CN117278145A

CN117278145A - Wireless channel scene recognition and model training method and device

Info

Publication number: CN117278145A
Application number: CN202210672514.6A
Authority: CN
Inventors: 乔亚娟; 熊奕洋; 董石磊; 赵慧杰
Original assignee: China Telecom Corp Ltd
Current assignee: China Telecom Corp Ltd
Priority date: 2022-06-14
Filing date: 2022-06-14
Publication date: 2023-12-22

Abstract

The embodiment of the invention provides a wireless channel scene recognition and model training method and device, and relates to the technical field of data processing, wherein the method comprises the following steps: acquiring CIR data representing a communication condition of a target wireless channel; calculating a channel characteristic parameter of the target wireless channel based on the CIR data, wherein the channel characteristic parameter comprises at least one of the following information: PL, SF, PDP, RMS delay spread, AS, DPS, K factor; inputting the channel characteristic parameters into a pre-trained channel scene recognition model to obtain an output result; and determining a wireless channel scene to which the target wireless channel belongs based on the output result. The scheme provided by the embodiment of the invention can identify the wireless channel scene to which the wireless channel belongs.

Description

Wireless channel scene recognition and model training method and device

Technical Field

The present invention relates to the field of data processing technologies, and in particular, to a method and an apparatus for identifying a wireless channel scene.

Background

The channel environments of different wireless channel scenes can have different influences on the wireless channel, for example, the wireless channel scenes can comprise urban scenes, rural scenes, indoor scenes and the like, the channel environments of the urban scenes contain more buildings, and the buildings can influence the communication quality of the wireless channel as barriers; the channel environment of the rural scene contains fewer obstacles and has smaller influence on the communication quality of the wireless channel; the signal strength of the wireless signal in the channel environment of the indoor scene is often weak, and the communication quality of the wireless channel is poor.

In order to improve the communication performance of the wireless channel, the wireless channel can be configured differently according to different wireless channel scenes, so that the wireless channel is adapted to the channel environment of the wireless channel scene to which the wireless channel belongs. Therefore, it is first necessary to determine the radio channel scene to which the radio channel belongs, and for this purpose, it is necessary to provide a radio channel scene identification method for identifying the radio channel scene to which the radio channel belongs.

Disclosure of Invention

An object of an embodiment of the present invention is to provide a method for identifying a wireless channel scene, so as to identify a wireless channel scene to which a wireless channel belongs. The specific technical scheme is as follows:

in a first aspect, an embodiment of the present invention provides a method for identifying a wireless channel scene, where the method includes:

acquiring Channel Impulse Response (CIR) data representing a communication condition of a target wireless channel;

calculating a channel characteristic parameter of the target wireless channel based on the CIR data, wherein the channel characteristic parameter comprises at least one of the following information: path loss PL, shadow fading SF, power delay spectrum PDP, root mean square RMS delay spread, angle spread AS, doppler power spectrum DPS, K factor;

inputting the channel characteristic parameters into a pre-trained channel scene recognition model to obtain an output result;

And determining a wireless channel scene to which the target wireless channel belongs based on the output result.

In one embodiment of the present invention, before the inputting the channel characteristic parameter into the pre-trained wireless channel scene recognition model to obtain the output result, the method further includes:

carrying out data preprocessing on the channel characteristic parameters, wherein the data preprocessing comprises at least one of data cleaning processing, data dimension reduction processing and data normalization processing;

inputting the channel characteristic parameters into a pre-trained channel scene recognition model to obtain an output result, wherein the method comprises the following steps:

inputting the channel characteristic parameters subjected to data preprocessing into a pre-trained channel scene recognition model to obtain an output result.

In a second aspect, an embodiment of the present invention provides a model training method, where the method includes:

acquiring sample CIR data representing a communication condition of a sample wireless channel;

calculating a sample channel characteristic parameter of the sample wireless channel based on the sample CIR data, wherein the sample channel characteristic parameter comprises at least one of the following information: path loss PL, shadow fading SF, power delay spectrum PDP, root mean square RMS delay spread, angle spread AS, doppler power spectrum DPS, K factor;

Inputting the sample channel characteristic parameters into an initial channel scene recognition model to obtain an output result;

determining a wireless channel scene to which the sample wireless channel belongs based on the output result;

based on the determined wireless channel scene and the sample wireless channel scene, calculating the model loss of the initial channel scene identification model, wherein the sample wireless channel scene is: a predetermined wireless channel scene to which the sample wireless channel belongs;

and adjusting model parameters of the initial channel scene recognition model based on the model loss to obtain a channel scene recognition model.

