram of the OFDM system of the present invention;
FIG. 2(b) is a schematic diagram of comb pilot insertion in the OFDM system of the present invention;
FIG. 2(c) is a schematic diagram of two-dimensional decimation pilot insertion in the OFDM system of the present invention;
FIG. 3 is a block diagram of the workflow of the present invention;
fig. 4(a) is a graph of the influence of the linear interpolation method on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 5 ms;
fig. 4(b) is a graph of the influence of the linear interpolation method on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 10 ms;
fig. 4(c) is a graph of the influence of the linear interpolation method of the present invention on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 20 ms;
fig. 4(d) is a graph of the influence of the linear interpolation method of the present invention on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 40 ms;
fig. 4(e) is a graph of the influence of the linear interpolation method on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 80 ms;
fig. 4(f) is a graph of the influence of the linear interpolation method of the present invention on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 160 ms;
fig. 5(a) is a diagram illustrating an influence of the nearest neighbor interpolation method of the present invention on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 5 ms;
fig. 5(b) is a diagram illustrating an influence of the nearest neighbor interpolation method of the present invention on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 10 ms;
fig. 5(c) is a graph of the influence of the nearest neighbor interpolation method of the present invention on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 20 ms;
fig. 5(d) is a diagram illustrating an influence of the nearest neighbor interpolation method of the present invention on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 40 ms;
fig. 5(e) is a diagram illustrating an influence of the nearest neighbor interpolation method of the present invention on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 80 ms;
fig. 5(f) is a diagram illustrating an influence of the nearest neighbor interpolation method of the present invention on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 160 ms;
fig. 6(a) is a diagram of the influence of the cubic spline interpolation method of the present invention on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 5 ms;
fig. 6(b) is a diagram of the influence of the cubic spline interpolation method of the present invention on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 10 ms;
fig. 6(c) is a diagram of the influence of the cubic spline interpolation method of the present invention on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 20 ms;
fig. 6(d) is a diagram of the influence of the cubic spline interpolation method of the present invention on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 40 ms;
fig. 6(e) is a diagram of the influence of the cubic spline interpolation method of the present invention on the performance of the system approaching the optimal machine learning assisted robust precoding algorithm when the channel sounding period T is 80 ms;
FIG. 6(f) is a diagram illustrating the influence of cubic spline interpolation on the performance of the optimal machine learning-assisted robust precoding algorithm system when the channel sounding period T is 160ms
Detailed Description
In order to better understand the purpose, structure and function of the present invention, the following describes a low complexity linear precoding algorithm based on an OFDM system in further detail with reference to the accompanying drawings.
The invention provides a low-complexity linear precoding algorithm, which adopts an OFDM system to carry out orthogonal frequency division multiplexing on a channel, converts the channel into a plurality of narrow-band sub-channels, considers that parameters such as Lagrangian operators and the like of each sub-channel in the precoding matrix solving process are slowly changed in the time and frequency directions, only selects sub-carriers in partial time and frequency directions to insert pilot frequency, and respectively obtains necessary parameters. And estimating channel parameters of other frequencies by an interpolation algorithm by using the acquired parameters of the known frequency points, thereby directly calculating all precoding matrixes. Fig. 1 is a diagram illustrating a massive MIMO system model according to the present invention.
As shown in fig. 3, the present invention comprises the steps of:
step 1, as shown in fig. 2, inserting pilots into the OFDM system from the time and frequency dimensions, respectively, and obtaining a channel matrix of the pilot signals.
Pilot frequency is inserted at a certain time and frequency interval in two dimensions of time domain and frequency domain, and a channel matrix of the pilot frequency signal can be estimated according to the received pilot frequency signal.
Consider a simple littleIn large-scale MIMO downlink transmission of multiple users, known pilot symbols are inserted into signals at a base station end, and a receiving end estimates channel characteristics by using the pilot symbols; wherein the number of base station end antennas is M t The number of users is K, and each user is provided with M k A root antenna; setting downlink data sent to the kth user asPrecoding matrix for user k isThe corresponding frequency domain channel matrix isReceived noise of n k Noise power of σ 2 (ii) a The received signal y of user k k Is shown as
If the transmitted signals of K users are jointly expressed asThe received signals of all users are represented asSystem transfer model is rewritten as
y=HPx+n
Wherein H is a channel matrix, P is a precoding matrix, x is a transmission signal, y is a reception signal, and n is noise; extracting a received pilot signal y from a received signal y (k) P (k) Estimating the channel frequency response { H } of the sub-channel with the inserted pilot frequency P (k)}。
And 2, acquiring an instantaneous correlation coefficient through the channel matrix, and outputting the instantaneous correlation coefficient as the input of the convolutional neural network and the Lagrangian operator of the pilot channel and the user power vector parameter.
