CN114785643B - OFDM system channel estimation method based on deep learning - Google Patents
OFDM system channel estimation method based on deep learning Download PDFInfo
- Publication number
- CN114785643B CN114785643B CN202210409459.1A CN202210409459A CN114785643B CN 114785643 B CN114785643 B CN 114785643B CN 202210409459 A CN202210409459 A CN 202210409459A CN 114785643 B CN114785643 B CN 114785643B
- Authority
- CN
- China
- Prior art keywords
- channel
- output
- mpdcnn
- ofdm
- dimensional convolution
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000013135 deep learning Methods 0.000 title claims abstract description 11
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 6
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 claims description 17
- 239000000969 carrier Substances 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 7
- 239000003638 chemical reducing agent Substances 0.000 claims description 6
- 125000004122 cyclic group Chemical group 0.000 claims description 6
- 230000004044 response Effects 0.000 claims description 4
- 239000000654 additive Substances 0.000 claims description 3
- 230000000996 additive effect Effects 0.000 claims description 3
- 238000011176 pooling Methods 0.000 claims description 3
- 238000004891 communication Methods 0.000 abstract description 8
- 238000011056 performance test Methods 0.000 abstract 1
- 238000005562 fading Methods 0.000 description 4
- 238000004088 simulation Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
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
- H04L25/0228—Channel estimation using sounding signals with direct estimation from sounding signals
- H04L25/023—Channel estimation using sounding signals with direct estimation from sounding signals with extension to other symbols
- H04L25/0232—Channel estimation using sounding signals with direct estimation from sounding signals with extension to other symbols by interpolation between sounding signals
-
- 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
-
- 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
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L25/03012—Arrangements for removing intersymbol interference operating in the time domain
-
- 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
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L25/03159—Arrangements for removing intersymbol interference operating in the frequency domain
-
- 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
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L25/03165—Arrangements for removing intersymbol interference using neural networks
-
- 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)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Power Engineering (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Radio Transmission System (AREA)
Abstract
The invention belongs to the technical field of information and communication, and relates to an OFDM system channel estimation method based on deep learning. In order to further improve the performance of OFDM system channel estimation, the invention provides an OFDM system channel estimation method based on deep learning, which comprises the following steps: firstly, estimating channel state information of a pilot frequency position from a received signal by using a linear estimation method; secondly, obtaining a channel state information estimated value of the whole OFDM time-frequency resource block by using an interpolation algorithm; finally, the multi-scale parallel expansion convolutional neural network provided by the invention is used for further reducing the noise of the estimated channel state information. The performance test is carried out on the method under the complex TDL mixed scene, and the result shows that the method of the invention exceeds the existing method.
Description
Technical Field
The invention belongs to the technical field of information and communication, and relates to an OFDM system channel estimation method based on deep learning.
Background
Accurate channel estimation and effective channel information feedback are basic conditions for guaranteeing the performance of a wireless communication system. In recent years, deep Learning (DL) or AI has been widely studied in wireless communication systems, both in academia and industry. OFDM is one of the core technologies of the 4G and 5G standards, which can well support multiple access and has stable performance in a frequency selective fading environment. In a wireless communication system, a transmission signal is subject to various fading and multipath propagation. In order to demodulate the transmitted signal, a pilot signal is designed to estimate the channel information. The pilot signal is transmitted with the data signal, precoded using the same data signal, and subjected to similar channel fading. Since the position and sequence of the pilot signals are known to the receiver, the receiver can estimate the channel using the received signals.
Disclosure of Invention
In order to further improve the performance of OFDM system channel estimation, the invention provides an OFDM system channel estimation method based on deep learning.
