CN114785433A - Channel scene recognition method, network device and storage medium - Google Patents
Channel scene recognition method, network device and storage medium Download PDFInfo
- Publication number
- CN114785433A CN114785433A CN202110088977.3A CN202110088977A CN114785433A CN 114785433 A CN114785433 A CN 114785433A CN 202110088977 A CN202110088977 A CN 202110088977A CN 114785433 A CN114785433 A CN 114785433A
- Authority
- CN
- China
- Prior art keywords
- channel
- scene
- value
- pilot
- symbol
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000012549 training Methods 0.000 claims abstract description 34
- 238000013528 artificial neural network Methods 0.000 claims abstract description 15
- 238000010586 diagram Methods 0.000 claims description 17
- 230000004044 response Effects 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 2
- 239000000126 substance Substances 0.000 claims 1
- 238000004891 communication Methods 0.000 abstract description 6
- 238000012545 processing Methods 0.000 description 10
- 238000012360 testing method Methods 0.000 description 7
- 238000005562 fading Methods 0.000 description 4
- 238000010606 normalization Methods 0.000 description 4
- 241000209094 Oryza Species 0.000 description 3
- 235000007164 Oryza sativa Nutrition 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 235000009566 rice Nutrition 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 2
- 239000013256 coordination polymer Substances 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000007480 spreading Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/391—Modelling the propagation channel
Abstract
The embodiment relates to the field of communication, in particular to a method for identifying a channel scene, network equipment and a storage medium, comprising the following steps: acquiring channel characteristics of a channel under a channel scene to be identified, wherein the channel characteristics comprise: the channel time domain characteristic is used for representing the multipath time delay expansion information and the direct path information of the channel scene to be identified, and the channel time correlation characteristic is used for representing the Doppler expansion information and the frequency offset information of the channel scene to be identified; inputting the channel characteristics into a preset channel scene recognition model, and obtaining a recognition result of the channel scene to be recognized, wherein the channel scene recognition model is obtained after training a preset neural network based on a characteristic data training set, and the characteristic data training set comprises channel characteristics of channels in at least 2 channel sample scenes. By adopting the embodiment, the channel scene can be quickly and accurately identified.
Description
Technical Field
The present disclosure relates to the field of communications, and in particular, to a method for identifying a channel scene, a network device, and a storage medium.
Background
In a wireless communication system, system performance is mainly affected by a wireless channel. Due to the fact that channel environments in wireless communication are complex and changeable, signals under different channel scenes have obvious differences in the aspects of energy, power, time delay, channel response and the like, different channel scenes are accurately identified, and then an optimal receiving algorithm and configuration parameters matched with channels are adopted for different channel scenes, so that the method has important significance for improving the performance of the whole communication system.
However, the currently adopted identification method is inaccurate in identifying the channel scene, or the identification steps are complex, and the implementation complexity is high.
Disclosure of Invention
The embodiments of the present application mainly aim to provide a method, a network device, and a storage medium for identifying a channel scene, which can quickly and accurately identify the channel scene.
In order to achieve the above object, an embodiment of the present application provides a method for channel scene identification, including: acquiring channel characteristics of a channel under a channel scene to be identified, wherein the channel characteristics comprise: the channel time domain characteristic is used for representing the multipath time delay expansion information and the direct path information of the channel scene to be identified, and the channel time correlation characteristic is used for representing the Doppler expansion information and the frequency offset information of the channel scene to be identified; inputting the channel characteristics into a preset channel scene recognition model, and obtaining a recognition result of the channel scene to be recognized, wherein the channel scene recognition model is obtained after training a preset neural network based on a characteristic data training set, and the characteristic data training set comprises channel characteristics of channels in at least 2 channel sample scenes.
In order to achieve the above object, an embodiment of the present application further provides a network device, including: at least one processor, and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of channel scene recognition described above.
In order to achieve the above object, an embodiment of the present application further provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the method for channel scene identification as described above is implemented.
According to the method for identifying the channel scene, the channel characteristics comprise a channel frequency domain characteristic, a channel time domain characteristic and a channel time correlation characteristic, the channel frequency domain characteristic reflects fading variation of the channel scene in a frequency domain, the channel time domain characteristic reflects multipath time delay expansion information and direct path information of the channel scene, the channel time correlation characteristic reflects Doppler expansion information and frequency offset information of the channel scene, the three characteristics describe more comprehensive channel characteristics from different dimensions, and accuracy of channel scene identification is improved; the feature data training set comprises channel features of channels under a plurality of channel scenes, so that a channel scene recognition model obtained by training a preset neural network based on the feature data training set is more accurate, the accuracy of channel scene recognition is improved, meanwhile, the acquisition modes of the channel frequency domain features, the channel time domain features and the channel time correlation features are simple, excessive computing resources are not consumed, the speed of channel scene recognition is improved, and due to the accuracy of the channel features, the channel scene model can recognize more channel scenes, and the application scenes of the recognition method are increased.
