CN113497770A - Fast time-varying channel parameter estimation method and device for OFDM system - Google Patents

Fast time-varying channel parameter estimation method and device for OFDM system Download PDF

Info

Publication number
CN113497770A
CN113497770A CN202010251557.8A CN202010251557A CN113497770A CN 113497770 A CN113497770 A CN 113497770A CN 202010251557 A CN202010251557 A CN 202010251557A CN 113497770 A CN113497770 A CN 113497770A
Authority
CN
China
Prior art keywords
channel
neural network
pilot
network model
data
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
Application number
CN202010251557.8A
Other languages
Chinese (zh)
Inventor
高飞飞
杨玉雯
钱婧
汪浩
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tsinghua University
HiSilicon Technologies Co Ltd
Original Assignee
Tsinghua University
HiSilicon Technologies Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Tsinghua University, HiSilicon Technologies Co Ltd filed Critical Tsinghua University
Priority to CN202010251557.8A priority Critical patent/CN113497770A/en
Publication of CN113497770A publication Critical patent/CN113497770A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • 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
    • H04L25/0224—Channel estimation using sounding signals
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Mathematical Physics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • Mathematical Analysis (AREA)
  • Computing Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computational Mathematics (AREA)
  • Molecular Biology (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Signal Processing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Power Engineering (AREA)
  • Databases & Information Systems (AREA)
  • Algebra (AREA)
  • Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)

Abstract

The embodiment of the invention provides a method and a device for estimating fast time-varying channel parameters of an OFDM system, wherein the method comprises the following steps: acquiring frequency domain sampling data and pilot frequency information of a received signal; estimating a base coefficient corresponding to time domain impact response of a channel according to the frequency domain sampling data and the pilot frequency information, and calculating impact response data of the time domain channel according to the base coefficient; inputting the impulse response data into a preset neural network model, and determining channel parameters of the received signals according to the output result of the neural network model; the preset neural network model is obtained after training according to the impact response data corresponding to the signal sample with the channel parameter label. The method estimates the base coefficient corresponding to the time domain impulse response of the channel through the pilot frequency information, does not need prior knowledge, does not relate to a large amount of matrix operations, has lower complexity, effectively improves the estimation performance compared with the current LS and LMMSE estimators, and simultaneously reduces the calculation complexity, thereby having stronger practicability.

