WO2022261842A1 - 预编码矩阵确定方法、装置、用户设备、基站及存储介质 - Google Patents
预编码矩阵确定方法、装置、用户设备、基站及存储介质 Download PDFInfo
- Publication number
- WO2022261842A1 WO2022261842A1 PCT/CN2021/100213 CN2021100213W WO2022261842A1 WO 2022261842 A1 WO2022261842 A1 WO 2022261842A1 CN 2021100213 W CN2021100213 W CN 2021100213W WO 2022261842 A1 WO2022261842 A1 WO 2022261842A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- matrix
- module
- network
- sub
- pilot
- Prior art date
Links
- 239000011159 matrix material Substances 0.000 title claims abstract description 356
- 238000000034 method Methods 0.000 title claims abstract description 85
- 238000006243 chemical reaction Methods 0.000 claims description 62
- 238000012549 training Methods 0.000 claims description 33
- 230000005540 biological transmission Effects 0.000 claims description 30
- 238000012545 processing Methods 0.000 claims description 24
- 230000017105 transposition Effects 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 5
- 238000004088 simulation Methods 0.000 claims description 4
- 230000008054 signal transmission Effects 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 abstract description 20
- 230000009466 transformation Effects 0.000 abstract description 6
- 230000006870 function Effects 0.000 description 43
- 238000013507 mapping Methods 0.000 description 23
- 230000004913 activation Effects 0.000 description 22
- 238000010586 diagram Methods 0.000 description 14
- 238000005516 engineering process Methods 0.000 description 14
- 238000004891 communication Methods 0.000 description 13
- 230000008859 change Effects 0.000 description 7
- 238000004590 computer program Methods 0.000 description 4
- 238000010606 normalization Methods 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 4
- 230000005236 sound signal Effects 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 3
- 235000000332 black box Nutrition 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 239000003795 chemical substances by application Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000003068 static 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
-
- 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
-
- 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
-
- 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
- H04L25/0242—Channel estimation channel estimation algorithms using matrix methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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
- H04L25/03891—Spatial equalizers
Definitions
- the present disclosure relates to the field of communication technologies, and in particular to a precoding matrix determination method, device, user equipment, base station, and storage medium.
- the data transmission process is adaptive to the change of the channel state, and the performance of data transmission is improved.
- the method for determining the precoding matrix mainly includes: determining the precoding matrix based on a black-box network or an unfolded network. Specifically, channel estimation is performed at the base station to construct a training data set, and then a black-box network structure or expanded network structure that can learn the mapping relationship from channel state information to precoding matrix is constructed at the base station, and based on the training data set, the The black-box network structure or the unfolded network structure is trained to obtain a trained network structure, and a precoding matrix is determined based on the trained network structure.
- the method of determining the precoding matrix based on the black box network lacks interpretability, and has poor mapping ability to nonlinear operations (such as inversion and other losses), resulting in poor calculation results.
- explicit channel state information needs to be obtained , affecting the calculation performance of the precoding matrix.
- the network design in the method of determining the precoding matrix based on the expanded network is more complicated, and the deep learning training framework cannot be used, and the efficiency is low.
- explicit channel state information needs to be obtained, which affects the calculation performance of the precoding matrix.
- the precoding matrix determination method, device, user equipment, base station, and storage medium proposed in the present disclosure are used to solve the problem of low calculation performance, poor calculation effect, and relatively complicated calculation method of the precoding matrix in the existing precoding matrix determination method technical problem.
- the method for determining the precoding matrix proposed in the embodiment of the first aspect of the present disclosure is applied to the base station, including:
- a precoding matrix W corresponding to the channel matrix H is determined based on the conversion matrix H p .
- the method for determining the precoding matrix proposed in the embodiment of the second aspect of the present disclosure is applied to the UE, including:
- a receiving module configured to receive received pilot information T from the UE, and determine a pilot vector T 2 based on the received pilot information T;
- a first processing module configured to determine a channel matrix H based on the pilot vector T2, and determine a conversion matrix Hp based on the channel matrix H ;
- the second processing module is configured to determine a precoding matrix W corresponding to the channel matrix H based on the conversion matrix H p .
- a processing module configured to determine transmission pilot information P based on the pilot data s, and send the transmission pilot information P to the base station.
- a user equipment provided in an embodiment of the fifth aspect of the present disclosure includes: a transceiver; a memory; and a processor, which are respectively connected to the transceiver and the memory, and configured to execute computer-executable instructions on the memory, The wireless signal transmission and reception of the transceiver is controlled, and the method proposed in the embodiment of the second aspect above can be implemented.
- a base station which includes: a transceiver; a memory; and a processor connected to the transceiver and the memory respectively, configured to execute computer-executable instructions on the memory , controlling the wireless signal sending and receiving of the transceiver, and implementing the method proposed in the embodiment of the first aspect above.
- the computer storage medium provided by the embodiment, wherein the computer storage medium stores computer-executable instructions; after the computer-executable instructions are executed by a processor, the method as described above can be implemented.
- the base station first receives the received pilot information T from the UE, and based on the received pilot
- the information T determines the pilot vector T 2 ; then, the channel matrix H is determined based on the pilot vector T 2 , and finally the conversion matrix H p is determined based on the channel matrix H, and the precoding matrix W is determined based on the conversion matrix H p . That is, in the embodiment of the present disclosure, the mapping relationship between the pilot vector T 2 and the precoding matrix W is learned, so as to determine the precoding matrix based on the pilot vector T 2 by using the implicit channel estimation technology based on the mapping relationship.
- the coding matrix W can ensure the precoding performance.
- the precoding matrix W is not determined directly based on the channel matrix H, but the conversion matrix H p that is more informative and linearly mapped to the precoding matrix W is determined based on the channel matrix H, and then , and then linearly map the precoding matrix H based on the conversion matrix H p , so that the determination of the precoding matrix has certain interpretability, ensures the calculation performance, and can obtain a better precoding matrix.
- the methods in the embodiments of the present disclosure are relatively simple and low in complexity.
- the second sub-network and the fourth sub-network are independently trained first, and then the precoding matrix determination network is trained as a whole, which can ensure that each sub-network will be in a
- the optimal state of the entire network is learned on the basis of the suboptimal, which ensures the training accuracy.
- FIG. 1 is a schematic flowchart of a method for determining a precoding matrix provided by an embodiment of the present disclosure
- FIG. 2 is a schematic structural diagram of a second sub-network provided by an embodiment of the present disclosure
- FIG. 3 is a schematic structural diagram of a third sub-network provided by an embodiment of the present disclosure.
- FIG. 4 is a schematic structural diagram of a fourth subnetwork provided by an embodiment of the present disclosure.
- FIG. 5 is a schematic flowchart of a method for determining a precoding matrix provided by another embodiment of the present disclosure
- FIG. 6 is a schematic flowchart of a method for determining a precoding matrix provided by another embodiment of the present disclosure.
- FIG. 7 is a schematic flowchart of a method for determining a precoding matrix provided by another embodiment of the present disclosure.