The technical scheme adopted by the invention comprises the following steps:
s1, assume that the system has N t Each OFDM symbol, N f Sub-carriers, M Tx Multiple transmit antennas and M Rx A plurality of receiving antennas; assuming that the Cyclic Prefix (CP) length of the OFDM system is greater than the maximum channel delay, removing the cyclic prefix and performing discrete fourier transform (discrete Fourier transform, DFT), mth Rx The signal on the ith subcarrier of the kth OFDM symbol received by the receiving antenna can be expressed as
wherein ,
and />
Respectively corresponding frequency domain channel and mth Tx A precoding factor of the root antenna; s is(s) k，i For transmitting symbols +.>
Is additive white gaussian noise; since both pilot and data signals are transmitted with the same transmit precoder, the above equation can be rewritten as +.>
wherein ,
is a pre-coded channel. For all OFDM symbols and subcarriers, there are
wherein ,
s k，i and />
Are respectively->
S and />
(k, i) -th elements of (a); />
Representing the product of the corresponding position elements of the matrix;
generally, at total N t In the OFDM symbols, the pilot occupies N p，t ＜N t And, at total N f Among the number of carriers, the pilot occupies N p，f ＜N f A plurality of; for the mth Rx A root receiving antenna for receiving pilot signals as
At the receiving end, the least square method LS is used for obtaining the known pilot frequency S p From observation of
Channel state information +.>
I.e.
Wherein/represents the division of the matrix's corresponding element positions;
s2, obtaining the channel estimation value of the pilot frequency position of the OFDM time-frequency resource block
Then, obtaining the channel state information of the whole OFDM time-frequency resource block by interpolation algorithm>
For the kth position of OFDM time-frequency resource block pilot frequency position p1 and kp2 OFDM symbols, using DFT interpolation, i.e
wherein ,
and />
Respectively represent the mth Rx On the root receiving antenna, kth p1 All pilot carriers (i) p ) A vector formed by a frequency domain channel and a corresponding delay domain impulse response vector; the same method can obtain +.>
and />
For each row of OFDM time-frequency resource block, linear interpolation method is adopted, namely
So far, the channel estimation value of the whole OFDM time-frequency resource block can be obtained
S3, judging whether the multi-scale parallel expansion convolutional neural network (multi-scale parallel dilated convolutional neural network, MPDCNN) has completed training and storing; if the MPDCNN is trained and saved, executing a step S5, otherwise executing a step S4;
s4, performing off-line training on the MPDCNN to reduce noise of the channel estimation value of the whole OFDM time-frequency resource block; the specific structure of the proposed MPDCNN is shown in the accompanying figure 2: the input data firstly passes through a layer of two-dimensional convolution in the MPDCNN, then sequentially passes through a plurality of multi-scale parallel expansion convolution blocks (multi-scale parallel dilated convolutional block, MPDCB) and a layer of two-dimensional convolution, and finally, the output after the two-dimensional convolution is added with the data of the input MPDCNN to be used as the output of the MPDCNN; the input data firstly sequentially passes through a layer of two-dimensional convolution and three parallel expansion convolution blocks (ilated convolutional block, DCB) with different expansion rates in the MPDCB; then, the outputs of the 3 DCBs are spliced according to the dimension of the tensor channel and then sent to a channel attention module (channel attention, CA), and the outputs after CA are subjected to one-layer two-dimensional convolution; finally, adding the output after two-dimensional convolution with the data input to the MPDCB to be used as the output of the MPDCB; the input data can pass through a plurality of residual blocks (ResBlock) in the DCB, and finally, the output of the last ResBlock is added with the input of the DCB to be used as the output of the DCB; the input data sequentially passes through a two-dimensional convolution layer, a parameterized rectifying linear unit (parametric rectifed linear unit, PRelu) and a two-dimensional convolution layer in the ResBlock; finally, adding the output of the second two-dimensional convolution layer and the input of ResBlock to be used as the output of ResBlock; the input data sequentially pass through two-dimensional convolution, PRelu, two-dimensional convolution, global average pooling, 1X1 convolution, PRelu, 1X1 convolution and Sigmoid function in the CA module; finally, multiplying the output of the Sigmoid function and the input of the CA module in the channel dimension according to broadcast multiplication to be used as the output of the CA module;
define the MPDCNN noise reducer used as
Where θ is a network parameter. Handle->
After being sent into the noise reducer, the noise-reduced output can be obtained:
the loss function used in the offline training phase is
S5, loading the saved MPDCNN to reduce the noise of the channel estimation value of the whole OFDM time-frequency resource block, namely
The invention has the beneficial effects that: the method has the advantages that the structural characteristics of the OFDM system channels are utilized by using the deep neural network to reduce noise of the time-frequency domain resource block channels with larger noise, and the method has outstanding performance on the problem of estimating the OFDM system channels.