In one embodiment of the present invention, the acquiring the sample CIR data representing the communication condition of the sample radio channel includes:

acquiring real CIR data representing the communication condition of a real wireless channel based on channel data of the real wireless channel measured in the field, and taking the real CIR data as sample CIR data;

and/or

Based on the channel data of the simulated wireless channel obtained by the channel simulation, the simulated CIR data representing the communication condition of the simulated wireless channel is obtained and used as sample CIR data.

In one embodiment of the present invention, when the sample channel feature parameter includes PDP, before the sample channel feature parameter is input into the initial channel scene recognition model to obtain an output result, the method further includes:

And removing sample channel characteristic parameters of sample wireless channels of which the PDP is larger than the preset PDP.

In one embodiment of the present invention, before the inputting the sample channel characteristic parameter into the initial channel scene recognition model, the method further includes:

performing data preprocessing on the sample channel characteristic parameters, wherein the data preprocessing comprises at least one of data cleaning processing, data dimension reduction processing and data normalization processing;

inputting the sample channel characteristic parameters into an initial channel scene recognition model to obtain an output result, wherein the method comprises the following steps:

and inputting the sample channel characteristic parameters subjected to data preprocessing into an initial channel scene recognition model to obtain an output result.

In one embodiment of the present invention, after the model parameters of the initial channel scene recognition model are adjusted based on the model loss, the method further includes:

acquiring test CIR data representing a communication condition of a test wireless channel;

calculating a test channel characteristic parameter of the test wireless channel based on the test CIR data, wherein the test channel characteristic parameter comprises at least one of the following information: PL, SF, PDP, RMS delay spread, AS, DPS, K factor;

Inputting the characteristic parameters of the test channel into the channel scene recognition model to obtain an output result;

determining a wireless channel scene to which the test wireless channel belongs based on the output result;

based on the determined wireless channel scene and a test wireless channel scene, calculating the accuracy of the channel scene identification model, wherein the test wireless channel scene is: and a predetermined wireless channel scene to which the test wireless channel belongs.

In a third aspect, an embodiment of the present invention provides a wireless channel scene identifying device, where the device includes:

the CIR acquisition module is used for acquiring CIR data representing the communication condition of the target wireless channel;

a characteristic parameter calculation module, configured to calculate a channel characteristic parameter of the target wireless channel based on the CIR data, where the channel characteristic parameter includes at least one of the following information: path loss PL, shadow fading SF, power delay spectrum PDP, root mean square RMS delay spread, angle spread AS, doppler power spectrum DPS, K factor;

the first result obtaining module is used for inputting the channel characteristic parameters into a pre-trained channel scene recognition model to obtain an output result;

And the first scene determining module is used for determining a wireless channel scene to which the target wireless channel belongs based on the output result.

In one embodiment of the invention, the apparatus further comprises:

the first preprocessing module is used for preprocessing the data of the channel characteristic parameters, wherein the data preprocessing comprises at least one of data cleaning processing, data dimension reduction processing and data normalization processing;

the first result obtaining module is specifically configured to:

In a fourth aspect, an embodiment of the present invention provides a model training apparatus, including:

the sample CIR acquisition module is used for acquiring sample CIR data representing the communication condition of a sample wireless channel;

a sample parameter calculation module, configured to calculate a sample channel feature parameter of the sample wireless channel based on the sample CIR data, where the sample channel feature parameter includes at least one of the following information: path loss PL, shadow fading SF, power delay spectrum PDP, root mean square RMS delay spread, angle spread AS, doppler power spectrum DPS, K factor;

The second result obtaining module is used for inputting the sample channel characteristic parameters into an initial channel scene recognition model to obtain an output result;

the second scene determining module is used for determining a wireless channel scene to which the sample wireless channel belongs based on the output result;

the loss calculation module is configured to calculate a model loss of the initial channel scene identification model based on the determined wireless channel scene and a sample wireless channel scene, where the sample wireless channel scene is: a predetermined wireless channel scene to which the sample wireless channel belongs;

and the parameter adjustment module is used for adjusting the model parameters of the initial channel scene recognition model based on the model loss to obtain the channel scene recognition model.