- FIG. 8 is a schematic structural diagram of an apparatus for determining a precoding matrix provided by an embodiment of the present disclosure.
- FIG. 9 is a schematic structural diagram of an apparatus for determining a precoding matrix provided by an embodiment of the present disclosure.
- Fig. 10 is a block diagram of a user equipment provided by an embodiment of the present disclosure.
- Fig. 11 is a block diagram of a base station provided by an embodiment of the present disclosure.
- first, second, third, etc. may use the terms first, second, third, etc. to describe various information, the information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of the embodiments of the present disclosure, first information may also be called second information, and similarly, second information may also be called first information.
- first information may also be called second information
- second information may also be called first information.
- the words "if” and "if” as used herein may be interpreted as “at” or "when” or "in response to a determination.”
- Fig. 1 is a schematic flowchart of a method for determining a precoding matrix provided by an embodiment of the present disclosure, which is applied to a base station. As shown in Fig. 1, the method for determining a precoding matrix may include the following steps:
- Step 101 Receive received pilot information T from a UE (User equipment, user equipment), and determine a pilot vector T 2 based on the received pilot information T.
- a UE User equipment, user equipment
- a UE may be a device that provides voice and/or data connectivity to a user.
- UE can communicate with one or more core networks via RAN (Radio Access Network, wireless access network).
- RAN Radio Access Network, wireless access network
- UE can be an Internet of Things terminal, such as a sensor device, a mobile phone (or called a "cellular" phone) and a device with an Internet of Things
- the computer of the terminal for example, may be a fixed, portable, pocket, hand-held, computer-built-in or vehicle-mounted device.
- station Station, STA
- subscriber unit subscriber unit
- subscriber station subscriber station
- mobile station mobile station
- mobile station mobile
- remote station remote station
- access point remote terminal
- user terminal or user agent.
- the UE may also be a device of an unmanned aerial vehicle.
- the UE may also be a vehicle-mounted device, for example, it may be a trip computer with a wireless communication function, or a wireless terminal connected externally to the trip computer.
- the UE may also be a roadside device, for example, it may be a street lamp, a signal lamp, or other roadside devices with a wireless communication function.
- the received pilot information T may specifically be the transmitted pilot information P obtained after the first sub-network at the UE side processes the pilot data s after being transmitted to the base station after channel interference earned.
- the UE first obtains pilot data s, and uses the first subnetwork to process the pilot data s to obtain transmission pilot information P, and then transmits the transmission
- the pilot information P is sent to the base station, so that the base station determines a precoding matrix based on the sent pilot information P.
- the pilot signal s may be pre-agreed by the base station and the UE on the same time-frequency resource.
- the first subnetwork at the UE end may specifically include a P-layer fully-connected layer; wherein, the dimension of the t-th fully-connected layer in the first subnetwork is q t ⁇ 1, and q t+1 ⁇ q t , where both P and q t are positive integers.
- the dimensions of the input information of the first fully connected layer of the first subnetwork may be 2 ⁇ K ⁇ L ⁇ 1, K is the number of UEs corresponding to the base station, and L is the pilot length; the last of the first subnetwork
- the dimension of the input information of a fully connected layer can be 2 ⁇ K ⁇ L ⁇ 1.
- the transmission pilot information P when the transmission pilot information P is transmitted to the base station via a channel, it will be interfered by the channel so that the information finally transmitted to the base station becomes the reception pilot information T, wherein the reception pilot information
- the frequency information T is related to the interference information of the channel (that is, the channel matrix H), so that the base station can subsequently determine the channel matrix H based on the received pilot information T, and then determine the precoding matrix W based on the channel matrix H, so that the subsequent transmission of data
- the data to be sent will be preprocessed through the precoding matrix W, so that the data transmission process can adapt to the change of the channel state and improve the performance of data transmission.
- the base station after the base station receives the received pilot information T, it will first determine the pilot vector T 2 based on the received pilot information, so that it can be subsequently determined based on the pilot vector T 2 Channel matrix H.
- T 2 [Re(T), Im(T); Re(s), Im(s); P u ]; where Re( ⁇ ) is real part, Im( ⁇ ) is the imaginary part; P u is the uplink signal-to-noise ratio.
- the base station may correspond to multiple UEs, wherein each UE corresponds to pilot data s, and each UE may be used to send the above-mentioned transmission to the base station Pilot information P, so that the UE receives received pilot information T from each UE, and then determines a channel matrix H corresponding to a channel between each UE and the base station based on the received pilot information T.
- the set of pilot data s corresponding to each UE is called pilot data set S, where, K is the number of UEs, s 1 , s 2 ,..., s K are mutually orthogonal.
- Step 102 Determine the channel matrix H based on the pilot vector T 2 , and determine the conversion matrix H p based on the channel matrix H.
- the method for determining the channel matrix H based on the pilot vector T 2 may include: inputting the pilot vector T 2 into the second sub-network to generate the channel matrix H.
- the second subnetwork may be a fully connected structure.
- the transmission pilot information P of each UE is also orthogonal to each other, so that P H P is a diagonal matrix, at this time It is easier to calculate, so it can be considered that there is a linear mapping between H and (Y; P), then the second subnetwork of the fully connected structure can be used to determine the channel matrix H based on the pilot vector T 2 .
- the second subnetwork may include sequentially connected F layers of fully connected layers, where F is a positive integer; wherein, the dimension of the tth fully connected layer of the second subnetwork is l t ⁇ 1, l t is a positive integer, and l 1 ⁇ 2 ⁇ M ⁇ N+2 ⁇ K ⁇ L+1, l l+1 ⁇ l t , where M is the number of base station antennas, and N is the channel noise .
- each fully connected layer of the second subnetwork can have a normalization layer
- the output of each fully connected layer can have an activation function layer.
- the activation function of the activation function layer of the first fully connected layer of the second subnetwork and the fully connected layer of the middle layer adopts the ReLU function:
- the activation function of the activation function layer of the last fully connected layer of the second subnetwork adopts the LeakyReLU function:
- X is the input data of the t-th fully-connected layer of the second subnetwork.
- the fully connected operation in the second subnetwork can be defined as:
- y i is the i-th element output by the second sub-network
- W i, j is the ( i , j)-th element in the fully connected weight matrix of the second sub-network
- bi is the fully connected weight matrix of the second sub-network
- Xi i is the i-th element input by the second sub-network.
- FIG. 2 is a schematic structural diagram of a second sub-network provided by an embodiment of the present disclosure.
- the second sub-network includes four layers of fully connected layers, wherein the first layer
- the dimensions of the fully connected layer, the second fully connected layer, the third fully connected layer, and the fourth fully connected layer are l 1 ⁇ 1, l 2 ⁇ 1, l 3 ⁇ 1, l 4 ⁇ 1, where , l 1 ⁇ 2 ⁇ M ⁇ N+2 ⁇ K ⁇ L+1, l 1 >l 2 >l 3 >l 4 .
- the input of each fully connected layer also has a normalization layer
- the output of each fully connected layer also has an activation function layer.