Description

Fast time-varying channel parameter estimation method and device for OFDM system
Technical Field
The present invention relates to the field of communications technologies, and in particular, to a method and an apparatus for estimating fast time-varying channel parameters of an OFDM system.
Background
OFDM is an efficient multi-carrier modulation technique, and is widely used by various communication standards at present due to its strong capability of resisting multipath fading and eliminating inter-symbol interference, high spectrum utilization rate, and simple implementation of the receiver. The channel estimation is a precondition for coherent detection of the OFDM system, and the accuracy of the estimation has an important significance on the overall performance of the system. In a high-speed mobile environment, the time-varying characteristic of the channel is more obvious, and due to the multipath effect and the Doppler frequency shift, the wireless channel is changed into a fast time-varying frequency selective fading channel, so that the orthogonality among subcarriers of the OFDM system is damaged, the interference among the subcarriers is caused, and the accuracy of channel estimation is directly influenced.
Channel estimation techniques fall into three categories: decision feedback based channel estimation, blind or semi-blind channel estimation, and pilot-assisted based channel estimation. The pilot-based estimation method is the most commonly used method in the OFDM system because it is easy to implement and can track the variation of the wireless channel. In a fast time-varying channel, the channel varies within one OFDM symbol period, and the number of unknowns to be estimated within one symbol is much larger than the number of pilots, which makes channel estimation difficult. Many current research designs reduce the amount of estimation required for the channel by simplifying the channel model, a more popular simplified model being a base extension model. After the channel simplification model is determined, the main estimation criteria include Least Square (LS) estimation, Linear Minimum Mean Square Error (LMMSE) estimation and the like. The LS estimator is widely applied due to the fact that the LS estimator is simple to implement and does not need channel statistical information or other prior information, and performance of the LS estimator is slightly poor. The LMMSE estimation performance is superior to LS, but the LMMSE estimation performance needs a plurality of priori knowledge, a large number of matrix operations are involved, and the calculation complexity is high.
Disclosure of Invention
In order to solve the above problem, embodiments of the present invention provide a method and an apparatus for estimating a fast time-varying channel parameter of an OFDM system.
In a first aspect, an embodiment of the present invention provides a method for estimating a fast time-varying channel parameter of an OFDM system, including: acquiring frequency domain sampling data and pilot frequency information of a received signal; estimating a base coefficient corresponding to time domain impact response of a channel according to the frequency domain sampling data and the pilot frequency information, and calculating impact response data of the time domain channel according to the base coefficient; inputting the impulse response data into a preset neural network model, and determining channel parameters of received signals according to the output result of the neural network model; and the preset neural network model is obtained after training according to the impact response data corresponding to the signal sample with the channel parameter label.
Further, the pilot information is obtained by setting pilot clusters at equal intervals in OFDM symbols by the signal transmitting end.
Further, the acquiring frequency domain sample data and pilot information of the received signal includes: and carrying out frequency domain transformation on the received signal to obtain frequency domain sampling data and pilot frequency information.
Further, the estimating, according to the frequency domain sampling data and the pilot information, a base coefficient corresponding to a time domain impulse response of a channel includes: and estimating a base coefficient corresponding to a multipath channel time domain impact response according to frequency domain sampling data and pilot frequency information based on a complex exponential base extension (CE-BEM for short) model of the fast time-varying channel.
Further, the inputting the impulse response data into a preset neural network model includes: and respectively splicing real parts and imaginary parts of the impulse response data of the time domain channel into real vectors which are used as the input of the neural network, and outputting the real vectors through an output layer after passing through a full-connection hidden layer.
Further, before inputting the impulse response data into the preset neural network model, the method further includes: acquiring a plurality of sample signals under a plurality of channel parameters; dividing the sample signal into a training set and a test set, training the training set to obtain an initial model, and verifying whether the model reaches a preset precision by using the test set; if so, the model is the final network model, otherwise, the initial model is subjected to parameter adjustment by adopting a gradient descent method, and the training data is retrained until the preset precision is reached.
Further, the parameter adjustment of the initial model by using the gradient descent method includes: and during each training, performing back propagation according to a preset loss function, and updating the parameter value of the next iteration according to the learning rate and the gradient information.
In a second aspect, an embodiment of the present invention provides an apparatus for estimating a fast time-varying channel parameter of an OFDM system, including: the analysis module is used for acquiring frequency domain sampling data and pilot frequency information of the received signal; the processing module is used for estimating a base coefficient corresponding to time domain impact response of the channel according to the frequency domain sampling data and the pilot frequency information and calculating impact response data of the time domain channel according to the base coefficient; the output module is used for inputting the impact response data into a preset neural network model and determining the channel parameters of the received signals according to the output result of the neural network model; and the preset neural network model is obtained after training according to the impact response data corresponding to the signal sample with the channel parameter label.