In one embodiment of the present invention, the sample CIR acquisition module is specifically configured to:

and/or

In one embodiment of the invention, the apparatus further comprises:

and the coefficient removing module is used for removing sample channel characteristic parameters of sample wireless channels of which the PDP is larger than a preset PDP.

In one embodiment of the invention, the apparatus further comprises:

the second preprocessing module is used for preprocessing the data of the sample channel characteristic parameters, wherein the data preprocessing comprises at least one of data cleaning processing, data dimension reduction processing and data normalization processing;

the second result obtaining module is specifically configured to:

In one embodiment of the invention, the apparatus further comprises:

the test parameter acquisition module is used for acquiring test CIR data representing the communication condition of a test wireless channel;

the test coefficient calculation module is used for calculating a test channel characteristic parameter of the test wireless channel based on the test CIR data, wherein the test channel characteristic parameter comprises at least one of the following information: PL, SF, PDP, RMS delay spread, AS, DPS, K factor;

the third result obtaining module is used for inputting the characteristic parameters of the test channel into the channel scene recognition model to obtain an output result;

A third scene determining module, configured to determine, based on the output result, a wireless channel scene to which the test wireless channel belongs;

the accuracy calculating module is used for calculating the accuracy of the channel scene identification model based on the determined wireless channel scene and the test wireless channel scene, wherein the test wireless channel scene is as follows: and a predetermined wireless channel scene to which the test wireless channel belongs.

In a fifth aspect, an embodiment of the present invention provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, where the processor, the communication interface, and the memory complete communication with each other through the communication bus;

a memory for storing a computer program;

a processor configured to implement the method steps of any one of the first or second aspects when executing a program stored on a memory.

In a sixth aspect, embodiments of the present invention provide a computer readable storage medium having a computer program stored therein, which when executed by a processor implements the method steps of any of the first or second aspects.

In a seventh aspect, embodiments of the present invention also provide a computer program product comprising instructions which, when run on a computer, cause the computer to perform the method steps of any of the first or second aspects described above.

The embodiment of the invention has the beneficial effects that:

in the embodiment of the invention, CIR data of a target wireless channel is firstly obtained, channel characteristic parameters of the target wireless channel are calculated based on the CIR data, the channel characteristic parameters are input into a channel scene identification model, and a wireless channel scene to which the target wireless channel belongs is determined based on an output result.

From the above, the channel characteristic parameters can represent the interfered condition of the target wireless channel, and the interference caused by the channel environments of different wireless channel scenes on the wireless channel is different, so that the identification of the wireless channel scene to which the target wireless channel belongs can be realized based on the channel characteristic parameters in the embodiment of the invention. And the channel scene recognition model is obtained by pre-training, and the trained channel scene recognition model can accurately mine the characteristics of the channel characteristic parameters and realize accurate recognition of the wireless channel scene based on the mined characteristics.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the embodiments or the description of the prior art will be briefly described, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings for a person having ordinary skill in the art.

Fig. 1 is a flow chart of a first method for identifying a wireless channel scene according to an embodiment of the present invention;

fig. 2 is a schematic structural diagram of a channel scene recognition model according to an embodiment of the present invention;

fig. 3 is a flow chart of a second method for identifying a wireless channel scene according to an embodiment of the present invention;

FIG. 4 is a flowchart of a first model training method according to an embodiment of the present invention;

FIG. 5 is a flowchart of a second model training method according to an embodiment of the present invention;

FIG. 6 is a flowchart of a third model training method according to an embodiment of the present invention;

FIG. 7 is a flowchart of a fourth model training method according to an embodiment of the present invention;

fig. 8 is a schematic diagram of a model training and wireless channel scene recognition flow provided in an embodiment of the present invention;

fig. 9 is a schematic structural diagram of a wireless channel scene recognition device according to an embodiment of the present invention;

FIG. 10 is a schematic structural diagram of a model training device according to an embodiment of the present invention;

fig. 11 is a schematic structural diagram of an electronic device according to an embodiment of the present invention;

fig. 12 is a schematic structural diagram of another electronic device according to an embodiment of the present invention.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, those of ordinary skill in the art will be able to devise all other embodiments that are obtained based on this application and are within the scope of the present invention.