Drawings
Fig. 1 is a flow chart of a method of channel scene identification provided in accordance with a first embodiment of the present invention;
fig. 2 is a flow chart of a method of channel scene recognition provided in accordance with a second embodiment of the present invention;
fig. 3 is a structural diagram of channel characteristics in the case of 2 pilot symbols provided in a second embodiment according to the present invention;
fig. 4 is a structural diagram of channel characteristics in the case of a single pilot symbol provided in a second embodiment according to the present invention;
fig. 5 is a structural diagram of another channel characteristic in the case of 2 pilot symbols provided in the second embodiment according to the present invention;
FIG. 6 is a schematic diagram of a convolutional neural network provided in accordance with a second embodiment of the present invention;
FIG. 7 is a schematic diagram of a fully-linked neural network architecture provided in accordance with a second embodiment of the present invention;
fig. 8 is a flowchart of a method of channel scene recognition provided in a third embodiment of the present invention;
fig. 9 is a block diagram of a network device provided in a fourth embodiment according to the present invention.
Detailed Description
To make the objects, technical solutions and advantages of the embodiments of the present application clearer, the embodiments of the present application will be described in detail below with reference to the accompanying drawings. However, it will be appreciated by those of ordinary skill in the art that in the various embodiments of the present application, numerous technical details are set forth in order to provide a better understanding of the present application. However, the technical solution claimed in the present application can be implemented without these technical details and various changes and modifications based on the following embodiments. The following embodiments are divided for convenience of description, and should not constitute any limitation to the specific implementation manner of the present application, and the embodiments may be mutually incorporated and referred to without contradiction.
There are various channel scene recognition methods, for example, the method one: processing and analyzing the current scene by using a feature matrix which is transmitted by a wireless channel and contains an arrival angle and path loss so as to obtain a scene classification result in real time; the classification process is divided into an uplink part and a downlink part, and in a characteristic matrix of a downlink, the mobile station judges four scenes, namely open land, viaduct, mountain land and city, according to a path loss model; in the feature matrix of the uplink link, the transmitting station calculates a corresponding angle reference value according to the arrival angle obtained by each measurement, and compares the difference value of the arrival angles measured before and after with the angle reference value, thereby judging whether the mobile station is in a moving or static scene. The method is classified by comparing the angle reference values, is easily influenced by the angle reference values and is inaccurate in identification. The second method comprises the following steps: the channel scene is identified through the machine learning model, the Rice factor and the root mean square characteristic are required for input parameters, the Rice factor in the method is difficult to obtain in an actual product, the calculation complexity of the root mean square characteristic is high, the realization cost is high, and meanwhile, if the obtained Rice factor and the obtained root mean square characteristic are not accurate, the model is inaccurate.
The first embodiment of the present invention relates to a method for channel scene recognition, and the flow thereof is shown in fig. 1.
Step 101: acquiring channel characteristics of a channel under a channel scene to be identified, wherein the channel characteristics comprise: the method comprises the following steps of characterizing the channel frequency domain characteristics of fading variation of a channel scene to be identified in a frequency domain, characterizing the channel time domain characteristics of multipath time delay expansion information and direct path information of the channel scene to be identified, and characterizing the channel time correlation characteristics of Doppler expansion information and frequency offset information of the channel scene to be identified.
Step 102: inputting the channel characteristics into a preset channel scene recognition model, and acquiring a recognition result of a channel scene to be recognized, wherein the channel scene recognition model is obtained after training a preset neural network based on a characteristic data training set, and the characteristic data training set comprises the channel characteristics of channels in at least 2 channel sample scenes.
According to the method for identifying the channel scene, the channel characteristics comprise a channel frequency domain characteristic, a channel time domain characteristic and a channel time correlation characteristic, the channel frequency domain characteristic reflects fading variation of the channel scene in a frequency domain, the channel time domain characteristic reflects multipath time delay expansion information and direct path information of the channel scene, the channel time correlation characteristic reflects Doppler expansion information and frequency offset information of the channel scene, the three characteristics describe more comprehensive channel characteristics from different dimensions, and accuracy of channel scene identification is improved; the feature data training set comprises channel features of channels under a plurality of channel scenes, so that a channel scene recognition model obtained by training a preset neural network based on the feature data training set is more accurate, the accuracy of channel scene recognition is improved, meanwhile, the acquisition modes of the channel frequency domain features, the channel time domain features and the channel time correlation features are simple, excessive computing resources are not required to be consumed, the speed of channel scene recognition is improved, and due to the accuracy of the channel features, the channel scene model can recognize more channel scenes, and the application scenes of the recognition method are increased.