In a third aspect, an embodiment of the present invention provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor executes the computer program to implement the steps of the method for estimating fast time varying channel parameters in an OFDM system according to the first aspect of the present invention.
In a fourth aspect, an embodiment of the present invention provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the steps of the fast time-varying channel parameter estimation method for the OFDM system according to the first aspect of the present invention.
According to the method and the device for estimating the fast time-varying channel parameters of the OFDM system, the base coefficient corresponding to the time domain impulse response of the channel is estimated through the pilot frequency information, prior knowledge is not needed, a large number of matrix operations are not involved, and the complexity is low. The channel parameters of the signals are obtained according to the impulse response of the time domain channel through the preset neural network model, and compared with the existing LS and LMMSE estimators, the estimation performance is effectively improved, and meanwhile, the calculation complexity is further reduced, so that the method has high practicability.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
Fig. 1 is a flowchart of a fast time varying channel parameter estimation method for an OFDM system according to an embodiment of the present invention;
fig. 2 is a schematic diagram of a pilot structure according to an embodiment of the present invention;
fig. 3 is a communication flow diagram of an OFDM system according to an embodiment of the present invention;
FIG. 4 is a graph of mean square error performance with signal to noise ratio variation for simulation experiments and conventional methods of the present invention;
fig. 5 is a structural diagram of a fast time-varying channel parameter estimation apparatus of an OFDM system according to an embodiment of the present invention;
fig. 6 is a schematic physical structure diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The research on a channel estimation method with low complexity and good performance has important academic and application values. In recent years, neural networks have been rapidly developed, which are capable of exploring inherent structures and characteristics based on learning of a large amount of data, giving accurate predictions when new observations are encountered, and whose combined application with wireless communication has been significantly successful in a variety of problems. Therefore, the neural network is applied to the estimation of the fast time-varying channel of the OFDM system, good performance can be obtained, the neural network can explore the inherent structural characteristics of the fast time-varying channel through offline training based on a large amount of data, and the neural network can still accurately estimate the channel even if the statistical characteristics of the channel change during online estimation.
Fig. 1 is a flowchart of a fast time-varying channel parameter estimation method for an OFDM system according to an embodiment of the present invention, and as shown in fig. 1, the fast time-varying channel parameter estimation method for an OFDM system according to an embodiment of the present invention includes:
101. and acquiring frequency domain sampling data and pilot frequency information of the received signal.
The "pilot" of the OFDM symbol subchannel is actually a subchannel signal composed of a certain frequency and having higher energy than the general signal, and the receiving end detects some points with higher energy, which are usually pilots. In the embodiment of the invention, the transmitting end of the signal is provided with pilot frequency clusters at equal intervals in advance in the transmitted OFDM symbols. In 101, the receiving end of the signal processes the received signal, and the frequency domain sampling data and the pilot frequency information are analyzed from the received signal. In a conventional method, a receiving end performs DFT on a received signal to obtain frequency domain sample data YpAnd pilot information Xp. Fig. 2 is a schematic diagram of a pilot structure according to an embodiment of the present invention, and as shown in fig. 2, each 9 pilot carriers may be set as a cluster, where the cluster includes 4 all-zero pilots on both sides and a non-zero pilot in the middle, each cluster is uniformly spaced by 23 carriers, and for transmitting data subcarriers, there are 8 pilot clusters in total. Fig. 3 is a communication flow chart of an OFDM system according to an embodiment of the present invention, and as shown in fig. 3, the communication flow of the OFDM system includes a processing step of a transmitting end for a transmitting signal and a processing step of a receiving end for a receiving signal.
102. And estimating a base coefficient corresponding to the time domain impulse response of the channel according to the frequency domain sampling data and the pilot frequency information, and calculating the impulse response data of the time domain channel according to the base coefficient.
The time domain channel impulse response is expressed as
Figure BDA0002435676170000051
Wherein B ═ B0,b1,…,bQ]The basis coefficients of the basis expansion model of dimension N (Q +1) are represented by Q +1 orthogonal basis expansion function vectors bqAnd (4) forming. The base coefficients are kept constant during one OFDM symbol period. Establishing a basis function bqFrequency domain response matrix Dq＝Fdiag{bq}FHSelecting a received pilot YpCorresponding row and sending pilot frequency XpElement composition D of corresponding columnPConstructing a matrix P ═ DpXpBased on LS estimation criteria, the estimated value of the base expansion coefficient is obtained
Figure BDA0002435676170000052
Where η is a small positive number. The impulse response data of the time domain channel is specifically a time domain channel matrix, and an initial estimation value of the time domain channel matrix is obtained according to an initial estimation basis coefficient
Figure BDA0002435676170000053
Wherein
Figure BDA0002435676170000054
In the form of vectorization of time domain channel matrix elements.
And 103, inputting the impulse response data into a preset neural network model, and determining channel parameters of the received signals according to the output result of the neural network model.