- the activation function of the activation function layer of the first layer of fully connected layer, the second layer of fully connected layer, and the third layer of fully connected layer adopts the ReLU function
- the activation function of the activation function layer of the fourth layer of fully connected layer adopts the LeakyReLU function
- the channel matrix H can be generated based on the input pilot vector T 2 by adopting the second sub-network structure.
- the conversion matrix Hp may be specifically determined based on the channel matrix H , so as to subsequently generate the precoding matrix W based on the conversion matrix Hp.
- the calculation formula of the precoding matrix W in the MMSE algorithm is the following formula 1:
- A, B, and C are all learnable parameters, which can be trained by the third sub-network.
- a H , B H , CH , D are learnable parameters, which can be trained by the third sub-network.
- FIG. 3 is a schematic structural diagram of a third sub-network provided by an embodiment of the present disclosure.
- the third sub-network may include: a sequentially connected matrix reconstruction ⁇ module, an operation module, and Inverse reconstruction of the ⁇ -1 module.
- the input end of the matrix reconstruction module is used as the input end of the third sub-network
- the output end of the inverse reconstruction is used as the output end of the third sub-network.
- the operation module may specifically include: a first multiplication module 301, a second multiplication module 302, a third multiplication module 303, a first transposition module, and a second transposition module.
- the output terminal of the matrix reconstruction module is respectively connected to the first input terminal of the first multiplication module 301, the input terminal of the first transpose module, the first input terminal of the second multiplication module 302, and the first input terminal of the third multiplication module 303.
- the second input end of the first multiplication module 301 is connected with the output end of the first transpose module, and the output end of the first multiplication module 301 is connected with the input end of the addition module 304, and the addition module 304 is used for performing adding p operation to the input data And output, the output end of the addition module 304 is respectively connected to the input end of the second transpose module and the input end of the matrix operation module, and the matrix operation module is used for performing matrix operation and output to the input data and the identity matrix I.
- the matrix operation module is used to perform Hadamard product operation on the input data and the identity matrix I and output it.
- the output end of the second transpose module is connected to the second input end of the second multiplying module 302, and the output end of the second multiplying module 302 is connected to the second input end of the splicing module; the output end of the matrix operation module is connected to the third multiplying
- the second input terminal of the module 303 and the output terminal of the third multiplication module 303 are connected to the third input terminal of the splicing module; the output terminal of the splicing module is connected to the input terminal of the inverse reconstruction module.
- the third sub-network will output the transformation matrix H p .
- Step 103 Determine the precoding matrix W corresponding to the channel matrix H based on the conversion matrix H p .
- the method for determining the precoding matrix W corresponding to the channel matrix H based on the conversion matrix Hp may include:
- A, B, C, and D are learnable parameters.
- the fourth sub-network can be made into a fully connected structure, and the conversion matrix H p to the precoding matrix W can be realized.
- the fourth subnetwork may include G layers of fully connected layers connected in sequence, and G is a positive integer; wherein, the dimension of the tth fully connected layer of the fourth subnetwork is f t ⁇ 1, f t is a positive integer, f 1 >3 ⁇ 6 ⁇ M ⁇ K, f t+1 ⁇ ft t .
- the dimension of the input data of the first fully connected layer of the fourth subnetwork may be 6 ⁇ M ⁇ K; the dimension of the input data of the last fully connected layer of the fourth subnetwork
- the dimension of the output information may be 2 ⁇ M ⁇ K.
- each fully connected layer of the fourth subnetwork can have a normalization layer
- the output of each fully connected layer can have an activation function layer.
- the activation function of the activation function layer of the first fully connected layer of the fourth subnetwork and the fully connected layer of the middle layer adopts the ReLU function:
- the activation function of the activation function layer of the last fully connected layer of the fourth subnetwork adopts the Tanh function:
- X is the input data of the fully connected layer of the tth layer of the fourth subnetwork.
- the activation function layer of the last fully connected layer in the fourth subnetwork is further provided with a power constraint layer, so as to ensure that the output precoding matrix can satisfy the transmission power constraint.
- FIG. 4 is a schematic structural diagram of a fourth sub-network provided by an embodiment of the present disclosure.
- the fourth sub-network includes four layers of fully connected layers, wherein the first layer
- the dimensions of the fully connected layer, the second fully connected layer, the third fully connected layer, and the fourth fully connected layer are f 1 ⁇ 1, f 2 ⁇ 1, f 3 ⁇ 1, f 4 ⁇ 1 respectively, where , f 1 >3 ⁇ 6 ⁇ M ⁇ K, f 1 >f 2 >f 3 >f 4 .
- the input of each fully connected layer also has a normalization layer
- the output of each fully connected layer also has an activation function layer.
- the activation function of the first fully connected layer of the fourth subnetwork, the second fully connected layer, and the activation function layer of the third fully connected layer adopts the ReLU function
- the activation function of the fourth fully connected layer adopts the LeakyReLU function
- the activation function layer of the fourth fully connected layer also has a power constraint layer.
- the precoding matrix W can be generated based on the input conversion matrix H p by adopting the fourth sub-network structure.
- the base station first receives the received pilot information T from the UE, and determines the pilot vector T 2 based on the received pilot information T; then , determine the channel matrix H based on the pilot vector T 2 , finally determine the conversion matrix H p based on the channel matrix H, and determine the precoding matrix W based on the conversion matrix H p . That is, in the embodiment of the present disclosure, the mapping relationship between the pilot vector T 2 and the precoding matrix W is learned, so as to determine the precoding matrix based on the pilot vector T 2 by using the implicit channel estimation technology based on the mapping relationship.
- the coding matrix W can ensure the precoding performance.
- the precoding matrix W is not determined directly based on the channel matrix H, but the conversion matrix H p that is more informative and linearly mapped to the precoding matrix W is determined based on the channel matrix H, and then , and then linearly map the precoding matrix H based on the conversion matrix H p , so that the determination of the precoding matrix has certain interpretability, ensures the calculation performance, and can obtain a better precoding matrix.
- the methods in the embodiments of the present disclosure are relatively simple and low in complexity.
- FIG. 5 is a schematic flowchart of a method for determining a precoding matrix provided by another embodiment of the present disclosure, which is applied to a base station. As shown in FIG. 5, the method may include:
- Step 501 based on the structure of the first subnetwork at the UE side, deploy a simulated subnetwork with the same structure as the first subnetwork at the base station side, and deploy a second subnetwork, a third subnetwork, and a fourth subnetwork at the base station side .
- Step 502 train the second sub-network based on the pilot data set S to obtain a pre-trained second sub-network.
- the pilot data set Wherein, s k is the pilot data of the Kth UE corresponding to the base station, and s 1 , s 2 ,..., s K are mutually orthogonal.
- the method for training the second subnetwork based on the pilot data set S to obtain the pretrained second subnetwork may include:
- Step a Obtain pilot data samples.
- N pilot data sets S may be obtained first as pilot data samples.
- the base station there may be 8 antennas at the base station, and the base station corresponds to 4 UEs configured with a single antenna, and 100,000 pilots are generated in a 2.4GHz outdoor picocell scenario Data set S is used as pilot data samples.