In order to obtain the instantaneous correlation coefficients required by the Lagrange operator under different scenes, on the premise that different users have the same optimal instantaneous correlation coefficient beta, the user moving speed and the uplink detection period are selected as input, and the optimal beta is found through a table look-up method. The Lagrange operator is obtained through a convolution neural network structure; and constructing a convolutional neural network with a precoding direction matrix, an instantaneous channel, a channel coupling array and a signal-to-noise ratio as input and an optimal Lagrange multiplier as output to obtain the optimal Lagrange multiplier.
Solving the lagrangian operator requires solving two problems, which specifically include:
1. and constructing a neural network which takes a precoding direction matrix, an instantaneous correlation coefficient, an instantaneous channel coupling array and a signal-to-noise ratio as input and takes a Lagrangian as output, and directly obtaining the Lagrangian through channel information with lower calculation complexity.
2. The expression of the channel a posteriori model is degraded for instantaneous correlation coefficients of 1 and 0. Simplifying the process of solving the Lagrange operator into a weighting model, and solving mu by using an instantaneous channel 1 Using machine learning, inputting channel coupling matrix and signal-to-noise ratio, and solving mu by Lagrange multiplier when output beta is 0 0 And obtaining the Lagrange operator mu of the pilot channel after weighting processing. The weighting factor is related to the instantaneous correlation coefficient.
In order to reduce the complexity of calculating the user power, a weighting method is adopted to obtain the user power. A precoding matrix when the instantaneous correlation coefficient is 1 is obtained during solving, and a power vector can be further solved; a power vector with an instantaneous correlation coefficient of 0 can be obtained by a neural network. The user power can be obtained by weighting processing.
The neural network is divided into an input layer, a convolutional layer, a pooling layer, an activation layer, a full-link layer and an output layer. The input layer inputs the data structure to be detected. Considering that the data structures of the instantaneous channel information and the channel coupling matrix are different, the channel information needs to be normalized firstly, and the normalized data samples enter the neural network from three input layers respectively. Then, the three paths of input information are respectively subjected to repeated convolution and pooling for multiple times to obtain three paths of input informationAnd activating. The convolution layer performs convolution operation on each input sample, and in order to reduce the complexity of the network as much as possible, the number of convolution kernels is reduced as much as possible on the premise of not influencing the performance. The pooling layer can reduce the number of features extracted by the convolutional layer, and redundant features are removed repeatedly through pooling. The role of the activation layer is to introduce a non-linear factor, activated using the Relu (-) function after pooling of each layer. Three-way data m having the same data structure after completion of the plurality of convolutions, pooling and activations 1 、m 2 And m 3 And respectively inputting the full-connection layers, and mapping the extracted features into predicted values. Meanwhile, the signal-to-noise ratio is also used as a node to participate in full connection, and the training result is adjusted. The result of the prediction finally output from the output layer is a K × 1 vector μ.
Obtaining Lagrange operator mu with instantaneous correlation coefficient of 1 by using instantaneous channel information 1 (ii) a Obtaining Lagrange operator mu with instantaneous correlation coefficient of 0 by utilizing statistical information 0 . Using a weighted model
And acquiring a Lagrangian mu corresponding to the current channel. Where η represents the correlation factor.
The user power vector is obtained by a weighted model method. Firstly, solving a neural network similar to a Lagrange operator mu to solve a power vector when an instantaneous correlation coefficient is zeroCombining the obtained power vectorsUsing a weighted model
Obtaining a power vector corresponding to a current channelWhere ξ represents the correlation factor.
And 3, estimating Lagrange operator parameters of the rest subcarriers by using an interpolation method through Lagrange operators and user power vector parameters of the known pilot channel.
Based on the channel being slowly varying, the parameters μ,Is also slowly varied, so that the analysis of the rest sub-carriers one by one is not needed, and only the mu, mu and mu of the sub-channel inserted with the pilot frequency are needed,Estimating the mu, mu of the rest sub-carriers by using an interpolation method,And further obtain the channel responses H of the remaining subcarriers.