In 103, the preset neural network model is obtained by training signal samples with channel parameter labels, and the impulse response data of the sample signals are input into the neural network in the training process. The signal sample is a received signal of which the channel parameter is known in advance, the impulse response data is acquired at the same time, the corresponding channel parameter is used as a label of each sample signal, the corresponding impulse response data is used as the input of a model, and the training of the model is carried out, wherein the channel parameter is mainly the signal-to-noise ratio of the channel. After the neural network model is established, a large number of sample signals are trained to obtain a preset neural network model, and for a received signal to be detected, the acquired data form corresponding to the impulse response is input into the preset neural network model, so that corresponding channel parameters can be quickly and accurately obtained. Accordingly, the same method as in steps 101 to 102 is used for the acquisition method of the impulse response data of the training received signal samples.
The method for estimating the fast time-varying channel parameters of the OFDM system estimates the base coefficients corresponding to the time domain impulse response of the channel through the pilot frequency information, does not need prior knowledge, does not relate to a large amount of matrix operation, and has lower complexity. The channel parameters of the signals are obtained according to the impulse response of the time domain channel through the preset neural network model, and compared with the existing LS and LMMSE estimators, the estimation performance is effectively improved, and meanwhile, the calculation complexity is further reduced, so that the method has high practicability.
Based on the content of the foregoing embodiment, as an optional embodiment, the pilot information is obtained by setting pilot clusters at equal intervals in an OFDM symbol by a signal transmitting end.
The sending end sets pilot frequency clusters with equal intervals in OFDM symbols, and each group of pilot frequency clusters can be composed of all-zero protection pilot frequencies at two sides and non-zero pilot frequency in the middle. In the transmitting end, a stream of binary bitsnSerial-to-parallel conversion is carried out to divide the data into parallel data streams, the data streams are modulated onto subcarriers with different frequencies (such as QPSK modulation), and pilot frequency information X is insertedpThe pilot structure comprises pilot clusters with equal spacing, and the length of one pilot cluster is Lp＝2Lg+1, i.e. L on both sidesgAll-zero pilot and a non-zero intermediate pilot, each pilot cluster marked as
Figure BDA0002435676170000061
M-0, 1, …, M-1, pilot cluster data
Figure BDA0002435676170000062
pmFor pilot start sub-carriers, corresponding received information
Figure BDA0002435676170000063
All pilot marks
Figure BDA0002435676170000064
And transforming the frequency domain transmission signal X into a time domain signal X through IFFT, adding a cyclic prefix in front of each OFDM symbol, and then transmitting the signal. For example, the frequency domain transmission signal X is subjected to IFFT conversionnConversion to time-domain signals
Figure BDA0002435676170000065
And adding a cyclic prefix with the length of 16 before each OFDM symbol, and transmitting the cyclic prefix through an antenna.
The method for estimating the fast time-varying channel parameters of the OFDM system can record channel information by arranging pilot frequency clusters at equal intervals in OFDM symbols by a signal sending end, and analyze the pilot frequency information by a receiving end to obtain the channel parameters.
Based on the content of the foregoing embodiment, as an alternative embodiment, the acquiring frequency domain sample data and pilot information of a received signal includes: and carrying out frequency domain transformation on the received signal to obtain frequency domain sampling data and pilot frequency information.
Receiving end receives time domain signal
Figure BDA0002435676170000066
After the cyclic prefix is removed, the vector is expressed as y ═ Htx + w, wherein HtIs an N multiplied by N dimensional channel time domain impact response matrix:
Figure BDA0002435676170000071
performing frequency domain transformation such as FFT on the received signal to obtain a frequency domain received signal Fy ═ FHtx + Fw, where F is a Fourier transform matrix,
Figure BDA0002435676170000072
the frequency-domain received signal may be represented as Y ═ HX + W, where Y ═ Fy, X ═ Fx, W ═ Fw are frequency-domain versions of the received signal, the transmitted signal, and gaussian white noise, respectively, and H ═ FHtFHFor the frequency-domain impulse response of the channel, FHIn the form of a conjugate of a fourier transform matrix. For each transmitted pilot cluster data
Figure BDA0002435676170000073
pmAcquiring a received signal of a pilot position for a pilot start subcarrier
Figure BDA0002435676170000074
Wherein 2BcThe diagonal width of the frequency domain impulse response matrix for +1 channel. Forming M blocks of received frequency domain sample data into a vector
Figure BDA0002435676170000075
Based on the content of the foregoing embodiment, as an optional embodiment, estimating, according to frequency domain sample data and pilot information, a base coefficient corresponding to a time domain impulse response of a channel, includes: and estimating a base coefficient corresponding to the multi-path channel time domain impact response according to the pilot frequency based on the CE-BEM model of the fast time-varying channel.
The method specifically comprises the following steps: simulating a fast time-varying channel based on a CE-BEM model, and estimating a base coefficient corresponding to a multi-path channel time domain impact response based on an LS (least squares) criterion according to a pilot frequency
Figure BDA0002435676170000076
The initial estimation of the time domain channel matrix is obtained by using the base coefficient and vectorized into
Figure BDA0002435676170000077
The method specifically comprises the following steps: the time domain channel impulse response is expressed as
Figure BDA0002435676170000078
Wherein B ═ B0,b1,…,bQ]A basis expansion model of dimension N (Q +1) composed of Q +1 orthogonal basis expansion function vectors bqAnd (4) forming. The basis matrix is selected in various forms, such as CE-BEM, polynomial basis expansion (P-BEM for short) model, and discrete Carlo basis expansion (DKL-BEM for short) model. The complex exponential base construction method is the simplest, but the model error is large; the DKL-BEM is optimal under the mean square error criterion, but the construction of the basis matrix requires a known doppler shift. When a CE-BEM is adopted to simulate a fast time-varying channel model, the basic function in the CE-BEM is
Figure BDA0002435676170000079
And channel frequency domain matrix half bandwidth Bc＝Q/2。
Figure BDA00024356761700000710