In order to identify a wireless channel scene to which a wireless channel belongs, the embodiment of the invention provides a wireless channel scene identification and model training method and device.

The embodiment of the invention provides a wireless channel scene identification method, which comprises the following steps:

acquiring CIR (Channel Impulse Response ) data representing a communication condition of a target wireless channel;

calculating a channel characteristic parameter of the target wireless channel based on the CIR data, wherein the channel characteristic parameter comprises at least one of the following information: PL (Path Loss), SF (Shadow Fading), power delay profile PDP (Power Delay Profile, power Shi Yanpu), RMS (Root Mean Square) delay Spread, angle Spread AS (Angular Spread), DPS (Doppler Power Spectra, doppler power spectrum), K factor;

From the above, the channel characteristic parameters can represent the interfered condition of the target wireless channel, and the interference caused by the channel environments of different wireless channel scenes on the wireless channel is different, so that the distinction of the wireless channel scenes to which the target wireless channel belongs can be realized based on the channel characteristic parameters in the embodiment of the invention. And the channel scene recognition model is obtained by pre-training, and the trained channel scene recognition model can accurately mine the characteristics of the channel characteristic parameters and realize accurate recognition of the wireless channel scene based on the mined characteristics.

Referring to fig. 1, a flowchart of a first wireless channel scene recognition method according to an embodiment of the present invention is shown, where the method includes the following steps S101 to S104.

S101: CIR data representing a communication condition of a target wireless channel is acquired.

The CIR data may be obtained by processing channel data of the target wireless channel based on SAGE (Space-Alternating Generalized Expectation-on-Maximization) with a relatively high convergence rate or other algorithms in the prior art.

S102: and calculating channel characteristic parameters of the target wireless channel based on the CIR data.

Wherein the channel characteristic parameter includes at least one of the following information: PL, SF, PDP, RMS delay spread, AS, DPS, K factor.

Specifically, the PL, PDP, SF belongs to a large-scale channel characteristic parameter, the large-scale channel characteristic parameter can reflect interference which is received by a target wireless channel to a large extent and is easy to perceive, the RMS delay spread, the K factor, the AS and the DPS belong to a small-scale channel characteristic parameter, and the small-scale channel characteristic parameter can reflect interference which is received by the target wireless channel to a small extent and is difficult to perceive.

In one embodiment of the invention, after CIR data is obtained, information such as time delay of a target wireless channel, distance between a target wireless channel transmitting end and a receiving end, pitch angles of the target wireless channel transmitting end and the receiving end antenna, azimuth angles of the target wireless channel transmitting end and the receiving end antenna and the like can be obtained through calculation based on the CIR data. The pitch angle includes EOA (Elevation of Arrival, vertical angle of arrival), EOD (Elevation of Departure, vertical angle of emission), and the azimuth angle includes: AOA (Azimuth of Arrival, horizontal angle of arrival), AOD (Azimuth of Departure, horizontal emission angle). The channel characteristic parameters can be obtained by processing the calculated information.

For a description of the specific meaning and calculation manner of each information that may be contained in the channel characteristic parameters, reference may be made to (one) - (seven) shown below, and details thereof will not be given here.

S103: and inputting the channel characteristic parameters into a pre-trained channel scene recognition model to obtain an output result.

The channel scene recognition model may be obtained by training an initial channel scene recognition model based on the sample channel characteristic parameters, and the training process of the initial channel scene recognition model may be referred to as an embodiment shown in fig. 4 below, which is not described in detail herein.

The initial channel scene recognition model may be a neural network model for classification, and may select a wireless channel scene to which the target wireless channel belongs from preset wireless channel scenes.