A second embodiment of the present invention relates to a method for channel scene recognition, and is a detailed description of the first embodiment, and a flow thereof is shown in fig. 2.
Step 201: and acquiring a pilot symbol of a specified channel in a channel scene to be identified and channel estimation on the pilot symbol.
In particular, the method for channel scene recognition is applied to a network device, such as a base station. The channel scene identification method can be applied to wireless communication systems, such as LTE and 5G NR wireless systems. The designated channel may be a Physical Uplink Shared Channel (PUSCH). A given channel may be configured with multiple pilot symbols, e.g., 1, 2, and more. For the sake of understanding, in the following, the wireless system of 5G NR is described by taking NR uplink PUSCH as an example.
For example, 2 pilot symbols are configured for the NR uplink PUSCH, the NR uplink PUSCH occupies 50 Resource Blocks (RB), occupies 14 pilot symbols, and 2 pilots adopt DMRS Type1 and are located in pilot symbols 2 and 11. 2 pilot symbols on the uplink PUSCH of the current mobile station NR and channel estimation on each pilot symbol can be obtained from the base station side; or the channel estimation on each pilot symbol may be obtained according to a channel estimation algorithm in the obtained channel data, which is not described herein any more.
Another example is: the method comprises the steps that 1 pilot symbol is configured on an NR uplink PUSCH, the NR uplink PUSCH occupies 50 Resource blocks (RB for short), 14 pilot symbols are occupied, and 1 pilot adopts DMRS Type1 and is located on a pilot symbol 2. A single pilot frequency symbol on the uplink PUSCH of the current mobile station NR and channel estimation on the pilot frequency symbol can be obtained from the base station side; or the channel estimation on a single pilot symbol may be obtained according to the obtained channel data and a channel estimation algorithm, which is not described herein any more.
Step 202: and acquiring channel frequency domain characteristics and channel time domain characteristics according to channel estimation on the pilot frequency symbols.
Specifically, the channel frequency domain characteristics include: the normalized power of the frequency domain channel response at different receiving antennas at the mobile station can be used to characterize the fading variation of the channel environment in which the mobile station is located in the frequency domain. The channel time domain characteristics may include: the normalized power of time domain channel response on different receiving antennas is used for representing the multipath time delay spread information and the direct path information of the channel environment where the mobile station is located, and the direct path information is information indicating whether the direct path is dominant or not.
In one example, the channel frequency domain characteristics may be obtained as follows: acquiring a frequency domain channel estimation value on a pilot frequency symbol according to channel estimation on the pilot frequency symbol; acquiring the frequency domain channel estimation value power of each subcarrier on each receiving antenna on the pilot frequency symbol according to the frequency domain channel estimation value on the pilot frequency symbol; normalizing the obtained frequency domain channel estimation value power to generate frequency domain normalized power; and marking the frequency domain normalized power as the channel frequency domain characteristic.
Specifically, the frequency domain channel estimation value on each pilot symbol may be obtained, and the frequency domain channel estimation value power on each pilot symbol may be calculated according to the frequency domain channel estimation value
Where, a denotes a symbol of a receiving antenna, a is 0,1, … … N, N is an integer greater than 1, isc denotes a symbol of a subcarrier, isc is 0,1 … … 299, sym denotes a symbol of a pilot symbol, and sym is 0, 1.
The current base station comprises N antennae, and the normalization processing of the power value of the frequency domain channel estimation value can be carried out through each pilot frequency symbol on each antenna to obtain the frequency domain normalized power
As shown in equation (1).
Wherein, the first and the second end of the pipe are connected with each other,
for frequency domain normalized power, isc denotes the index of the subcarrier and a denotes the index of the receive antenna.
In this example, the normalized power of the frequency domain of each antenna on the pilot symbol 2 channel and the normalized power of the frequency domain of each antenna on the pilot symbol 11 will be obtained separately. And taking each obtained frequency domain normalized power as the characteristic of the channel characteristic on the frequency domain dimension.
In one example, a frequency domain channel estimation value on a pilot symbol is extracted according to channel estimation on the pilot symbol; performing inverse Fourier transform on the frequency domain channel estimation value on the pilot frequency symbol, and converting the frequency domain channel estimation value into a time domain channel estimation value; acquiring the time domain channel estimation value power of each subcarrier on a pilot frequency symbol on at least one receiving antenna according to the time domain channel estimation value; normalizing the obtained time domain channel estimation value power to generate time domain normalized power; and marking the time domain normalized power as a channel time domain characteristic.