For the channel parameter of the first path, hl＝[h0,l,…,hQ,l]TThe BEM coefficient corresponding to the first path is shown. The frequency domain received signal under the basis extension model is represented as
Figure BDA0002435676170000081
Wherein z is a basis function bqIn the frequency domain, Δq＝diag{FL[hq,0,…,hq,L-1]TFrequency domain response with diagonal matrix and diagonal elements as base coefficients, FLRepresentation matrix
Figure BDA0002435676170000082
The first L columns of (a).
When a complex exponential base is used,
Figure BDA0002435676170000083
Dqis a strict diagonal band matrix. For each transmitted pilot cluster data
Figure BDA0002435676170000084
pmFor pilot start sub-carriers, corresponding received information
Figure BDA0002435676170000085
Can be expressed as
Figure BDA0002435676170000086
Wherein
Figure BDA0002435676170000087
Is (L)p-2Bc)×MLpThe matrix of (a) is,
Figure BDA0002435676170000088
is a MLp×MLpThe diagonal matrices of (2) are all transmit pilot correlation matrices. Forming M blocks of received frequency domain data into a vector
Figure BDA0002435676170000089
Which can be represented as Yp＝DpXPh+d+Wp＝Ph+d+Wp. Constructing matrix P ═ DpXp，
Figure BDA00024356761700000810
Based on LS estimation criterion, the estimation value of the base expansion coefficient is obtained
Figure BDA00024356761700000811
Where η is a small positive number. Obtaining an initial estimation value in a matrix form of time domain channel impulse response according to the initial estimation basis coefficient
Figure BDA00024356761700000812
Wherein
Figure BDA00024356761700000813
In the form of vectorization of time domain channel matrix elements.
Based on the content of the foregoing embodiment, as an alternative embodiment, inputting the impulse response data into a preset neural network model includes: and respectively splicing real parts and imaginary parts of the impulse response data of the time domain channel into real vectors which are used as the input of the neural network, and outputting the real vectors through an output layer after passing through a full-connection hidden layer.
Combining complex vector time domain channel impulse responses
Figure BDA00024356761700000814
The real part and the imaginary part are separately spliced into a real vector serving as the input of the neural network, the real vector passes through a full-connection hidden layer, the dimensionality of the output layer is the same as that of the input layer, the output layer does not adopt an activation function, and the hidden layer can adopt a ReLu activation function.
In particular, a full-connection network of 2 layers may be constructed, with initial channel estimates received
Figure BDA00024356761700000815
1536 × 1 complex vector, a real vector whose real part and imaginary part are connected to 3072 × 1 is taken as the input of the neural network, and then the real vector of 3072 × 1 is output as the finally estimated channel parameter through a fully connected hidden layer containing 5000 neurons
Figure BDA00024356761700000816
After receiving a signal to be estimated, DFT conversion is carried out to obtain a frequency domain sampling signal and pilot frequency information, a time domain base coefficient is estimated through a base extension model, initial time domain channel impulse response is calculated, real and imaginary parts are extracted and spliced into real vectors which are input into a corresponding fully connected neural network, a final estimated value of a time domain channel is obtained, and performance evaluation is achieved through calculating a mean square error between the estimated value and a real value.
According to the method for estimating the fast time-varying channel parameters of the OFDM system, the real part and the imaginary part of the impulse response data of the time domain channel are spliced into the real vector which is used as the input of the neural network, and the real vector and the imaginary vector are output through the output layer after passing through the full-connection hidden layer, so that the accurate characteristic extraction can be performed on the impulse response data of the time domain channel, and the accuracy of the output channel parameters is ensured.
Based on the content of the foregoing embodiment, as an optional embodiment, before inputting the impulse response data into the preset neural network model, the method further includes: acquiring a plurality of sample signals under a plurality of channel parameters; dividing the sample signal into a training set and a test set, training the training set to obtain an initial model, and verifying whether the model reaches a preset precision by using the test set; if so, the model is the final network model, otherwise, the initial model is subjected to parameter adjustment by adopting a gradient descent method, and the training data is retrained until the preset precision is reached.
In particular, it may be in [0,5,10,15,20,25,30 ]]And under the signal-to-noise ratio of dB, respectively and randomly generating various channel parameters, sending random signals and obtaining a plurality of training samples. Training the training set to obtain initial models under multiple signal-to-noise ratios, and verifying whether the models achieve convergence. The training set can adopt 10000 sample data to divide into 50 batchs, and the test set is 1000 samples, utilizes training data to obtain initial model, utilizes test data to verify the model, carries out the parameter continuously to the model through Adam optimizer, sets up the learning rate and is 0.001, and the epoch is 300 and can reach the convergence, according to being based on
Figure BDA0002435676170000091
And calculating the error.
In addition, the OFDM signal is transmitted from the transmitting end and then reaches the receiving end via a multipath time-varying channel, and a received signal sample can be generated by simulation under each set of preset channel parameters. In the simulation process of the channel, the length of the channel is assumed to be 6, the normalized Doppler frequency shift is a random number between 0.2 and 1, each channel tap is generated by adopting a Jakes model, and the fading model of the channel is as follows:
h(t)＝hc(t)+jhs(t)
Figure BDA0002435676170000092
wherein the content of the first and second substances,
Figure BDA0002435676170000093
θ,φ,ψnsubject to uniform distribution of [ -pi, pi), the statistical information of the channel is Rhh(τ)＝J0(wdτ)，J0Is a first order bessel function. Tap power distribution compliance
Figure BDA0002435676170000094
According to the method for estimating the fast time-varying channel parameters of the OFDM system, the initial model is subjected to parameter adjustment by adopting a gradient descent method, and training data are retrained until the preset precision is reached, so that the method is beneficial to improving the accuracy of the preset neural network model.