Drawings
Fig. 1 is a diagram illustrating pilot configuration in a 5G OFDM system used;
FIG. 2 is a flow chart of the proposed channel estimation method;
FIG. 3 is a block diagram of the proposed denoising network;
fig. 4 is a simulation curve of an OFDM system using different channel estimation methods.
Detailed Description
The following describes in further detail the embodiments of the present invention with reference to the drawings and examples.
A specific implementation method of the present invention based on the above method is given below, and parameters of the specific method are set as follows:
fig. 1 is a typical pilot resource allocation in a current 5G system. In the frequency domain, there are 12 subcarriers and 14 OFDM symbols in one Resource Block (RB). The pilot occupies 6 subcarriers and 2 OFDM symbols in one RB and one slot. In the simulation, 8 RBs shown in FIG. 1 are used in the frequency domain, and 14 OFDM symbols are used in the time domain, i.e., N t ＝14，N f ＝96，N p，t ＝2，N p，f =48. Number of transmitting antennas M Tx Number of receiving antennas m=32 Rx =2. The modulation scheme is QPSK.
The link-level simulator follows a 3GPP Tapped Delay Line (TDL) model, which has been calibrated. The carrier frequency is 3.5GHz, and the subcarrier space is 15KHz. A new hybrid training data construction method is presented to solve the generalization problem in DL. For training data, the channel model is a mixed model of TDL-A, TDL-B, TDL-C, TDL-D and TDL-E, wherein one sample randomly selects one channel model from the 5 models; the delay spread is randomly selected from 0 seconds to 300 ns; randomly generating the speed between 0km/h and 50 km/h; the signal-to-noise ratio of each sample is randomly chosen from 0dB to 20 dB.
In training the MPDCNN network, the training set and the test set contained 189000 samples (9000 samples per signal-to-noise ratio) and 21000 samples (1000 samples per signal-to-noise ratio), respectively. In MPDCNN, 4 different MPDCBs were used, each containing 4 different resblocks. Note that the first layer convolution of MPDCNN is 48 different 3×3×2 filters, the last layer convolution of MPDCNN is two different 3×3×48 filters, and the 1×1 convolution in CA is 48 different 1×1×48 filters. All other convolution layers of MPDCNN are 48 different 3×3×48 filters. An Adam optimizer was used and the learning rate was set to 1e -4 。
According to the above parameter settings, the specific steps of the simulation are as follows:
s1, assume that the system has N t Each OFDM symbol, N f Sub-carriers, M Tx Multiple transmit antennas and M Rx A plurality of receiving antennas; assuming that the Cyclic Prefix (CP) length of the OFDM system is greater than the maximum channel delay, removing the cyclic prefix and performing discrete fourier transform (discrete Fourier transform, DFT), mth Rx The signal on the ith subcarrier of the kth OFDM symbol received by the receiving antenna can be expressed as
wherein ,
and />
Respectively corresponding frequency domain channel and mth Tx A precoding factor of the root antenna; s is(s) k，i For transmitting symbols +.>
Is additive white gaussian noise; since both pilot and data signals are transmitted with the same transmit precoder, the above equation can be rewritten as
wherein ,
is a pre-coded channel. For all OFDM symbols and subcarriers, there are
wherein ,
s k，i and />
Are respectively->
S and />
(k, i) -th elements of (a); />
Representing the product of the corresponding position elements of the matrix;
generally, at total N t In the OFDM symbols, the pilot occupies N p，t ＜N t And, at total N f Among the number of carriers, the pilot occupies N p，f ＜N f A plurality of; for the mth Rx A root receiving antenna for receiving pilot signals as
At the receiving end, the least square method LS is used for obtaining the known pilot frequency S P From observation of
Channel state information +.>
I.e. < ->
Wherein/represents the division of the matrix's corresponding element positions;
s2, obtaining the channel estimation value of the pilot frequency position of the OFDM time-frequency resource block
Then, obtaining the channel state information of the whole OFDM time-frequency resource block by interpolation algorithm>
For the kth position of OFDM time-frequency resource block pilot frequency position p1 and kp2 OFDM symbols, using DFT interpolation, i.e
wherein ,
and />
Respectively represent the mth Rx On the root receiving antenna, kth p1 All pilot carriers (i) p ) A vector formed by a frequency domain channel and a corresponding delay domain impulse response vector; the same method can obtain +.>
and />
For each row of OFDM time-frequency resource block, linear interpolation method is adopted, namely
So far, the channel estimation value of the whole OFDM time-frequency resource block can be obtained