Specifically, the channel scene recognition model may be a BPNN (Back Propagation Neural Network, back propagation neural network model), or other models in the prior art, which is not limited in this embodiment.

In addition, the preset wireless channel scenario may include: NLOS (Non Line of Sight ) scenes, LOS (Line of Sight) scenes, urban scenes, suburban scenes, rural scenes, indoor non-factory scenes, indoor factory scenes, tunnel scenes and other static scenes, and dynamic scenes such as vehicles, trains and the like in the process of traveling can be included.

Referring to fig. 2, a schematic structural diagram of a channel scene recognition model according to an embodiment of the present invention is provided.

The channel scene recognition model is BPNN, and the BPNN model consists of an input layer, a hidden layer and an output layer.

The input layer comprises n input layer nodes, each input layer node is used for receiving one item of information in the channel characteristic parameters, and the number n of the input layer nodes is the same as the number of the channel characteristic parameters.

The output layer comprises m output layer nodes, each output layer node corresponds to a preset wireless channel scene and is used for outputting the probability that a target wireless channel belongs to the wireless channel scene corresponding to the output layer node, the probabilities output by the output layer nodes jointly form an output result of the channel scene recognition model, and the sum of the probabilities contained in the output result is 1. The number of the output layer nodes is the same as the number of the preset wireless channel scenes.

For example, if the preset wireless channel scene includes a city scene, a rural scene, a suburban scene, and an indoor scene, the number of output layer nodes is 4, and the output result of the channel scene identification model includes a probability that the target wireless channel belongs to the city scene, a probability that the target wireless channel belongs to the rural scene, a probability that the target wireless channel belongs to the subsuburban scene, and a probability that the target wireless channel belongs to the indoor scene.

The hidden layer contains j hidden layer nodes.

The arrow between the output layer node and the hidden layer node indicates that each input layer node respectively inputs the information received by the input layer node into each hidden layer node, and the arrow between the hidden layer node and the output layer node indicates that each hidden layer node processes the received information and then inputs the processing result into each output layer node.

In addition, the hidden layer node t can perform data processing according to the following formula to obtain an output result of the hidden layer node t:

wherein, the H is _t In order to hide the output result of layer node t, n is the number of nodes in the input layer, i is the number of the input layer nodes,for a first weight, X, between an input layer node i and a hidden layer node t _n Identifying channel characteristic parameters of a model for an input channel scene, a _t Is the first threshold of the hidden layer node t. f () is the activation function used by the hidden layer node t.

In particular, the method comprises the steps of,and a _t For the model parameters included in the channel scene recognition model, the activation function may be a Sigmoid function or other activation functions, which is not limited in this embodiment.

In addition, the output layer node k may perform data processing according to the following formula, to obtain an output result of the output layer node k:

Wherein Y is _k Output result for output layer node kT is the number of hidden layer nodes, j is the total number of hidden layer nodes,for a second weight between the output layer node k and the hidden layer node t, H _t B, as an output result of the hidden layer node t _k Is the second threshold of the output layer node k.

In particular, the method comprises the steps of,and b _k Model parameters contained in the model are identified for the channel scene.

Therefore, if the channel scene recognition model is BPNN, the number of nodes in the input layer can be adjusted based on the number of channel characteristic parameters, and the number of nodes in the output layer can be adjusted based on the number of preset wireless channel scenes, so that the more the number of preset wireless channel scenes is, the more the wireless channel scenes can be recognized by the channel scene recognition model, the structure of BPNN can be adjusted based on the actual application scene, and the embodiment of the invention can be matched with different application scenes.

S104: and determining a wireless channel scene to which the target wireless channel belongs based on the output result.

Specifically, if the output result includes probabilities that the target wireless channel belongs to each preset wireless channel scene, the wireless channel scene with the highest probability may be taken as the wireless channel scene to which the target wireless channel belongs.

In addition, the output result may be an identifier of a radio channel scenario to which the target radio channel belongs, and the radio channel scenario to which the target radio channel belongs may be directly determined based on the identifier.

Further, each item of information that may be included in the fading parameters is specifically described by the following (one) - (seven).