Based on the content of the foregoing embodiment, as an optional embodiment, the performing parameter adjustment on the initial model by using a gradient descent method includes: and during each training, performing back propagation according to a preset loss function, and updating the parameter value of the next iteration according to the learning rate and the gradient information.
The neural network can adopt a gradient descent method to adjust parameters of the initial model in a training stage, and specifically comprises the step of carrying out L training in batch processing each time1And (5) reversely propagating the error as a loss function, and updating the parameter value of the next iteration according to the learning rate and the gradient information. L can be as follows1Loss function:
Figure BDA0002435676170000101
FIG. 4 is a graph of mean square error performance with signal-to-noise ratio variation for simulation experiments and conventional methods of the present invention. As can be seen from the figure, the time-varying channel estimation method based on the neural network of the embodiment of the present invention is significantly superior to the LS estimator and the LMMSE estimator which select different basis functions. In the figure, CE-BEM LS estimation, P-BEM LMMSE estimation and DKL-BEM LMMSE estimation are divided into the three basic matrix models mentioned above and are realized based on LS estimation or LMMSE estimation; the neural network method is a method of an embodiment of the present invention.
The channel estimation method combines the traditional time-varying channel estimation with the strong learning capacity of the neural network, and the neural network can explore the inherent characteristics of the time-varying channel through off-line training of a large amount of data without channel prior knowledge, thereby obtaining better estimation performance.
Fig. 5 is a structural diagram of a fast time-varying channel parameter estimation apparatus of an OFDM system according to an embodiment of the present invention, and as shown in fig. 5, the fast time-varying channel parameter estimation apparatus of the OFDM system includes: a parsing module 501, a processing module 502 and an output module 503. The analysis module 501 is configured to obtain frequency domain sampling data and pilot frequency information of a received signal; the processing module 502 is configured to estimate a base coefficient corresponding to a time domain impulse response of a channel according to the frequency domain sampling data and the pilot frequency information, and calculate impulse response data of the time domain channel according to the base coefficient; the output module 503 is configured to input the impulse response data into a preset neural network model, and determine a channel parameter of the received signal according to an output result of the neural network model; and the preset neural network model is obtained after training according to the impact response data corresponding to the signal sample with the channel parameter label.
The device embodiment provided in the embodiments of the present invention is for implementing the above method embodiments, and for details of the process and the details, reference is made to the above method embodiments, which are not described herein again.
The fast time-varying channel parameter estimation device of the OFDM system provided by the embodiment of the invention estimates the base coefficient corresponding to the time domain impulse response of the channel through the pilot frequency information, does not need prior knowledge, does not relate to a large amount of matrix operation, and has lower complexity. The channel parameters of the signals are obtained according to the impulse response of the time domain channel through the preset neural network model, and compared with the existing LS and LMMSE estimators, the estimation performance is effectively improved, and meanwhile, the calculation complexity is further reduced, so that the method has high practicability.
Fig. 6 is a schematic entity structure diagram of an electronic device according to an embodiment of the present invention, and as shown in fig. 6, the electronic device may include: a processor 601, a communication Interface 602, a memory 603 and a bus 604, wherein the processor 601, the communication Interface 602 and the memory 603 complete communication with each other through the bus 604. The communication interface 602 may be used for information transfer of an electronic device. The processor 601 may call logic instructions in the memory 603 to perform a method comprising: acquiring frequency domain sampling data and pilot frequency information of a received signal; estimating a base coefficient corresponding to time domain impact response of a channel according to the frequency domain sampling data and the pilot frequency information, and calculating impact response data of the time domain channel according to the base coefficient; inputting the impulse response data into a preset neural network model, and determining channel parameters of the received signals according to the output result of the neural network model; and the preset neural network model is obtained after training according to the impact response data corresponding to the signal sample with the channel parameter label.
In addition, the logic instructions in the memory 603 may be implemented in the form of software functional units and stored in a computer readable storage medium when the logic instructions are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the above-described method embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In another aspect, an embodiment of the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program is implemented to perform the transmission method provided in the foregoing embodiments when executed by a processor, and for example, the method includes: acquiring frequency domain sampling data and pilot frequency information of a received signal; estimating a base coefficient corresponding to time domain impact response of a channel according to the frequency domain sampling data and the pilot frequency information, and calculating impact response data of the time domain channel according to the base coefficient; inputting the impulse response data into a preset neural network model, and determining channel parameters of the received signals according to the output result of the neural network model; and the preset neural network model is obtained after training according to the impact response data corresponding to the signal sample with the channel parameter label.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims (10)