S3, judging whether the multi-scale parallel expansion convolutional neural network (multi-scale parallel dilated convolutional neural network, MPDCNN) has completed training and storing; if the MPDCNN is trained and saved, executing a step S5, otherwise executing a step S4;
s4, performing off-line training on the MPDCNN to reduce noise of the channel estimation value of the whole OFDM time-frequency resource block; the specific structure of the proposed MPDCNN is shown in the accompanying figure 2: the input data firstly passes through a layer of two-dimensional convolution in the MPDCNN, then sequentially passes through a plurality of multi-scale parallel expansion convolution blocks (multi-scale parallel dilated convolutional block, MPDCB) and a layer of two-dimensional convolution, and finally, the output after the two-dimensional convolution is added with the data of the input MPDCNN to be used as the output of the MPDCNN; the input data firstly sequentially passes through a layer of two-dimensional convolution and three parallel expansion convolution blocks (ilated convolutional block, DCB) with different expansion rates in the MPDCB; then, the outputs of the 3 DCBs are spliced according to the dimension of the tensor channel and then sent to a channel attention module (channel attention, CA), and the outputs after CA are subjected to one-layer two-dimensional convolution; finally, adding the output after two-dimensional convolution with the data input to the MPDCB to be used as the output of the MPDCB; the input data can pass through a plurality of residual blocks (ResBlock) in the DCB, and finally, the output of the last ResBlock is added with the input of the DCB to be used as the output of the DCB; the input data sequentially passes through a two-dimensional convolution layer, a parameterized rectifying linear unit (parametric rectifed linear unit, PRelu) and a two-dimensional convolution layer in the ResBlock; finally, adding the output of the second two-dimensional convolution layer and the input of ResBlock to be used as the output of ResBlock; the input data sequentially pass through two-dimensional convolution, PRelu, two-dimensional convolution, global average pooling, 1X1 convolution, PRelu, 1X1 convolution and Sigmoid function in the CA module; finally, multiplying the output of the Sigmoid function and the input of the CA module in the channel dimension according to broadcast multiplication to be used as the output of the CA module;
define the MPDCNN noise reducer used as
Where θ is a network parameter. Handle->
After being sent into the noise reducer, the noise-reduced output can be obtained:
the loss function used in the offline training phase is
S5, loading the saved MPDCNN to reduce the noise of the channel estimation value of the whole OFDM time-frequency resource block, namely
Fig. 4 is a simulation curve of different channel estimation methods. Wherein LS corresponds to the channel estimation result obtained after executing the first two steps of the method of the invention; reenet corresponds to channel estimation and results obtained using the method of literature 2; the channel net corresponds to the channel estimation result obtained by using the method in document 1; MPDCNN corresponds to the performance results of the proposed method. From the figure, the performance of the method provided by the invention is far superior to that of the other three methods, and the effectiveness of the method is proved.
Claims (2)
1. The OFDM system channel estimation method based on deep learning is characterized by comprising the following steps:
s1, defining system N t Each OFDM symbol, N f Sub-carriers, M Tx Multiple transmit antennas and M Rx A plurality of receiving antennas; the Cyclic Prefix (CP) length of the OFDM system is greater than the maximum channel delay, and after the cyclic prefix is removed and Discrete Fourier Transform (DFT) is performed, the mth Rx The kth OFDM symbol received by the receiving antenna, the signal on the ith subcarrier is expressed as:
wherein ,
and />
Respectively corresponding frequency domain channel and mth Tx A precoding factor of the root antenna; s is(s) k，i For transmitting symbols +.>
Is additive white gaussian noise; since both pilot and data signals are transmitted with the same transmit precoder, the above equation is rewritten as:
wherein ,
is a pre-coded channel; for all OFDM symbols and subcarriers, there are:
wherein ,
s k，i and />
Are respectively->
S and />
(k, i) -th element of (c); />
Representing the product of the corresponding position elements of the matrix;
at total N t In the OFDM symbols, the pilot occupies N p，t ＜N t And, at total N f Among the number of carriers, the pilot occupies N p，f ＜N f A plurality of; for the mth Rx The root receiving antenna receives the pilot signals:
wherein ,
S p and />
Are respectively->
S and />
A sub-matrix formed by corresponding pilot frequency position elements;
at the receiving end, the least square method LS is used for obtaining the known pilot frequency S p From observation of
Channel state information +.>
I.e.