Specifically, after obtaining the frequency domain channel estimation value on each pilot frequency symbol, the frequency domain estimation value is converted into a time domain estimation value through inverse fourier transform IFFT, and the time domain channel estimation value power on each pilot frequency symbol can be determined according to the time domain estimation value. For example, if the number of IFFT points is 1024, the power of the time domain channel estimation value obtained on the pilot symbol for each subcarrier is expressed as
Where a denotes a symbol of a receiving antenna, a is 0,1, … … N, N is an integer greater than 1, isc denotes a symbol of a subcarrier, isc is 0,1 … … 299,sym denotes a pilot symbol, and sym is 0, 1. In this example, the time domain channel estimation value powers of two CP lengths before and after the time domain theoretical peak are taken. If the time domain theoretical peak point position is i 511 and the CP length is 144, then the time domain channel estimation power when i 223,224.
Wherein, the first and the second end of the pipe are connected with each other,
denotes the time domain normalized power, isc denotes the index of the subcarrier, and a denotes the index of the receiving antenna. In this example, the time-domain normalized power of each antenna on the pilot symbol 2 channel and the time-domain normalized power of each antenna on the pilot symbol 11 would be obtained separately. And taking each obtained time domain normalized power as the characteristic of the channel characteristic on the time domain dimension.
It should be noted that there may be only one pilot symbol, and if there is only one pilot symbol, the above formula sym takes 0. And obtaining the frequency domain normalized power value and the time domain normalized power value of each receiving antenna on the pilot frequency symbol.
Step 203: it is determined whether the number of the acquired pilot symbols is 1, and if the number of the acquired pilot symbols is 1, step 204 is executed, otherwise, step 205 is executed.
Step 204: and acquiring the constellation diagram rotation angle of each data symbol after channel equalization as the time correlation characteristic of the channel.
Specifically, there are various ways to obtain the constellation rotation angle of each data symbol, and one way to obtain the constellation rotation angle of each data symbol is described below.
Specifically, the center of a constellation diagram cluster on each data symbol is calculated by using the soft information and the modulation mode after channel equalization, and each data symbol has a corresponding starA seating chart for performing the following processing for each data symbol: and calculating the central positions of four clusters in the constellation diagram on the data symbol, and obtaining the constellation diagram rotation angle on the data symbol according to the angle difference between the central positions and the ideal constellation diagram position. For example, there are 13 target user data symbols, and the data structure of the input neural network is as follows:
where sym denotes the index of the data symbol, rot denotes the angle of rotation, and est denotes the estimate, i.e.
Representing the estimated rotation angle on the data symbol sym.
For a single pilot frequency symbol, the frequency offset cannot be estimated using a time domain correlation method, in this example, the constellation rotation angle of each data symbol after channel equalization is obtained, frequency offset information can be reflected by the constellation rotation angle of the data symbol, and the constellation rotation angle of each data symbol can be fed back to the physical layer in real time to perform frequency offset compensation, so as to improve the receiving performance of the single pilot frequency symbol.
Step 205: and acquiring a normalized correlation value and a phase value of the channel response in time as a channel time correlation characteristic according to the channel estimation on at least two pilot symbols.
In one example, a channel estimation value on each pilot symbol is obtained according to channel estimation on at least two pilot symbols; obtaining a channel correlation value between channels on every two pilot symbols on at least one receiving antenna according to the channel estimation value on each pilot symbol; carrying out normalization processing on the channel correlation value to generate a normalized channel correlation value; and acquiring a phase value of the channel correlation value, and taking the phase value of the channel correlation value and the normalized channel correlation value as a channel time correlation characteristic.
Specifically, channel estimation values of each antenna on each pilot symbol are obtained
Where a denotes a symbol of a receiving antenna, a is 0,1, … … N, N is an integer greater than 1, isc denotes a symbol of a subcarrier, isc is 0,1 … … 299, sym denotes a symbol of a pilot symbol, and sym is 0, 1. Calculating channel estimation values on 2 pilot symbols
And
of (d) a channel correlation value gammacorrAnd normalizing the channel correlation values on different antennas to obtain normalized channel correlation values
Calculating the phase values of the channel correlation values on the two pilot symbols: thetacorr＝angle(γcorr). And taking the normalized channel correlation value and the phase value of the channel correlation value as the channel time correlation characteristic.
Step 206: and obtaining the channel characteristics according to the channel frequency domain characteristics, the channel time domain characteristics and the channel time correlation characteristics.
The obtained channel frequency domain features, channel time domain features, and channel time correlation features may be combined into a vector as channel features. Fig. 3 is a schematic structural diagram of channel characteristics in the case of 2 pilot symbols; fig. 4 is a structural diagram of channel characteristics in the case of a single pilot symbol.