And 4, forming matrixes Am and Bk, m by Lagrange operator parameters of all subcarriers and instantaneous channel information, wherein the eigenvector corresponding to the minimum generalized eigenvalue of the matrix pair (Am; Bk, m) is the optimal precoding matrix.
After obtaining the Lagrange operator mu, obtaining the direction information of the precoding matrix by using a method for solving the generalized eigenvalue, and combining the obtained user power vectorThe parameters can be equalized to obtain a complete precoding matrix quickly.
In the scheme, the linear precoding algorithm adopts an algorithm structure method of estimating the channel by using an interpolation method, which is derived from the traditional precoding matrix algorithm, but the characteristic of an OFDM system is utilized, and the interpolation method is combined with the interpolation method, so that short-time large-scale data transmission can be realized.
The linear precoding algorithm adopts an interpolation method to estimate the channel parameters of the unknown subcarriers, thereby reducing the complexity of the algorithm and improving the operation efficiency and the practicability.
Under the assumption of slow fading of the channel, the method of inserting the pilot frequency by the OFDM technology is the optimal solution.
The OFDM technology adopted by the linear precoding algorithm has great advantages in resisting frequency selective fading or narrow-band interference. In this multi-carrier system, only a small fraction of the carriers are interfered with, with high accuracy.
In order to verify the performance of the optimized linear precoding algorithm, simulation is carried out on the optimized linear precoding algorithm. In the simulation, each time slot is set to contain a plurality of blocks, and each block contains one OFDM symbol. The first subframe of each sounding period is used for uplink sounding process, and other subframes are used for downlink transmission. The base station designs a precoding matrix by utilizing the channel information obtained in the uplink detection process, and the precoding matrix is used as the transmission process of all downlink blocks in the next time slot. And the rest subframes except the first subframe of each sounding period are used for downlink transmission. In the transmission process, the generated user data passes through a downlink channel after being subjected to user selection, coding modulation and precoding under the guidance of channel information. After channel estimation, the sum rate of the transmission can be counted.
The MATLAB software is used for calculating and drawing images, the influence of three interpolation methods, namely linear interpolation (figure 4), nearest neighbor interpolation (figure 5) and cubic spline interpolation (figure 6), on the performance of the system approaching to the optimal machine learning auxiliary robust pre-coding algorithm is compared, the difference between the performance of the system approaching to the optimal machine learning auxiliary robust pre-coding algorithm is found to be not large with the performance of the algorithm and the rate of all parameters, and the method effectively reduces the complexity on the premise of almost no performance loss.
It is to be understood that the present invention has been described with reference to certain embodiments, and that various changes in the features and embodiments, or equivalent substitutions may be made therein by those skilled in the art without departing from the spirit and scope of the invention. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the invention without departing from the essential scope thereof. Therefore, it is intended that the invention not be limited to the particular embodiment disclosed, but that the invention will include all embodiments falling within the scope of the appended claims.
Claims (9)
1. A low-complexity linear precoding algorithm based on an OFDM system is characterized by comprising the following steps:
step 1, inserting pilot frequency into an OFDM system from time and frequency dimensions respectively to obtain a channel matrix of the pilot frequency signal;
step 2, obtaining instantaneous correlation coefficient through a channel matrix, using the instantaneous correlation coefficient as the input of a convolutional neural network, and using a Lagrange operator of a pilot channel and a user power vector parameter as the output;
step 3, estimating Lagrange operator parameters of the rest subcarriers by using an interpolation method through Lagrange operators of the known pilot channel and user power vector parameters;
step 4, Lagrange operator parameters of all subcarriers and instantaneous channel information form a matrix A m And B k,m Matrix pair (A) m ;B k,m ) The eigenvector corresponding to the minimum generalized eigenvalue is the optimal precoding matrix.