Wherein/represents the division of the matrix's corresponding element positions;
s2, obtaining the channel estimation value of the pilot frequency position of the OFDM time-frequency resource block
Then, the channel state information of the whole OFDM time-frequency resource block is obtained by interpolation algorithm>
For the kth position of OFDM time-frequency resource block pilot frequency position p1 and kp2 The OFDM symbols adopt a DFT interpolation method, namely:
wherein ,
and />
Respectively represent the mth Rx On the root receiving antenna, kth p1 All pilot carriers (i) p ) A vector formed by a frequency domain channel and a corresponding delay domain impulse response vector; the same method can obtain +.>
and />
For each row of the OFDM time-frequency resource block, a linear interpolation method is adopted, namely:
so far, the channel estimation value of the whole OFDM time-frequency resource block is obtained
S3, judging whether the multi-scale parallel expansion convolutional neural network (MPDCNN) has completed training and storing; if the MPDCNN is trained and saved, executing a step S5, otherwise executing a step S4;
s4, performing off-line training on the MPDCNN to reduce noise of the channel estimation value of the whole OFDM time-frequency resource block; define the MPDCNN noise reducer used as
Wherein θ is a network parameter, handle->
After being sent into the noise reducer, the noise-reduced output is obtained:
the loss function used in the offline training phase is:
s5, loading the stored MPDCNN to reduce noise of the channel estimation value of the whole OFDM time-frequency resource block, namely:
2. the method for estimating the channel of the OFDM system based on deep learning as claimed in claim 1, wherein in the step S4, the training process of MPDCNN is:
the method comprises the steps that input data firstly pass through a layer of two-dimensional convolution in an MPDCNN, then pass through a plurality of multi-scale parallel expansion convolution blocks (MPDCB) and a layer of two-dimensional convolution in sequence, and finally, the output after the two-dimensional convolution is added with the data of the input MPDCNN to be used as the output of the MPDCNN;
the input data firstly sequentially passes through a layer of two-dimensional convolution and three parallel expansion convolution blocks (DCBs) with different expansion rates in the MPDCB; then, the outputs of the 3 DCBs are spliced according to the dimension of the tensor channel and then sent to a channel attention module (CA), and the outputs after CA are subjected to one-layer two-dimensional convolution; finally, adding the output after two-dimensional convolution with the data input to the MPDCB to be used as the output of the MPDCB;
the input data can pass through a plurality of residual blocks (ResBlock) in the DCB, and finally, the output of the last ResBlock is added with the input of the DCB to be used as the output of the DCB;
the input data sequentially passes through a two-dimensional convolution layer, a parameterized rectifying linear unit (PRelu) and a two-dimensional convolution layer in the Resblock; finally, adding the output of the second two-dimensional convolution layer and the input of ResBlock to be used as the output of ResBlock;
the input data sequentially pass through two-dimensional convolution, PRelu, two-dimensional convolution, global average pooling, 1X1 convolution, PRelu, 1X1 convolution and Sigmoid function in the CA module; and finally, multiplying the output of the Sigmoid function and the input of the CA module in the channel dimension according to broadcast multiplication to obtain the output of the CA module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210409459.1A CN114785643B (en) | 2022-04-19 | 2022-04-19 | OFDM system channel estimation method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210409459.1A CN114785643B (en) | 2022-04-19 | 2022-04-19 | OFDM system channel estimation method based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114785643A CN114785643A (en) | 2022-07-22 |
| CN114785643B true CN114785643B (en) | 2023-04-25 |
Family
ID=82430759
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210409459.1A Active CN114785643B (en) | 2022-04-19 | 2022-04-19 | OFDM system channel estimation method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114785643B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110365613A (en) * | 2019-07-01 | 2019-10-22 | 重庆大学 | A kind of channel estimation methods based on neural network prediction |