It should be noted that the channel characteristics further include: and the spatial characteristics are used for characterizing the idle information of the mobile station, and comprise the change information of the beam arrival angle and/or the channel spatial correlation information. The obtained channel frequency domain features, channel time domain features, and channel time correlation features and spatial features may be combined into a vector, and the vector is used as the channel features. Fig. 5 is a schematic structural diagram of channel characteristics in the case of 2 pilot symbols, where the spatial characteristics in this example include information about changes in the angle of arrival of the beam.
The following procedure may be employed to obtain the channel spatial correlation information: acquiring a pilot frequency symbol of a specified channel in a channel scene to be identified and channel estimation on the pilot frequency symbol; obtaining a channel space correlation value between channels on every two receiving antennas under at least one pilot frequency symbol according to channel estimation on the same pilot frequency symbol on at least two receiving antennas; carrying out normalization processing on the channel space correlation value to generate a normalized channel space correlation value; and taking the normalized channel space correlation value as channel space correlation information.
Specifically, a spatial channel estimation value of each receiving antenna on the same pilot symbol is obtained, and a channel spatial correlation value between channels of each two receiving antennas on the same pilot symbol is calculated according to the spatial channel estimation value of each receiving antenna on the same pilot symbol, where a calculation manner of the channel spatial correlation value is substantially the same as a calculation manner of the channel correlation value in step 205, and is not described herein again. And obtaining the channel space correlation values under different pilot symbols to carry out normalization processing to obtain normalized channel space correlation values, and taking the normalized channel space correlation values as channel space correlation information.
Step 207: and inputting the channel characteristics into a preset channel scene recognition model, and acquiring a recognition result of the channel scene to be recognized.
Specifically, the channel scene recognition model may be trained in advance, and a feature data training set is collected, where the feature data training set includes channel features of channels in different channel sample scenes. The acquisition mode is as follows: the method comprises the steps that different channel sample scene models can be established through simulation to obtain channel data, or the channel data can be collected in different real channel environments, the collected channel data can be divided into a training set and a testing set, and the channel data can be processed; acquiring channel characteristics; the processing procedure of the channel data is substantially the same as the processing procedure of the channel data in the identification stage, i.e. the processing procedure from step 201 to step 206, and is not described herein again. Adding labels of channel scene categories to the channel characteristics in the test set, inputting each channel characteristic in the training set into a preset neural network for training to obtain appropriate model parameters, then testing the channel characteristics in the test set, and testing the model to obtain a test result; and continuously adjusting and optimizing the model parameters until the model convergence or the accuracy of the recognition result reaches a threshold value, and storing the trained model parameters.
The preset neural network can adopt a convolutional neural network as shown in fig. 6, and the normalized power of the frequency domain channel on different receiving antennas is subjected to feature extraction through a convolutional layer; and (3) performing feature extraction on time domain channel normalized power on different receiving antennas through two convolutional layers, inputting the time domain channel normalized power, a channel time correlation value and a phase value into a full link neural network together, and training the values together to obtain the channel scene recognition model.
Alternatively, a fully-linked neural network architecture such as that shown in FIG. 7 may be employed. The normalized power of the frequency domain channel on different receiving antennas, the normalized power of the time domain channel on different receiving antennas and the rotation angle of the constellation diagram on different data symbols form a one-dimensional vector, and the one-dimensional vector is input into a neural network for training and identification. The neural network model adopted in this example is a fully-linked four-layer neural network, including: an input layer, two hidden layers and an output layer adopt an optimizer of sgd, and an activation function is logistic. It will be appreciated that other configurations of neural networks may also be used.
In one example, the type of channel scenario includes any combination of the following types: the mobile station comprises a line-of-sight type, a non-line-of-sight type, a delay spread type obtained by dividing the delay spread degree based on a channel and a speed type obtained by dividing the moving speed of the mobile station.
The delay spread types include: the time delay expansion is small and the time delay expansion is large; the speed types include: low, medium and high speed. The channel scene type may be any combination of the above, for example, a line-of-sight type, a scene type with a small delay spread and at a low speed.
Inputting the acquired channel characteristics into the channel scene recognition model, outputting the matching probability of each channel sample scene, and determining the type of the current channel scene to be recognized according to the matching probability. After the scene type of the channel scene to be identified is determined, the receiving end may adopt a modulation mode matched with the type, or adopt a high-matching filter, or perform operations such as reasonable resource allocation or modulation system parameter configuration, so as to improve the performance and throughput of the whole wireless system.