- the 100,000 pilot data samples may be divided into training samples including 90,000 samples and testing samples including 10,000 samples.
- Step b Determine a plurality of training data for training the second sub-network based on the pilot data samples.
- the training data used for training the second subnetwork includes a pilot vector T 2 .
- Step c input the plurality of training data for training the second sub-network into the second sub-network to obtain the predicted channel matrix H 1 , and calculate the loss function MSE based on the predicted channel matrix H 1 and the actual channel matrix H 2 .
- the loss function MSE can be:
- ⁇ is the Euclidean norm.
- Step d Update the parameters of the second sub-network according to the loss function MSE.
- the number of times of training of the second subnetwork may be set to 100 times.
- the network parameters are set as follows: Adam optimizer, initial learning rate 0.01, dynamic learning rate change strategy, and model strategy saved according to the test set.
- Step 503 train the fourth sub-network to obtain a pre-trained fourth sub-network.
- the method for training the fourth subnetwork based on the pilot data set S to obtain the pretrained fourth subnetwork may include:
- Step 1 Obtain a transformation matrix H pK. corresponding to the actual channel matrix of the channel between the Kth UE and the base station.
- the actual channel matrix H 2K of the channel between the Kth UE and the base station is determined, and the actual channel of the channel between the Kth UE and the base station is determined based on the actual channel matrix H 2K matrix corresponding to the transformation matrix H pK .
- Step 2 Input the conversion matrix H pK into the fourth subnetwork to generate a predictive precoding matrix W K corresponding to the channel between the Kth UE and the base station, and calculate a loss function Loss based on the predictive precoding matrix W K .
- the loss function Loss may be:
- ⁇ 2 is the channel noise power of the channel between the UE and the base station.
- Step 3 The parameters of the fourth sub-network are updated according to the loss function Loss.
- the number of times of training of the second subnetwork may be set to 100 times.
- the network parameters are set as follows: Adam optimizer, initial learning rate 0.01, dynamic learning rate change strategy, and model strategy saved according to the test set.
- channel matrices mentioned in the embodiments of the present disclosure can be regarded as channel matrices corresponding to perfect channels.
- Step 504 Concatenate the simulated sub-network, the pre-trained second sub-network, the third sub-network, and the pre-trained fourth sub-network in sequence to obtain a precoding matrix determination network, and determine the precoding matrix based on the pilot data set S
- the network is trained to obtain the optimal precoding matrix to determine the network.
- the network parameters when training the precoding matrix to determine the network parameters, are set to: Adam optimizer, initial learning rate 0.01, dynamic learning rate change strategy, save the model strategy according to the test set, and the number of training iterations is 100, the loss function is the above Loss function.
- the input of the simulation sub-network is the pilot data s
- the output is the transmission pilot information P.
- the precoding matrix determination network will learn "based on the transmission The pilot information P determines the received pilot information T, and then determines the pilot vector T 2 "based on the received pilot information T to obtain the pilot vector T 2 , and then inputs the pilot vector T 2 to the second subnetwork for subsequent training steps.
- the data T 1 can also be determined based on the pilot data set S; And the data T1 is used as the input of the simulation sub-network, so that the simulation sub-network outputs the transmission pilot information P, so as to train the precoding matrix determination network.
- the obtained optimal precoding matrix determines that the loss function of the network is the smallest, and the corresponding total rate sum of the K UEs is the largest.
- Step 505 Determine the network parameters corresponding to the simulated sub-network after training, and send the network parameters corresponding to the simulated sub-network to the UE.
- the network parameters corresponding to the simulated sub-networks in the optimal precoding matrix determination network can be obtained, and the network parameters can be fed back The link is fed back to the UE, so that the UE can adjust the first subnetwork deployed on the UE side according to the network parameter.
- step 505 the network for determining the optimal precoding matrix can also be tested, and then the next steps can be performed, namely: determining based on the optimal precoding matrix
- the network determines the precoding matrix.
- Step 506 Receive received pilot information T from the UE, and determine a pilot vector T 2 based on the received pilot information T.
- Step 507 Determine the channel matrix H based on the pilot vector T 2 , and determine the conversion matrix H p based on the channel matrix H.
- Step 508 Determine the precoding matrix W corresponding to the channel matrix H based on the conversion matrix Hp.
- the base station first receives the received pilot information T from the UE, and determines the pilot vector T 2 based on the received pilot information T; then , determine the channel matrix H based on the pilot vector T 2 , finally determine the conversion matrix H p based on the channel matrix H, and determine the precoding matrix W based on the conversion matrix H p . That is, in the embodiment of the present disclosure, the mapping relationship between the pilot vector T 2 and the precoding matrix W is learned, so as to determine the precoding matrix based on the pilot vector T 2 by using the implicit channel estimation technology based on the mapping relationship.
- the coding matrix W can ensure the precoding performance.
- the precoding matrix W is not determined directly based on the channel matrix H, but the conversion matrix H p that is more informative and linearly mapped to the precoding matrix W is determined based on the channel matrix H, and then , and then linearly map the precoding matrix H based on the conversion matrix H p , so that the determination of the precoding matrix has certain interpretability, ensures the calculation performance, and can obtain a better precoding matrix.
- the methods in the embodiments of the present disclosure are relatively simple and low in complexity.
- the second sub-network and the fourth sub-network are trained independently first, and then the precoding matrix determination network is trained as a whole, which can ensure that each sub-network is The optimal state of the entire network is learned on the basis of optimality, which ensures the training accuracy.
- FIG. 6 is a schematic flowchart of a method for determining a precoding matrix provided by another embodiment of the present disclosure, which is applied to a UE. As shown in FIG. 6 , the method for determining a precoding matrix may include the following steps:
- Step 601 Determine transmission pilot information P based on pilot data s, and transmit transmission pilot information P to a base station.
- step 601 reference may be made to the description of the foregoing embodiments, and details are not described here in the embodiments of the present disclosure.
- the base station first receives the received pilot information T from the UE, and determines the pilot vector T 2 based on the received pilot information T; then , determine the channel matrix H based on the pilot vector T 2 , finally determine the conversion matrix H p based on the channel matrix H, and determine the precoding matrix W based on the conversion matrix H p . That is, in the embodiment of the present disclosure, the mapping relationship between the pilot vector T 2 and the precoding matrix W is learned, so as to determine the precoding matrix based on the pilot vector T 2 by using the implicit channel estimation technology based on the mapping relationship.
- the coding matrix W can ensure the precoding performance.
- the precoding matrix W is not determined directly based on the channel matrix H, but the conversion matrix H p that is more informative and linearly mapped to the precoding matrix W is determined based on the channel matrix H, and then , and then linearly map the precoding matrix H based on the conversion matrix H p , so that the determination of the precoding matrix has certain interpretability, ensures the calculation performance, and can obtain a better precoding matrix.
- the methods in the embodiments of the present disclosure are relatively simple and low in complexity.
- FIG. 7 is a schematic flowchart of a method for determining a precoding matrix provided by another embodiment of the present disclosure, which is applied to a UE. As shown in FIG. 7 , the method for determining a precoding matrix may include the following steps:
- Step 701 receiving network parameters sent by the base station.