1. A fast time-varying channel parameter estimation method for an OFDM system is characterized by comprising the following steps:
acquiring frequency domain sampling data and pilot frequency information of a received signal;
estimating a base coefficient corresponding to time domain impulse response of a channel according to the frequency domain sampling data and the pilot frequency information, and calculating impulse response data of the time domain channel according to the base coefficient;
inputting the impulse response data into a preset neural network model, and determining channel parameters of the received signals according to an output result of the neural network model;
and the preset neural network model is obtained after training according to the impact response data corresponding to the signal sample with the channel parameter label.
2. The method as claimed in claim 1, wherein the pilot information is obtained by setting pilot clusters at equal intervals in OFDM symbols at a signal transmitting end.
3. The method of claim 1, wherein the obtaining frequency domain sample data and pilot information of the received signal comprises:
and carrying out frequency domain transformation on the received signal to obtain frequency domain sampling data and pilot frequency information.
4. The method of claim 1, wherein the estimating the basis coefficients corresponding to the time-domain impulse response of the channel according to the frequency-domain sample data and the pilot information comprises:
and estimating a base coefficient corresponding to the time domain impulse response of the multipath channel based on a complex exponential base extension model of the fast time-varying channel according to the frequency domain sampling data and the pilot frequency information.
5. The method according to claim 1, wherein the inputting the impulse response data into a preset neural network model comprises:
and respectively splicing real parts and imaginary parts of the impulse response data of the time domain channel into real vectors which are used as the input of the neural network, and outputting the real vectors through an output layer after passing through a full-connection hidden layer.
6. The method according to claim 1, wherein before inputting the impulse response data into the neural network model, the method further comprises:
acquiring a plurality of sample signals under a plurality of channel parameters;
dividing the sample signal into a training set and a test set, training the training set to obtain an initial model, and verifying whether the model reaches a preset precision by using the test set; if so, the model is the final network model, otherwise, the initial model is subjected to parameter adjustment by adopting a gradient descent method, and the training data is retrained until the preset precision is reached.
7. The method of claim 6, wherein the parametrizing the initial model by using a gradient descent method comprises:
and during each training, performing back propagation according to a preset loss function, and updating the parameter value of the next iteration according to the learning rate and the gradient information.
8. An apparatus for estimating fast time-varying channel parameters of an OFDM system, comprising:
the analysis module is used for acquiring frequency domain sampling data and pilot frequency information of the received signal;
the processing module is used for estimating a base coefficient corresponding to the time domain impact response of the channel according to the frequency domain sampling data and the pilot frequency information and calculating the impact response data of the time domain channel according to the base coefficient;
the output module is used for inputting the impact response data into a preset neural network model and determining the channel parameters of the received signals according to the output result of the neural network model;
and the preset neural network model is obtained after training according to the impact response data corresponding to the signal sample with the channel parameter label.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executing the program implements the steps of the method for fast time varying channel parameter estimation for an OFDM system as claimed in any one of claims 1 to 7.
10. A non-transitory computer readable storage medium having stored thereon a computer program, which when executed by a processor, performs the steps of the method for fast time varying channel parameter estimation in an OFDM system according to any of claims 1 to 7.
CN202010251557.8A 2020-04-01 2020-04-01 Fast time-varying channel parameter estimation method and device for OFDM system Pending CN113497770A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010251557.8A CN113497770A (en) 2020-04-01 2020-04-01 Fast time-varying channel parameter estimation method and device for OFDM system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010251557.8A CN113497770A (en) 2020-04-01 2020-04-01 Fast time-varying channel parameter estimation method and device for OFDM system