It is worth mentioning that in the process of model training, the test set is labeled according to the channel scene type, so that the channel scene model obtained by training can also identify the time delay expansion type and the moving speed condition of the mobile station, and the accuracy and the applicability of the channel scene identification model identification result including the matching probability of the channel scene to be identified belonging to each channel sample scene are further improved.
The channel scene identification method in this embodiment can perform online identification in real time, is not only suitable for the case where multiple pilot symbols exist in one slot of an NR system, but also suitable for the case where a single pilot symbol exists in one slot, and solves the problem that doppler spreading and frequency offset estimation cannot be performed in the case of a single pilot.
A third embodiment of the present invention relates to a method for channel scene recognition, and this embodiment is a further improvement of the above embodiments, and the flow is shown in fig. 8.
And 301, acquiring channel characteristics of a channel under the scene of the channel to be identified.
Step 302: inputting the channel characteristics into a preset channel scene recognition model, and acquiring a recognition result of the channel scene to be recognized, wherein the recognition result comprises the matching probability of the channel scene to be recognized belonging to each channel sample scene.
Step 303: and judging whether all the matching probabilities are smaller than a preset threshold value, if so, executing the step 304, and otherwise, executing the step 307.
Specifically, the preset threshold may be set according to the actual situation, for example, the preset threshold is β ═ 0.2, if the matching probabilities are all smaller than the preset threshold, it indicates that the current channel scene to be identified does not belong to the channel sample scene in the feature data training set, step 304 may be executed, and the unrecognized prompt information may be output.
Step 304: and judging that the channel scene to be identified does not belong to the channel sample scene in the characteristic data training set. After this step is performed, step 305 is performed.
Step 305: and storing the channel data of the channel scene to be identified.
Step 306: and after the stored channel data reach a storage threshold value, adding the stored channel into a characteristic data training set for retraining the channel scene recognition model.
Specifically, a storage threshold may be set, and after the storage amount in the storage library reaches the storage threshold, all stored channel data is added to the feature data training set for retraining the channel scene recognition model.
Step 307: and determining that the channel scene to be identified belongs to the channel sample scene with the maximum matching probability.
Specifically, the channel scene type corresponding to the maximum matching probability may be selected as the type of the channel scene to be identified.
In the embodiment, in the channel scene recognition model recognition process, channel data of a new channel scene is stored, the channel scene recognition model is retrained to recognize the new channel scene, and samples in the characteristic data training set are continuously expanded through the stored channel data, so that the channel scene recognition model is continuously optimized to recognize more channel scenes.
A fourth embodiment of the present invention relates to a network device, a block diagram of which is shown in fig. 9, the network device including: at least one processor 401; and a memory 402 communicatively coupled to the at least one processor 401; the memory 402 stores instructions executable by the at least one processor 401, and the instructions are executed by the at least one processor 701 to enable the at least one processor 401 to perform the above-described channel scene recognition method.
The memory and the processor are connected by a bus, which may include any number of interconnected buses and bridges, linking together one or more of the various circuits of the processor and the memory. The bus may also link various other circuits such as peripherals, voltage regulators, power management circuits, and the like, which are well known in the art, and therefore, will not be described any further herein. A bus interface provides an interface between the bus and the transceiver. The transceiver may be one element or a plurality of elements, such as a plurality of receivers and transmitters, providing a means for communicating with various other apparatus over a transmission medium. The data processed by the processor is transmitted over a wireless medium via an antenna, which further receives the data and transmits the data to the processor.
The processor is responsible for managing the bus and general processing and may also provide various functions including timing, peripheral interfaces, voltage regulation, power management, and other control functions. And the memory may be used to store data used by the processor in performing operations.
A fifth embodiment of the present invention relates to a computer-readable storage medium, which stores a computer program, and the computer program, when executed by a processor, implements the method for channel scene recognition described above.
Those skilled in the art can understand that all or part of the steps in the method of the foregoing embodiments may be implemented by a program to instruct related hardware, where the program is stored in a storage medium and includes several instructions to enable a device (which may be a single chip, a chip, etc.) or a processor (processor) to execute all or part of the steps of the method described in the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk, and various media capable of storing program codes.
It will be understood by those of ordinary skill in the art that the foregoing embodiments are specific examples of practicing the invention, and that various changes in form and detail may be made therein without departing from the spirit and scope of the invention in practice.
Claims (12)
1. A method for channel scene recognition, comprising:
acquiring channel characteristics of a channel under a channel scene to be identified, wherein the channel characteristics comprise: the channel time domain characteristic is used for representing the multipath time delay expansion information and the direct path information of the channel scene to be identified, and the channel time correlation characteristic is used for representing the Doppler expansion information and the frequency offset information of the channel scene to be identified;
inputting the channel characteristics into a preset channel scene recognition model, and obtaining a recognition result of the channel scene to be recognized, wherein the channel scene recognition model is obtained after training a preset neural network based on a characteristic data training set, and the characteristic data training set comprises the channel characteristics of channels in at least 2 channel sample scenes.