- Step 702 adjust the first sub-network based on network parameters.
- Step 703 Determine transmission pilot information P based on the pilot data s, and transmit transmission pilot information P to the base station.
- steps 701-703 For details about steps 701-703, reference may be made to the description of the foregoing embodiments, and the embodiments of the present disclosure will not repeat them here.
- the base station first receives the received pilot information T from the UE, and determines the pilot vector T 2 based on the received pilot information T; then , determine the channel matrix H based on the pilot vector T 2 , finally determine the conversion matrix H p based on the channel matrix H, and determine the precoding matrix W based on the conversion matrix H p . That is, in the embodiment of the present disclosure, the mapping relationship between the pilot vector T 2 and the precoding matrix W is learned, so as to determine the precoding matrix based on the pilot vector T 2 by using the implicit channel estimation technology based on the mapping relationship.
- the coding matrix W can ensure the precoding performance.
- the precoding matrix W is not determined directly based on the channel matrix H, but the conversion matrix H p that is more informative and linearly mapped to the precoding matrix W is determined based on the channel matrix H, and then , and then linearly map the precoding matrix H based on the conversion matrix H p , so that the determination of the precoding matrix has certain interpretability, ensures the calculation performance, and can obtain a better precoding matrix.
- the methods in the embodiments of the present disclosure are relatively simple and low in complexity.
- FIG. 8 is a schematic structural diagram of an apparatus for determining a precoding matrix provided by an embodiment of the present disclosure. As shown in FIG. 8, the apparatus may include:
- a receiving module 801 configured to receive received pilot information T from the UE, and determine a pilot vector T 2 based on the received pilot information T;
- the first processing module 802 is configured to determine a channel matrix H based on the pilot vector T2, and determine a conversion matrix Hp based on the channel matrix H ;
- the second processing module 803 is configured to determine a precoding matrix W corresponding to the channel matrix H based on the conversion matrix H p .
- the base station will first receive the received pilot information T from the UE, and will determine the pilot vector T2 based on the received pilot information T; then , determine the channel matrix H based on the pilot vector T 2 , finally determine the conversion matrix H p based on the channel matrix H, and determine the precoding matrix W based on the conversion matrix H p . That is, in the embodiment of the present disclosure, the mapping relationship between the pilot vector T 2 and the precoding matrix W is learned, so as to determine the precoding matrix based on the pilot vector T 2 by using the implicit channel estimation technology based on the mapping relationship.
- the coding matrix W can ensure the precoding performance.
- the precoding matrix W is not determined directly based on the channel matrix H, but the conversion matrix H p that is more informative and linearly mapped to the precoding matrix W is determined based on the channel matrix H, and then , and then linearly map the precoding matrix H based on the conversion matrix H p , so that the determination of the precoding matrix has certain interpretability, ensures the calculation performance, and can obtain a better precoding matrix.
- the methods in the embodiments of the present disclosure are relatively simple and low in complexity.
- the received pilot information T is: the transmitted pilot information P obtained after the first subnetwork at the UE side processes the pilot data s is transmitted to The base station is obtained after channel interference;
- T 2 [Re(T), Im(T); Re(s), Im(s); P u ];
- Re( ⁇ ) is the real part
- Im( ⁇ ) is the imaginary part
- P u is the uplink signal-to-noise ratio
- the pilot data is pre-agreed by the base station and the UE on the same time-frequency resource.
- the receiving module is further configured to:
- the pilot vector T 2 is input into the second sub-network to generate the channel matrix H.
- the second subnetwork is a fully connected structure
- the second sub-network includes F layers of fully-connected layers connected in sequence, and F is a positive integer; wherein, the dimension of the t-th layer of the fully-connected layer of the second sub-network is 1 t ⁇ 1, and 1 t is a positive integer, l 1 ⁇ 2 ⁇ M ⁇ N+2 ⁇ K ⁇ L+1, l t+1 ⁇ l t , where M is the number of base station antennas, K is the number of UEs corresponding to the base station, and L is the pilot length, N is the channel noise.
- the full connection operation in the second subnetwork is defined as:
- y i is the i-th element output by the second sub-network
- W i, j is the (i, j)-th element in the fully connected weight matrix of the second sub-network
- b i is the i-th element of the second sub-network output
- Xi is the i -th element input to the second sub-network.
- the first processing module is further configured to:
- the third subnetwork includes a matrix reconstruction module, an operation module, and an inverse reconstruction module connected in sequence.
- the input terminal of the matrix reconstruction module is used as the input terminal of the third sub-network
- the output terminal of the inverse reconstruction is used as the output terminal of the third sub-network.
- the operation module includes: a first multiplication module, a second multiplication module, a third multiplication module, a first transposition module, a second transposition module, and a splicing module , add module, and matrix operation module;
- the output terminal of the matrix reconstruction module is respectively connected to the first input terminal of the first multiplication module, the input terminal of the first transpose module, the first input terminal of the second multiplication module, the The first input end of the third multiplication module and the first input end of the splicing module;
- the second input end of the first multiplication module is connected to the output end of the first transpose module, the output end of the first multiplication module is connected to the input end of the addition module, and the addition module is used for
- the input data performs the operation of adding ⁇ and outputs, and the output of the adding module is respectively connected to the input of the second transpose module and the input of the matrix operation module, and the matrix operation module is used to compare the input data with the identity matrix I do matrix operations and output;
- the output end of the second transposition module is connected to the second input end of the second multiplication module, and the output end of the second multiplication module is connected to the second input end of the splicing module; the matrix operation module The output end of the described third multiplication module is connected to the second input end of the module, and the output end of the third multiplication module is connected to the third input end of the splicing module; the output end of the splicing module is connected to the reverse The input of the reconstruction module.
- the second processing module is further configured to:
- A, B, C, and D are learnable parameters.
- the fourth subnetwork is a fully connected structure
- the fourth subnetwork includes G layers of fully connected layers connected in sequence, and G is a positive integer; wherein, the dimension of the tth fully connected layer of the fourth subnetwork is f t ⁇ 1, and f t is a positive integer, f 1 >3 ⁇ 6 ⁇ M ⁇ K, f t +1 ⁇ ft .
- the above-mentioned device is also used for:
- s k is the pilot data corresponding to the Kth UE, and s 1 , s 2 ,..., s K are mutually orthogonal.
- the above-mentioned device is also used for:
- the fourth sub-network is trained to obtain a pre-trained fourth sub-network.
- the above-mentioned device is also used for:
- the above-mentioned device is also used for:
- the precoding matrix determination network is trained.
- the above-mentioned device is also used for:
- FIG. 9 is a schematic structural diagram of an apparatus for determining a precoding matrix provided by another embodiment of the present disclosure. As shown in FIG. 9, the apparatus may include:
- a processing module configured to determine transmission pilot information P based on the pilot data s, and send the transmission pilot information P to the base station.