Publications (1)

Publication Number Publication Date
CN113497770A true CN113497770A (en) 2021-10-12

Family

ID=77994537

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010251557.8A Pending CN113497770A (en) 2020-04-01 2020-04-01 Fast time-varying channel parameter estimation method and device for OFDM system

Country Status (1)

Country Link
CN (1) CN113497770A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115022135A (en) * 2022-07-15 2022-09-06 北京理工大学 Dual-channel selection LMMSE estimation and ML detection method based on AmBC
WO2023092310A1 (en) * 2021-11-23 2023-06-01 Oppo广东移动通信有限公司 Information processing method, model generation method, and devices

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1576741A1 (en) * 2002-12-27 2005-09-21 Da Tang Mobile Communications Equipment Co., Ltd. A calibrating method for time-varying channel in a time-slot division mobile communication system
CN102833194A (en) * 2012-08-27 2012-12-19 重庆邮电大学 OFDM (orthogonal frequency division multiplexing) channel estimation method based on symmetrical basis expansion model for quick time-varying channel
CN106254284A (en) * 2016-08-01 2016-12-21 西安电子科技大学昆山创新研究院 A kind of Quick-Change channel method of estimation based on LEO Satellite
CN110247719A (en) * 2019-06-17 2019-09-17 华北电力大学 The playback of 5G time varying channel and emulation mode based on machine learning

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1576741A1 (en) * 2002-12-27 2005-09-21 Da Tang Mobile Communications Equipment Co., Ltd. A calibrating method for time-varying channel in a time-slot division mobile communication system
CN102833194A (en) * 2012-08-27 2012-12-19 重庆邮电大学 OFDM (orthogonal frequency division multiplexing) channel estimation method based on symmetrical basis expansion model for quick time-varying channel
CN106254284A (en) * 2016-08-01 2016-12-21 西安电子科技大学昆山创新研究院 A kind of Quick-Change channel method of estimation based on LEO Satellite
CN110247719A (en) * 2019-06-17 2019-09-17 华北电力大学 The playback of 5G time varying channel and emulation mode based on machine learning

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
YUWEN YANG,FEIFEI GAO,XIAOLI MA,SHUN ZHANG: "Deep Learning-Based Channel Estimation for Doubly Selective Fading Channels", 《IEEE ACCESS》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023092310A1 (en) * 2021-11-23 2023-06-01 Oppo广东移动通信有限公司 Information processing method, model generation method, and devices
CN115022135A (en) * 2022-07-15 2022-09-06 北京理工大学 Dual-channel selection LMMSE estimation and ML detection method based on AmBC
CN115022135B (en) * 2022-07-15 2023-10-20 北京理工大学 AmBC-based dual-selection channel LMMSE estimation and ML detection method

Similar Documents

Publication Publication Date Title
CN113890796B (en) High-speed channel estimation device and method based on OTFS (optical transport System) modulation and demodulation
CN113507426B (en) OTFS modulation-based joint channel estimation and signal detection method and device
CN102664841B (en) Method for SC-FDE (single carrier-frequency domain equalization) system low complexity RLS self-adaption channel estimation
CN109756432B (en) OFDM channel estimation method and device
EP1842305A2 (en) Delay restricted channel estimation for multi-carrier systems
CN1937598A (en) Channel estimation method in orthogonal frequency-division multiplexing system and channel estimation device
CN110048972B (en) Underwater sound orthogonal frequency division multiplexing channel estimation method and system
CN112910810B (en) Synchronization method, apparatus and medium for mimo-ofdm system
CN113497770A (en) Fast time-varying channel parameter estimation method and device for OFDM system
US20060159203A1 (en) Method and apparatus for channel estimation of transmission channels with memory in digital telecommunications systems, and corresponding computer-program product
US20090180557A1 (en) Channel estimation device and related method of an orthogonal frequency division multiplexing system
Liu et al. RecNet: Deep learning-based OFDM receiver with semi-blind channel estimation
Zhou et al. Channel estimation and data detection for OFDM systems over fast-fading and dispersive channels
CN113055318B (en) Channel estimation method
Doukopoulos et al. Adaptive algorithms for blind channel estimation in OFDM systems
Nazzal et al. Sparse coding for transform domain-based sparse OFDM channel estimation
Jiang et al. A joint OFDM channel estimation and ICI cancellation for double selective channels
Kao et al. The Research of Frequency Domain Least Squares Channel Estimation in OFDM Systems.
Obradovic et al. EM-based semi-blind channel estimation method for MIMO-OFDM communication systems
Aboutorab et al. Channel estimation and ICI cancellation for high mobility pilot-aided MIMO-OFDM systems
Fathi et al. Deep learning for channel estimation and signal detection
CN114143145B (en) Channel estimation method based on deep learning
Al-Makhlasawy et al. New Joint Channel Estimation and Interference Cancellation Method for FBMC Systems Based on Recurrent Neural Networks
Inkilä OFDM Channel Estimation With Machine Learning
CN117914653A (en) Time-varying MIMO-OFDM system block plane channel estimation method based on deep learning

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
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20211012