2. The method of channel scene recognition according to claim 1, wherein the channel characteristics further comprise:
and the spatial characteristics are used for characterizing the idle information of the mobile station, and comprise the change information of the beam arrival angle and/or the channel spatial correlation information.
3. The method according to claim 1, wherein obtaining the recognition result of the channel scene to be recognized comprises:
obtaining the matching probability of the channel scene to be identified belonging to each channel sample scene;
judging whether each matching probability is smaller than a preset threshold value; if the channel sample scenes are smaller than the preset threshold value, judging that the channel scenes to be identified do not belong to the channel sample scenes in the characteristic data training set; otherwise, determining that the channel scene to be identified belongs to the channel sample scene with the maximum matching probability.
4. The method according to claim 3, wherein if it is determined that each of the matching probabilities is smaller than a preset threshold, the method further comprises:
storing the channel data of the channel scene to be identified;
and after the stored channel data reach a storage threshold value, adding the stored channel to the characteristic data training set for retraining the channel scene recognition model.
5. The method of any of claims 1 to 4, wherein the type of the channel scene comprises any combination of the following types: the mobile station comprises a line-of-sight type, a non-line-of-sight type, a delay spread type obtained by dividing the delay spread degree based on a channel and a speed type obtained by dividing the moving speed of the mobile station.
6. The method according to claim 1, wherein the obtaining the channel characteristics of the channel in the channel scene to be identified comprises:
acquiring a pilot frequency symbol of a specified channel in the channel scene to be identified and channel estimation on the pilot frequency symbol;
acquiring the channel frequency domain characteristics and the channel time domain characteristics according to the channel estimation on the pilot frequency symbols;
judging whether the number of the obtained pilot symbols is 1 or not, if so, obtaining the constellation diagram rotation angle of each data symbol after channel equalization as the channel time correlation characteristic; otherwise, acquiring a normalized correlation value and a phase value of channel response in time according to channel estimation on at least two pilot symbols as the channel time correlation characteristic;
and obtaining the channel characteristics according to the channel frequency domain characteristics, the channel time domain characteristics and the channel time correlation characteristics.
7. The method of claim 6, wherein obtaining the channel frequency domain characteristics according to the channel estimation on the pilot symbols comprises:
acquiring a frequency domain channel estimation value on the pilot frequency symbol according to the channel estimation on the pilot frequency symbol;
acquiring the frequency domain channel estimation value power of each subcarrier on each receiving antenna on the pilot symbols according to the frequency domain channel estimation value on the pilot symbols;
normalizing the obtained frequency domain channel estimation value power to generate frequency domain normalized power;
and marking the frequency domain normalized power as the channel frequency domain characteristic.
8. The method of claim 6, wherein obtaining the time domain channel characteristics according to the frequency domain channel estimation on the pilot symbols comprises:
extracting a frequency domain channel estimation value on the pilot frequency symbol according to the channel estimation on the pilot frequency symbol;
performing inverse Fourier transform on the frequency domain channel estimation value on the pilot frequency symbol, and converting the frequency domain channel estimation value into a time domain channel estimation value;
acquiring the time domain channel estimation value power of each subcarrier on at least one receiving antenna on the pilot frequency symbol according to the time domain channel estimation value;
normalizing the obtained time domain channel estimation value power to generate time domain normalized power;
and marking the time domain normalized power as the channel time domain characteristic.
9. The method according to claim 6, wherein said obtaining a normalized correlation value and a phase value of a channel response in time as the channel time correlation characteristic according to channel estimation on at least two pilot symbols comprises:
acquiring a channel estimation value on each pilot frequency symbol according to channel estimation on at least two pilot frequency symbols;
obtaining a channel correlation value between channels on every two pilot symbols on at least one receiving antenna according to the channel estimation value on each pilot symbol;
normalizing the channel correlation value to generate a normalized channel correlation value;
and acquiring a phase value of the channel correlation value, and taking the phase value of the channel correlation value and the normalized channel correlation value as the channel time correlation characteristic.
10. The method of channel scene recognition according to claim 2, further comprising:
acquiring a pilot frequency symbol of a specified channel in the channel scene to be identified and channel estimation on the pilot frequency symbol;
obtaining a channel space correlation value between channels on every two receiving antennas under at least one pilot frequency symbol according to channel estimation on the same pilot frequency symbol on at least two receiving antennas;
normalizing the channel space correlation value to generate a normalized channel space correlation value;
and taking the normalized channel space correlation value as the channel space correlation information.