- the base station will first receive the received pilot information T from the UE, and will determine the pilot vector T2 based on the received pilot information T; then , determine the channel matrix H based on the pilot vector T 2 , finally determine the conversion matrix H p based on the channel matrix H, and determine the precoding matrix W based on the conversion matrix H p . That is, in the embodiment of the present disclosure, the mapping relationship between the pilot vector T 2 and the precoding matrix W is learned, so as to determine the precoding matrix based on the pilot vector T 2 by using the implicit channel estimation technology based on the mapping relationship.
- the coding matrix W can ensure the precoding performance.
- the precoding matrix W is not determined directly based on the channel matrix H, but the conversion matrix H p that is more informative and linearly mapped to the precoding matrix W is determined based on the channel matrix H, and then , and then linearly map the precoding matrix H based on the conversion matrix H p , so that the determination of the precoding matrix has certain interpretability, ensures the calculation performance, and can obtain a better precoding matrix.
- the methods in the embodiments of the present disclosure are relatively simple and low in complexity.
- the processing module is further configured to:
- the pilot data s is input to the first sub-network to output the transmission pilot information P; wherein, the pilot data is pre-agreed by the base station and the UE on the same time-frequency resource.
- the first subnetwork is a fully connected structure
- the first sub-network includes a P-layer fully-connected layer, and P is a positive integer; wherein, the dimension of the t-th layer fully-connected layer in the first sub-network is q t ⁇ 1, q t is a positive integer, and q t+1 ⁇ q t .
- the above-mentioned device is also used for:
- the first sub-network is adjusted based on the network parameters.
- the computer storage medium provided by the embodiments of the present disclosure stores an executable program; after the executable program is executed by a processor, the method shown in any one of FIG. 1 or FIG. 5 or FIG. 6 or FIG. 7 can be implemented.
- the present disclosure further proposes a computer program product, including a computer program, and when the computer program is executed by a processor, the method as shown in any one of FIG. 1 or FIG. 5 or FIG. 6 or FIG. 7 is implemented.
- the present disclosure further proposes a computer program.
- the program When the program is executed by a processor, the method as shown in any one of FIG. 1 or FIG. 5 or FIG. 6 or FIG. 7 can be realized.
- Fig. 10 is a block diagram of a user equipment UE1000 provided by an embodiment of the present disclosure.
- UE1000 may be a mobile phone, a computer, a digital broadcasting terminal device, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, and the like.
- UE1000 may include at least one of the following components: a processing component 1002, a memory 1004, a power supply component 1006, a multimedia component 1008, an audio component 1010, an input/output (I/O) interface 1012, a sensor component 1013, and a communication component 1016.
- Processing component 1002 generally controls the overall operations of UE 1000, such as those associated with display, phone calls, data communications, camera operations, and recording operations.
- the processing component 1002 may include at least one processor 1020 to execute instructions to complete all or part of the steps of the above-mentioned method.
- processing component 1002 can include at least one module to facilitate interaction between processing component 1002 and other components.
- processing component 1002 may include a multimedia module to facilitate interaction between multimedia component 1008 and processing component 1002 .
- the memory 1004 is configured to store various types of data to support operations at the UE 1000 . Examples of such data include instructions for any application or method operating on UE1000, contact data, phonebook data, messages, pictures, videos, etc.
- the memory 1004 can be realized by any type of volatile or non-volatile storage device or their combination, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), Magnetic Memory, Flash Memory, Magnetic or Optical Disk.
- SRAM static random access memory
- EEPROM electrically erasable programmable read-only memory
- EPROM erasable Programmable Read Only Memory
- PROM Programmable Read Only Memory
- ROM Read Only Memory
- Magnetic Memory Flash Memory
- Magnetic or Optical Disk Magnetic Disk
- the power supply component 1006 provides power to various components of the UE 1000 .
- Power component 1006 may include a power management system, at least one power supply, and other components associated with generating, managing, and distributing power for UE 1000 .
- the multimedia component 1008 includes a screen providing an output interface between the UE 1000 and the user.
- the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from a user.
- the touch panel includes at least one touch sensor to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense a boundary of a touch or slide action, but also detect a wake-up time and pressure related to the touch or slide operation.
- the multimedia component 1008 includes a front camera and/or a rear camera. When UE1000 is in operation mode, such as shooting mode or video mode, the front camera and/or rear camera can receive external multimedia data. Each front camera and rear camera can be a fixed optical lens system or have focal length and optical zoom capability.
- the audio component 1010 is configured to output and/or input audio signals.
- the audio component 1010 includes a microphone (MIC), which is configured to receive an external audio signal when the UE 1000 is in an operation mode, such as a call mode, a recording mode and a voice recognition mode. Received audio signals may be further stored in memory 1004 or sent via communication component 1016 .
- the audio component 1010 also includes a speaker for outputting audio signals.
- the I/O interface 1012 provides an interface between the processing component 1002 and a peripheral interface module, which may be a keyboard, a click wheel, a button, and the like. These buttons may include, but are not limited to: a home button, volume buttons, start button, and lock button.
- the sensor component 1013 includes at least one sensor, which is used to provide various aspects of status assessment for the UE 1000 .
- the sensor component 1013 can detect the open/close state of the device 1000, the relative positioning of components, such as the display and the keypad of the UE1000, the sensor component 1013 can also detect the position change of the UE1000 or a component of the UE1000, and the user and Presence or absence of UE1000 contact, UE1000 orientation or acceleration/deceleration and temperature change of UE1000.
- the sensor assembly 1013 may include a proximity sensor configured to detect the presence of nearby objects in the absence of any physical contact.
- the sensor assembly 1013 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications.
- the sensor component 1013 may also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor or a temperature sensor.
- Communication component 1016 is configured to facilitate wired or wireless communications between UE 1000 and other devices.
- UE1000 can access wireless networks based on communication standards, such as WiFi, 2G or 3G, or their combination.
- the communication component 1016 receives broadcast signals or broadcast related information from an external broadcast management system via a broadcast channel.
- the communication component 1016 also includes a near field communication (NFC) module to facilitate short-range communication.
- NFC near field communication
- the NFC module may be implemented based on Radio Frequency Identification (RFID) technology, Infrared Data Association (IrDA) technology, Ultra Wide Band (UWB) technology, Bluetooth (BT) technology and other technologies.
- RFID Radio Frequency Identification
- IrDA Infrared Data Association
- UWB Ultra Wide Band
- Bluetooth Bluetooth
- UE 1000 may be powered by at least one Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array ( FPGA), controller, microcontroller, microprocessor or other electronic components for implementing the above method.
- ASIC Application Specific Integrated Circuit
- DSP Digital Signal Processor
- DSPD Digital Signal Processing Device
- PLD Programmable Logic Device
- FPGA Field Programmable Gate Array
- controller microcontroller, microprocessor or other electronic components for implementing the above method.
- Fig. 11 is a block diagram of a base station 1100 provided by an embodiment of the present disclosure.
- base station 1100 may be provided as a base station.
- base station 1100 includes processing component 1126 , which further includes at least one processor, and a memory resource represented by memory 1132 for storing instructions executable by processing component 1122 , such as application programs.
- the application program stored in memory 1132 may include one or more modules each corresponding to a set of instructions.