2. The OFDM system-based low-complexity linear precoding algorithm as claimed in claim 1, wherein in step 1, considering massive MIMO downlink transmission of single cell and multiple users, known pilot symbols are inserted into the signal at the base station, and the receiving end estimates the channel characteristics by using the pilot symbols; wherein the number of base station end antennas is M t The number of users is K, and each user is provided with M k A root antenna; setting downlink data sent to the kth user asPrecoding matrix for user k isThe corresponding frequency domain channel matrix isReceived noise of n k Noise power of σ 2 (ii) a The received signal y of user k k Is shown as
If the transmitted signals of K users are jointly expressed asThe received signals of all users are represented asSystem transfer model is rewritten as
y=HPx+n
Wherein H is a channel matrix, P is a precoding matrix, x is a transmission signal, y is a reception signal, and n is noise; extracting the received pilot signal y from the received signal y (k) P (k) Estimating the channel frequency response { H } of the sub-channel with the inserted pilot frequency P (k)}。
3. The low complexity linear precoding algorithm based on the OFDM system as claimed in claim 1, wherein in the step 2, in order to obtain instantaneous correlation coefficients required for computing lagrangian operators in different scenarios, on the premise that different users have the same optimal instantaneous correlation coefficient β, the user movement speed and the uplink sounding period are selected as inputs, and the optimal β is found through a table lookup method.
4. The OFDM system based low complexity linear precoding algorithm as claimed in claim 1, wherein the lagrangian in step 2 is obtained by a convolutional neural network structure; and constructing a convolutional neural network with a precoding direction matrix, an instantaneous channel, a channel coupling array and a signal-to-noise ratio as input and an optimal Lagrange multiplier as output to obtain the optimal Lagrange multiplier.
5. The method of claim 1The low-complexity linear precoding algorithm based on the OFDM system is characterized in that the convolutional neural network is divided into an input layer, a convolutional layer, a pooling layer, an activation layer, a full-link layer and an output layer; the input layer inputs a data structure to be detected; considering that the data structures of instantaneous channel information and a channel coupling matrix are different, firstly, the channel information needs to be normalized, and normalized data samples respectively enter a neural network from three input layers; then, carrying out repeated convolution, pooling and activation on the three paths of input information; the convolution layer performs convolution operation on each input sample; the pooling layer reduces the number of the features extracted from the convolutional layer, and redundant features are removed through pooling; the function of the activation layer is to introduce a nonlinear factor, and after each layer is pooled, the activation layer is activated by using a Relu (·) function; three-way data m having the same data structure after completion of the plurality of convolutions, pooling and activations 1 、m 2 And m 3 Respectively inputting the full-connection layers, and mapping the extracted features into predicted values; meanwhile, the signal-to-noise ratio is also used as a node to participate in full connection, and the training result is adjusted; the result of the prediction output from the output layer is finally a K × 1 vector μ;
obtaining Lagrange operator mu with instantaneous correlation coefficient of 1 by using instantaneous channel information 1 (ii) a Obtaining Lagrange operator mu with instantaneous correlation coefficient of 0 by utilizing statistical information 0 (ii) a Using a weighted model
6. The low complexity linear precoding algorithm based on the OFDM system as claimed in claim 1, wherein the step 4 is to approach the minimum generalized eigenvalue by an iterative method after obtaining the lagrangian, and the optimal precoding direction matrix is an eigenvector corresponding to the minimum generalized eigenvalue.
7. The OFDM system-based low-complexity linear precoding algorithm of claim 1, wherein the user power vector in step 2 is obtained by a weighted model method; firstly, solving a neural network similar to a Lagrange operator mu to solve a power vector when an instantaneous correlation coefficient is zeroCombining the obtained power vectorsUsing a weighted model
8. The OFDM system based low complexity linear precoding algorithm of claim 1, wherein the channel-based in step 3 is slowly-varying, so that the parameter Lagrangian μ, the power vector corresponding to the current channelThe method is also slow-varying, so that the other subcarriers do not need to be analyzed one by one, and only the Lagrange operator mu of the subchannel with the inserted pilot frequency and the power vector corresponding to the current channel are neededEstimating the Lagrangian mu of the rest sub-carriers by interpolationPower vector corresponding to current channelAnd then obtain the channel responses H of the remaining subcarriers.