| CN111404849A (en) * | 2020-03-20 | 2020-07-10 | 北京航空航天大学 | OFDM channel estimation and signal detection method based on deep learning |
| CN114244675A (en) * | 2021-12-29 | 2022-03-25 | 电子科技大学 | MIMO-OFDM system channel estimation method based on deep learning |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7248559B2 (en) * | 2001-10-17 | 2007-07-24 | Nortel Networks Limited | Scattered pilot pattern and channel estimation method for MIMO-OFDM systems |
-
2022
- 2022-04-19 CN CN202210409459.1A patent/CN114785643B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110365613A (en) * | 2019-07-01 | 2019-10-22 | 重庆大学 | A kind of channel estimation methods based on neural network prediction |
| CN111404849A (en) * | 2020-03-20 | 2020-07-10 | 北京航空航天大学 | OFDM channel estimation and signal detection method based on deep learning |
| CN114244675A (en) * | 2021-12-29 | 2022-03-25 | 电子科技大学 | MIMO-OFDM system channel estimation method based on deep learning |
Non-Patent Citations (5)
| Title |
|---|
| Motorola.R1-071350 "Channel Estimation Performance for DL Control Channel in E-UTRA".3GPP tsg_ran\WG1_RL1.2007,(第TSGR1_48b期),全文. * |
| Samsung.R1-134196 "Link abstraction method for SLML receiver".3GPP tsg_ran\WG1_RL1.2013,(第TSGR1_74b期),全文. * |
| Xiaoyan Kuai ; Xiaojun Yuan ; Ying-Chang Liang.Message-Passing Based OFDM Receiver for Time-Varying Sparse Multipath Channels.Vehicular Technology, IEEE Transactions on.2018,第67卷(第10期),全文. * |
| 廖勇 ; 花远肖 ; 姚海梅 ; .基于深度学习的OFDM信道估计.重庆邮电大学学报(自然科学版).2019,(第03期),全文. * |
| 李勇 ; 袁晓军. MIMO多向中继信道在完整数据交换模型下的自由度.中国科学院大学学报.2020,全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114785643A (en) | 2022-07-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102687441B (en) | Channel estimation for OFDM systems | |
| CN102035767B (en) | Channel estimation method and device | |
| CN101951353B (en) | Channel estimation method for orthogonal frequency division multiplexing (OFDM) system under interference environment | |
| EP3154231A1 (en) | Doubly-selective channel compensation method, system and related device | |
| CN103685096A (en) | Optimal pilot frequency based MIMO-OFDM (Multiple Input Multiple Output-Orthogonal Frequency Division Multiplexing) system channel estimation method | |
| CN102571647B (en) | Frequency offset estimation method and device | |
| EP2381632B1 (en) | Method and a channel estimating arrangement for performing channel estimation | |
| CN107222438A (en) | The simplification BEM channel estimation methods of high-speed mobile SC FDMA systems | |
| CN103595679A (en) | Method for reducing peak-to-average ratio of LTE uplink single-carrier frequency-division multiple access signal | |
| CN102271102B (en) | Channel estimating method and equipment based on sliding window | |
| CN114363135B (en) | OTFS signal processing method and device | |
| CN102045285A (en) | Channel estimation method and device and communication system | |
| JP6220844B2 (en) | MIMO system testing apparatus and testing method | |
| CN114785643B (en) | OFDM system channel estimation method based on deep learning | |
| CN101848183A (en) | Channel estimation method and device in multiple input multiple output OFDM (Orthogonal Frequency Division Multiplexing) system | |
| CN101197796A (en) | Wireless sensor network channel evaluation method based on SC-FDE and virtual multi-antenna | |
| Lien et al. | Extended Kalman filter for channel and carrier frequency offset estimation | |
| Aida et al. | LMMSE channel estimation for block-pilot insertion in OFDM systems under time varying conditions | |
| CN114172779A (en) | Channel estimation method, device, equipment and storage medium | |
| CN104486274B (en) | A kind of method for transmitting signals of multiple antennas single carrier-frequency division multiple access system | |
| CN103179058A (en) | Method and apparatus for estimating channel impulse response length | |
| CN1275396C (en) | Guide signal structuring method in estimation of time-domain multiple transceiving channel | |
| CN113055318A (en) | Channel estimation method | |
| KR101501334B1 (en) | Method for Estimating Frequency Offset based on OFDM System and Apparatus Thereof | |
| Hao et al. | Semi-blind channel estimation of MIMO-OFDM systems based on RBF network |
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 |