- the processing component 1126 is configured to execute instructions, so as to execute any of the aforementioned methods applied to the base station, for example, the method shown in FIG. 1 .
- Base station 1100 may also include a power component 1126 configured to perform power management of base station 1100, a wired or wireless network interface 1150 configured to connect base station 1100 to a network, and an input-output (I/O) interface 1158.
- the base station 1100 can operate based on an operating system stored in the memory 1132, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, Free BSDTM or similar.
Landscapes
- Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- General Physics & Mathematics (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Power Engineering (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
Description
Claims (25)
- 一种预编码矩阵确定方法,其特征在于,应用于基站,包括:接收来自用户设备UE的接收导频信息T,基于所述接收导频信息T确定导频向量T 2;基于所述导频向量T 2确定信道矩阵H,并基于所述信道矩阵H确定转换矩阵H p;基于所述转换矩阵H p确定所述信道矩阵H对应的预编码矩阵W。
- 如权利要求1所述的方法,其特征在于,所述接收导频信息T为:所述UE端的第一子网络对导频数据s进行处理后所得到的发送导频信息P在传输至所述基站时经由信道干扰后所得;所述基于所述接收导频信息T确定导频向量T 2,包括:T 2=[Re(T),Im(T);Re(s),Im(s);P u];其中,Re(·)为取实部,Im(·)为取虚部;P u为上行链路信噪比。
- 如权利要求2所述的方法,其特征在于,所述导频数据为所述基站与所述UE在相同时频资源预先约定的。
- 如权利要求1所述的方法,其特征在于,所述基于所述导频向量T 2确定信道矩阵H,包括:将所述导频向量T 2输入至第二子网络中以生成所述信道矩阵H。
- 如权利要求4所述的方法,其特征在于,所述第二子网络为全连接结构;所述第二子网络包括依次连接的F层全连接层,F为正整数;其中,所述第二子网络的第t层全连接层的维度为l t×1,l t为正整数,l 1≥2×M×N+2×K×L+1,l t+1<l t,其中,M为基站端天线数量,K为所述基站对应的UE数量,L为导频长度,N为信道噪声。
- 如权利要求4所述的方法,其特征在于,所述第二子网络中的全连接操作定义为:y i=∑ jW i,jx i+b i;其中,y i为所述第二子网络输出的第i个元素,W i,j为所述第二子网络的全连接权重矩阵中第(i,j)个元素,b i为所述第二子网络的全连接偏置中第i个元素,X i为所述第二子网络输入的第i个元素。
- 如权利要求4所述的方法,其特征在于,所述基于所述信道矩阵H确定转换矩阵H p,包括:将所述信道矩阵H输入至第三子网络中以生成转换矩阵H p,所述转换矩阵H p=[Z HH;Z +H;H],其中,Z=(HH H+ρI),ρ为下行链路信噪比的倒数,Z +=ZοI,I为单位矩阵。
- 如权利要求7所述的方法,其特征在于,所述第三子网络包括依次连接的矩阵重构模块、运算模块、以及逆重构模块;其中,所述矩阵重构模块的输入端作为所述第三子网络的输入端,所述逆重构的输出端作为所述第三子网络的输出端。
- 如权利要求8所述的方法,其特征在于,所述运算模块包括:第一乘模块、第二乘模块、第三乘模块、第一转置模块、第二转置模块、拼接模块、加模块、以及矩阵运算模块;其中,所述矩阵重构模块的输出端分别连接于所述第一乘模块的第一输入端、所述第一转置模块的输入端、所述第二乘模块的第一输入端、所述第三乘模块的第一输入端、以及所述拼接模块的第一输入端;所述第一乘模块的第二输入端与所述第一转置模块的输出端连接,所述第一乘模块的输出端与所述加模块的输入端连接,所述加模块用于对输入数据执行加ρ操作并输出,所述加模块的输出端分别连接于第二转置模块的输入端和所述矩阵运算模块的输入端,所述矩阵运算模块用于对输入数据与单位矩阵I做矩阵运算并输出;所述第二转置模块的输出端连接于所述第二乘模块的第二输入端,所述第二乘模块的输出端连接于所述拼接模块的第二输入端;所述矩阵运算模块的输出端连接于所述第三乘 模块的第二输入端,所述第三乘模块的输出端连接于所述拼接模块的第三输入端;所述拼接模块的输出端连接于所述逆重构模块的输入端。
- 如权利要求7所述的方法,其特征在于,所述基于所述转换矩阵H p确定所述信道矩阵H对应的预编码矩阵,包括:将所述转换矩阵H p输入至第四子网络中以生成所述预编码矩阵W,其中,W=Z -1H≈(AZ+BZ ++C)H=AZH+BZ +H+CH+D;其中,A、B、C、D为可学习参数。
- 如权利要求10所述的方法,其特征在于,所述第四子网络为全连接结构;所述第四子网络包括依次连接的G层全连接层,G为正整数;其中,所述第四子网络的第t层全连接层的维度为f t×1,f t为正整数,f t>3×6×M×K,f t+1<f t。
- 如权利要求12所述的方法,其特征在于,所述方法还包括:对所述第四子网络进行训练以得到预训练第四子网络。
- 如权利要求13所述的方法,其特征在于,所述方法还包括:基于所述第一子网络的结构在所述基站端部署模拟子网络,所述模拟子网络与所述第一子网络结构相同。
- 如权利要求14所述的方法,其特征在于,所述方法还包括:依次连接所述模拟子网络、所述预训练第二子网络、所述第三子网络、所述预训练第 四子网络以得到预编码矩阵确定网络;对所述预编码矩阵确定网络进行训练。
- 如权利要求15所述的方法,其特征在于,所述方法还包括:确定训练完成后的所述模拟子网络对应的网络参数;将所述网络参数发送至UE。
- 一种预编码矩阵确定方法,其特征在于,应用于UE,包括:基于导频数据s确定发送导频信息P,并向基站发送所述发送导频信息P。
- 如权利要求17所述的方法,其特征在于,所述基于导频数据s确定发送导频信息P包括:将所述导频数据s输入至第一子网络以输出所述发送导频信息P;其中,所述导频数据为所述基站与所述UE在相同时频资源预先约定的。
- 如权利要求17所述的方法,其特征在于,所述第一子网络为全连接结构;所述第一子网络包括P层全连接层,P为正整数;其中,所述第一子网络中的第t层全连接层的维度为q t×1,q t为正整数,且q t+1<q t。
- 如权利要求17所述的方法,其特征在于,所述方法还包括:接收基站发送的网络参数;基于所述网络参数调整所述第一子网络。
- 一种预编码矩阵确定装置,其特征在于,包括:接收模块,用于接收来自UE的接收导频信息T,基于所述接收导频信息T确定导频向量T 2;第一处理模块,用于基于所述导频向量T 2确定信道矩阵H,并基于所述信道矩阵H 确定转换矩阵H p;第二处理模块,用于基于所述转换矩阵H p确定所述信道矩阵H对应的预编码矩阵W。
- 一种预编码矩阵确定装置,其特征在于,包括:处理模块,用于基于导频数据s确定发送导频信息P,并向基站发送所述发送导频信息P。
- 一种用户设备,其特征在于,包括:收发器;存储器;处理器,分别与所述收发器及所述存储器连接,配置为通过执行所述存储器上的计算机可执行指令,控制所述收发器的无线信号收发,并能够实现权利要求17至20任一项所述的方法。
- 一种基站,其特征在于,包括:收发器;存储器;处理器,分别与所述收发器及所述存储器连接,配置为通过执行所述存储器上的计算机可执行指令,控制所述收发器的无线信号收发,并能够实现权利要求1至16任一项所述的方法。
- 一种计算机存储介质,其中,所述计算机存储介质存储有计算机可执行指令;所述计算机可执行指令被处理器执行后,能够实现权利要求1至16或17至20任一项所述的方法。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/CN2021/100213 WO2022261842A1 (zh) | 2021-06-15 | 2021-06-15 | 预编码矩阵确定方法、装置、用户设备、基站及存储介质 |
EP21945434.5A EP4354754A1 (en) | 2021-06-15 | 2021-06-15 | Precoding matrix determination method and apparatus, user equipment, base station and storage medium |