9. The OFDM system-based low-complexity linear precoding algorithm as claimed in claim 1, wherein after the Lagrangian μ is obtained in step 4, a method for solving a generalized eigenvalue is used to obtain the direction information of the precoding matrix, and the obtained power vector corresponding to the current channel is combinedThe parameters quickly get a complete precoding matrix.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210488206.8A CN114900214B (en) | 2022-05-06 | 2022-05-06 | Low-complexity linear precoding algorithm based on OFDM system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210488206.8A CN114900214B (en) | 2022-05-06 | 2022-05-06 | Low-complexity linear precoding algorithm based on OFDM system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114900214A true CN114900214A (en) | 2022-08-12 |
CN114900214B CN114900214B (en) | 2024-01-26 |
Family
ID=82719802
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210488206.8A Active CN114900214B (en) | 2022-05-06 | 2022-05-06 | Low-complexity linear precoding algorithm based on OFDM system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114900214B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105471795A (en) * | 2015-11-18 | 2016-04-06 | 中国科学院上海微系统与信息技术研究所 | Pilot frequency design and channel estimation method for offset orthogonal multicarrier system |
CN113055061A (en) * | 2021-04-26 | 2021-06-29 | 中国科学技术大学 | Low-complexity precoding method of large-scale MIMO system |
CN114143896A (en) * | 2021-12-10 | 2022-03-04 | 东南大学 | Large-scale MIMO cross-frequency cooperation robust transmission method |
CN114337751A (en) * | 2021-12-07 | 2022-04-12 | 重庆邮电大学 | Power distribution method of time reversal OFDM multi-user communication system |
-
2022
- 2022-05-06 CN CN202210488206.8A patent/CN114900214B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105471795A (en) * | 2015-11-18 | 2016-04-06 | 中国科学院上海微系统与信息技术研究所 | Pilot frequency design and channel estimation method for offset orthogonal multicarrier system |
CN113055061A (en) * | 2021-04-26 | 2021-06-29 | 中国科学技术大学 | Low-complexity precoding method of large-scale MIMO system |
CN114337751A (en) * | 2021-12-07 | 2022-04-12 | 重庆邮电大学 | Power distribution method of time reversal OFDM multi-user communication system |
CN114143896A (en) * | 2021-12-10 | 2022-03-04 | 东南大学 | Large-scale MIMO cross-frequency cooperation robust transmission method |
Also Published As
Publication number | Publication date |
---|---|
CN114900214B (en) | 2024-01-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Shin et al. | Enhanced channel-estimation technique for MIMO-OFDM systems | |
CN112468196B (en) | Power distribution method in de-cellular large-scale MIMO system based on PZF precoding | |
CN107086886B (en) | Double-layer precoding design for large-scale MIMO system fusion zero forcing and Taylor series expansion | |
Su et al. | Investigation on key technologies in large-scale MIMO | |
CN108712198B (en) | Mixed precoding method based on sub-band equivalent channel matrix condition number | |
Jamali et al. | A low-complexity recursive approach toward code-domain NOMA for massive communications | |
CN113824477B (en) | Multi-user large-scale MIMO optimization method assisted by discrete lens antenna array | |
CN115733530A (en) | Combined precoding method for reconfigurable intelligent surface assisted millimeter wave communication | |
Tang et al. | Kalman filter channel estimation in 2× 2 and 4× 4 STBC MIMO-OFDM systems | |
Feng et al. | mmWave RIS-assisted SIMO channel estimation based on global attention residual network | |
Stephen et al. | Uplink channel estimation and data transmission in millimeter-wave CRAN with lens antenna arrays | |
Zhang et al. | Robust multi-branch space–time beamforming for OFDM system with interference | |
CN117220740A (en) | Beam forming method of communication and interference integrated system in non-cooperative scene | |
Pascual-Iserte et al. | An approach to optimum joint beamforming design in a MIMO-OFDM multiuser system | |
CN109039402B (en) | MIMO topological interference alignment method based on user compression | |
CN107733487B (en) | Signal detection method and device for large-scale multi-input multi-output system | |
Yang et al. | Optimal beamforming in cooperative cognitive backscatter networks for wireless-powered IoT | |
CN114900214B (en) | Low-complexity linear precoding algorithm based on OFDM system | |
Li et al. | Spatial multiuser access with MIMO smart antennas for OFDM systems | |
CN113691287B (en) | Large-scale MIMO system adaptive modulation method based on average bit error rate constraint | |
Liu et al. | A distributed detection algorithm for uplink massive MIMO systems | |
Ji et al. | Multipath extraction based UL/DL channel estimation for FDD massive MIMO-OFDM systems | |
Ravindran et al. | Direction of arrival and channel estimation using machine learning for multiple input multiple output system | |
CN114978254B (en) | Machine learning auxiliary low-complexity robust precoding algorithm based on OFDM system | |
CN108347267B (en) | Adaptive hybrid detection receiving method for large-scale MIMO |
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 |