(one) calculate the path loss PL according to the following formula:

wherein, P is as described above _dB Path loss for target wireless channel, d ₀ For a preset reference distance, d is the actual distance between the transmitting end and the receiving end of the target wireless channel, gamma is the preset path loss index, PL () represents the calculated path loss, PL (d ₀ ) Is free path loss without any system loss.

Specifically, PL (d) can be calculated according to the following formula ₀ )：

Wherein G is _t For the transmitting antenna gain of the target wireless channel transmitting end, G _r Gain of receiving antenna of receiving end of target wireless channel, lambda is wavelength, d ₀ Is a preset reference distance.

And (II) in the propagation process of the wireless signal, shadow fading SF can be generated when the average signal power meets an obstacle, and the specific formula is as follows:

wherein P is _dB For calculating the shadow fading SF, shadow fading X _δ Is a random variable in a Gaussian distribution with 0 as the mean and delta as the standard deviation, namely X _δ ～N(0,δ ² ) The meaning of other parameters included in the formula is the same as that of the formula PL, and this embodiment will not be repeated.

Specifically, in theory, the more obstacles contained in the channel environment of the wireless channel scene, the larger the shadow fading SF.

The (III) power delay spectrum PDP represents the desire to receive signal power at a certain delay, and can be calculated according to the following formula:

P(t,τ)＝E(|h(t,τ)| ² )

where P (t, τ) is the time t, h (t, τ) is the PDP in the case where the delay is τ, and h (t, τ) is the CIR of the target radio channel in the case where the delay is τ. The absolute value is represented by the absolute value, and the mathematical expectation is calculated by the E ().

(IV) Root Mean Square (RMS) delay spread can represent the dispersion condition of a target wireless channel from the angle of a delay domain, and the RMS delay spread can be calculated according to the following formula:

wherein sigma _τ For the RMS delay spread of the target radio channel, l is the number of the paths contained in the target radio channel, τ _l And P (τ) _l ) Respectively representing the time delay and the power of the first path in the target wireless channel.

The (fifth) angle spread AS is a parameter describing the angular level dispersion of the target wireless channel, which is the square root of the second order center distance of the angular power spectrum (Power Azimuth Spectral, PAS), and is specifically expressed AS follows:

where θ represents the angle of the target wireless channel,representing the average angle, P (θ) represents the power received at angle θ.

The Doppler power spectrum DPS is used to represent the frequency dispersion degree of the target wireless channel, and the Doppler power spectrum can be used to calculate the Doppler spread, and the specific formula is as follows:

S(f)＝|fft(h(t))| ²

wherein B is _D For the DPS of the target wireless channel,representing the average Doppler shift, S (f) represents the Doppler power spectrum, FFT transformation, and h (t) is the CIR at time t.

The K factor is the ratio of the power of the master MPC to the power of the reflected MPC in the target wireless channel, and the specific formula is as follows:

Wherein K is the K factor, P of the target wireless channel ₀ The power of the main path is represented, l represents the number of the path, P _l Is the power of the first path.

Referring to fig. 3, a flow chart of a second method for identifying a wireless channel scene according to an embodiment of the present invention, compared with the embodiment shown in fig. 1, the method further includes the following step S105 before the step S103, and in addition, the step S103 may be implemented by the following step S103A.

S105: and carrying out data preprocessing on the channel characteristic parameters.

The data preprocessing comprises at least one of data cleaning processing, data dimension reduction processing and data normalization processing.

Specifically, abnormal data can be removed by performing data cleaning processing on the channel characteristic parameters, and in the embodiment of the invention, data with an absolute value of a difference value larger than a preset difference value from an average value of historical data can be used as the abnormal data, and the abnormal data in the channel characteristic parameters can be removed.

In addition, the dimension reduction processing of the channel characteristic parameters can reduce the dimension and the data volume of the channel characteristic parameters, so that the calculation complexity of the data processing of the channel characteristic parameters is reduced. In one embodiment of the present invention, the data dimension reduction process may be performed on the channel characteristic parameters based on PCA (Principal Components Analysis, principal component analysis), where the core idea of PCA is to convert the channel characteristic parameters into a set of values of linear uncorrelated principal components through orthogonal transformation, and PCA is capable of projecting the multidimensional channel characteristic parameters onto a preserving characteristic direction with maximum variance. In this embodiment, the channel characteristic parameter may be subjected to the dimension reduction processing by other methods, which is not limited in this embodiment.