CN202180001868.1A CN115956344A (zh) | 2021-06-15 | 2021-06-15 | 预编码矩阵确定方法、装置、用户设备、基站及存储介质 |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/CN2021/100213 WO2022261842A1 (zh) | 2021-06-15 | 2021-06-15 | 预编码矩阵确定方法、装置、用户设备、基站及存储介质 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022261842A1 true WO2022261842A1 (zh) | 2022-12-22 |
Family
ID=84526807
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2021/100213 WO2022261842A1 (zh) | 2021-06-15 | 2021-06-15 | 预编码矩阵确定方法、装置、用户设备、基站及存储介质 |
Country Status (3)
Country | Link |
---|---|
EP (1) | EP4354754A1 (zh) |
CN (1) | CN115956344A (zh) |
WO (1) | WO2022261842A1 (zh) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107332598A (zh) * | 2017-06-26 | 2017-11-07 | 浙江理工大学 | 一种基于深度学习的mimo***联合预编码和天线选择方法 |
CN107769895A (zh) * | 2016-08-19 | 2018-03-06 | 普天信息技术有限公司 | 一种干扰对齐方法和*** |
CN111092641A (zh) * | 2019-12-18 | 2020-05-01 | 重庆邮电大学 | 基于毫米波mimo***深度学习的混合预编码设计方法 |
WO2021084343A1 (en) * | 2019-10-30 | 2021-05-06 | Telefonaktiebolaget Lm Ericsson (Publ) | Uplink covariance estimation for su-mimo precoding in wireless cellular systems |
-
2021
- 2021-06-15 EP EP21945434.5A patent/EP4354754A1/en active Pending
- 2021-06-15 CN CN202180001868.1A patent/CN115956344A/zh active Pending
- 2021-06-15 WO PCT/CN2021/100213 patent/WO2022261842A1/zh active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107769895A (zh) * | 2016-08-19 | 2018-03-06 | 普天信息技术有限公司 | 一种干扰对齐方法和*** |
CN107332598A (zh) * | 2017-06-26 | 2017-11-07 | 浙江理工大学 | 一种基于深度学习的mimo***联合预编码和天线选择方法 |
WO2021084343A1 (en) * | 2019-10-30 | 2021-05-06 | Telefonaktiebolaget Lm Ericsson (Publ) | Uplink covariance estimation for su-mimo precoding in wireless cellular systems |
CN111092641A (zh) * | 2019-12-18 | 2020-05-01 | 重庆邮电大学 | 基于毫米波mimo***深度学习的混合预编码设计方法 |
Non-Patent Citations (2)
Title |
---|
KAREEM M. ATTIAH; FOAD SOHRABI; WEI YU: "Deep Learning Approach to Channel Sensing and Hybrid Precoding for TDD Massive MIMO Systems", ARXIV.ORG, CORNELL UNIVERSITY LIBRARY, 201 OLIN LIBRARY CORNELL UNIVERSITY ITHACA, NY 14853, 21 November 2020 (2020-11-21), 201 Olin Library Cornell University Ithaca, NY 14853 , XP081820028 * |
WANG SEN: "Research on Millimeter Wave Channel Estimation and Precoding Based on Deep Learning", CHINA MASTER’S THESES FULL-TEXT DATABASE, 31 January 2021 (2021-01-31), XP093015326 * |
Also Published As
Publication number | Publication date |
---|---|
EP4354754A1 (en) | 2024-04-17 |
CN115956344A (zh) | 2023-04-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2021248371A1 (zh) | 一种接入方法、接入装置及存储介质 | |
US20220247472A1 (en) | Coding method, decoding method, and device | |
JP2023525845A (ja) | パイロット情報シンボル送信方法、チャネル推定方法及び通信機器 | |
WO2021184217A1 (zh) | 信道状态信息测量方法、装置及计算机存储介质 | |
WO2019218832A1 (zh) | 上行数据传输指示方法、终端和网络侧设备 | |
US20140146192A1 (en) | Acoustic configuration of a wireless camera | |
WO2022261842A1 (zh) | 预编码矩阵确定方法、装置、用户设备、基站及存储介质 | |
WO2023159614A1 (zh) | 一种预编码矩阵确定方法及设备/存储介质/装置 | |
CN111278115B (zh) | 传输方法、配置方法及相关设备 | |
US20230137584A1 (en) | Multiple-input multiple-output mode configuration method and apparatus, and storage medium | |
WO2023056646A1 (zh) | 一种信道状态信息上报方法、装置及存储介质 | |
WO2023155106A1 (zh) | 信道状态指示反馈信息传输方法、通信设备和存储介质 | |
WO2023197329A1 (zh) | 定位方法、装置、通信设备和存储介质 | |
WO2022217609A1 (zh) | 信号检测网络的确定方法和装置 | |
CN116830671B (zh) | Trp选择方法及装置、通信设备、通信***及存储介质 | |
WO2023179540A1 (zh) | 信道预测方法、装置及无线通信设备 | |
WO2024130747A1 (zh) | 一种资源限制指示方法、装置及存储介质 | |
WO2024119382A1 (zh) | 信道状态信息上报方法、装置及存储介质 | |
WO2024087118A1 (zh) | 幅度系数的上报方法、装置及存储介质 | |
WO2022205229A1 (zh) | 用于pusch的通信方法、装置及存储介质 | |
WO2023040886A1 (zh) | 数据采集方法及装置 | |
WO2022237822A1 (zh) | 训练数据集获取方法、无线传输方法、装置及通信设备 | |
EP4354755A1 (en) | Information feedback method and apparatus, user equipment, base station, system model and storage medium | |
WO2023206104A1 (zh) | 无线通信方法、装置、通信设备及存储介质 | |
WO2024031635A1 (zh) | 一种多点协作传输方法、装置及存储介质 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 21945434 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 18569877 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2021945434 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
ENP | Entry into the national phase |
Ref document number: 2021945434 Country of ref document: EP Effective date: 20240111 |