11. A network device, comprising:
at least one processor, and,
a memory communicatively coupled to the at least one processor; wherein the content of the first and second substances,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of channel scene recognition according to any one of claims 1 to 10.
12. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the method of channel scene recognition according to any one of claims 1 to 10.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110088977.3A CN114785433A (en) | 2021-01-22 | 2021-01-22 | Channel scene recognition method, network device and storage medium |
| PCT/CN2021/140106 WO2022156469A1 (en) | 2021-01-22 | 2021-12-21 | Channel scene recognition method, network device, and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110088977.3A CN114785433A (en) | 2021-01-22 | 2021-01-22 | Channel scene recognition method, network device and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114785433A true CN114785433A (en) | 2022-07-22 |
Family
ID=82407656
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110088977.3A Pending CN114785433A (en) | 2021-01-22 | 2021-01-22 | Channel scene recognition method, network device and storage medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN114785433A (en) |
| WO (1) | WO2022156469A1 (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023051528A1 (en) * | 2021-09-30 | 2023-04-06 | 中兴通讯股份有限公司 | Channel scenario recognition method and apparatus, electronic device, and storage medium |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107734507A (en) * | 2016-08-12 | 2018-02-23 | 索尼公司 | Wireless scene identification device and method and Wireless Telecom Equipment and system |
| CN106548136A (en) * | 2016-10-19 | 2017-03-29 | 中科院成都信息技术股份有限公司 | A kind of wireless channel scene classification method |
| CN110113119A (en) * | 2019-04-26 | 2019-08-09 | 国家无线电监测中心 | A kind of Wireless Channel Modeling method based on intelligent algorithm |
| CN111835444B (en) * | 2020-06-17 | 2021-04-27 | 武汉大学 | Wireless channel scene identification method and system |
-
2021
- 2021-01-22 CN CN202110088977.3A patent/CN114785433A/en active Pending
- 2021-12-21 WO PCT/CN2021/140106 patent/WO2022156469A1/en active Application Filing
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023051528A1 (en) * | 2021-09-30 | 2023-04-06 | 中兴通讯股份有限公司 | Channel scenario recognition method and apparatus, electronic device, and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2022156469A1 (en) | 2022-07-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR100930799B1 (en) | Automated Clustering Method and Multipath Clustering Method and Apparatus in Mobile Communication Environment | |
| CN111869291A (en) | Channel mapping in a wireless system | |
| CN105790813B (en) | Code book selection method based on deep learning under a kind of extensive MIMO | |
| US20160366565A1 (en) | Grouping wireless devices in a communications network | |
| CN113595666B (en) | Channel prediction method and system of large-scale MIMO system based on time-frequency joint correlation | |
| CN115486035A (en) | Class of NN parameters for channel estimation | |
| CN112040400A (en) | Single-site indoor fingerprint positioning method based on MIMO-CSI, storage medium and equipment | |
| CN113037410A (en) | Channel identification method, device, transmission method, transmission equipment, base station and medium | |
| CN113114399B (en) | Three-dimensional spectrum situation complementing method and device based on generation countermeasure network | |
| CN112867021B (en) | Improved TrAdaBoost-based indoor positioning method for transfer learning | |
| Du et al. | SVM-assisted adaptive kernel power density clustering algorithm for millimeter wave channels | |
| CN114785433A (en) | Channel scene recognition method, network device and storage medium | |
| CN112564835B (en) | 5G wireless channel multipath clustering calculation method based on KNN and SVM algorithm | |
| CN112543409B (en) | Time of arrival (TOA) estimation method and base station | |
| Su et al. | Channel estimation in massive MIMO systems using a modified Bayes-GMM method | |
| Gölz et al. | Improving inference for spatial signals by contextual false discovery rates | |
| CN111612097B (en) | GRU network-based master user number estimation method | |
| CN114925720A (en) | Small sample modulation signal identification method based on space-time mixed feature extraction network | |
| WO2023005746A1 (en) | Channel modeling method and apparatus, storage medium, and electronic apparatus | |
| CN114978254B (en) | Machine learning auxiliary low-complexity robust precoding algorithm based on OFDM system | |
| WO2023071760A1 (en) | Beam domain division method and apparatus, storage medium, and electronic device | |
| CN115334523B (en) | Network coverage measuring and calculating method, device, equipment and computer readable storage medium | |
| WO2023116703A1 (en) | Precoding matrix obtaining method and apparatus, electronic device, and storage medium | |
| WO2024011741A1 (en) | Data-efficient updating for channel classification | |
| TWI777840B (en) | Number of repetitions prediction method and number of repetitions prediction device |
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 |