Specifically, the process of performing dimension reduction processing on the channel characteristic parameters by adopting PCA can be realized according to the following formula:

K(x _m ,x _n )＝exp(-γ||x _m ,x _n || ² )

wherein K (x _m ,x _n ) In order to adopt PCA to carry out the result that the dimension reduction process is carried out on the channel characteristic parameters, exp () represents an exponential function based on e, gamma is a preset super parameter, and x _m And x _n Representing the mth kind contained in the channel characteristic parametersAnd the nth information.

The above formula is a schematic formula for performing dimension reduction processing on the channel characteristic parameters by adopting PCA, and in practice, the channel characteristic parameters may not only contain x _m And x _n Two kinds of information.

Furthermore, the channel characteristic parameters can be unified into a preset interval by carrying out data normalization processing on the channel characteristic parameters, and the data with values in the unified preset interval can be processed, so that the data processing speed can be improved. Specifically, the data normalization processing may be performed on the channel characteristic parameter based on a z-score (z-score) normalization method, or may be performed by other methods in the prior art, which is not limited in this embodiment.

Specifically, the process of normalizing the channel characteristic parameters in the z-score normalization manner may be implemented according to the following formula:

Wherein j is the dimension number of the channel characteristic parameter, each item of information contained in the channel characteristic parameter corresponds to one dimension, i is the number of data contained in each item of dimension information,for the ith piece of data in the jth dimension information contained in the pre-processing channel characteristic parameters, is->For the ith piece of data in the jth dimension information contained in the normalized processed channel characteristic parameters, a>Mean value of j-th dimension information, sigma _xj Is the standard deviation of the j-th dimension information.

S103A: inputting the channel characteristic parameters subjected to data preprocessing into a pre-trained channel scene recognition model to obtain an output result.

Specifically, step S103A is similar to step S103 described above, and the difference is only that the data in the input channel scene recognition model is different, which is not described in detail in this embodiment.

As can be seen from the above, the embodiment of the present invention can remove abnormal data in the channel characteristic parameters by performing data preprocessing on the channel characteristic parameters, reduce the dimension of the channel characteristic parameters, and unify the channel characteristic parameters into a preset interval, so that the data processing speed can be increased in the subsequent process of processing the channel characteristic parameters by using the channel scene recognition model.

Referring to fig. 4, a flowchart of a first model training method according to an embodiment of the present invention is shown, where the method includes the following steps S401 to S406.

S401: sample CIR data representing a communication condition of a sample radio channel is acquired.

In one embodiment of the present invention, the above step S401 may be implemented by the following step a and/or step B.

Step A: based on the channel data of the real wireless channel measured in the field, real CIR data representing the communication condition of the real wireless channel is acquired as sample CIR data.

In one embodiment of the invention, channel data of real wireless channels belonging to different wireless communication scenes can be measured in the field, the channel data are processed to obtain real CIR data, and the wireless communication scene acquired to obtain the channel data is determined as a sample wireless communication scene actually corresponding to the real CIR data.

Specifically, when channel data is acquired, time domain data and frequency domain data of a real wireless channel can be acquired from time domain dimension and frequency domain dimension respectively, and as the channel data, the channel data can be processed based on SAGE or other algorithms in the prior art to obtain real CIR data.

And (B) step (B): based on the channel data of the simulated wireless channel obtained by the channel simulation, simulated CIR data representing the communication condition of the simulated wireless channel is obtained as sample CIR data.

In one embodiment of the invention, the communication conditions of the wireless channel in different simulation wireless communication scenes can be simulated through a simulation tool, so that the channel data of the simulation wireless channel can be obtained. For example, the communication situation of the wireless channel in the urban scene, the communication situation in the rural scene, and the like can be simulated by a simulation tool. And processing the channel data to obtain simulated CIR data, and determining a simulated wireless communication scene where the simulated wireless channel is located as a sample wireless communication scene actually corresponding to the simulated CIR data.

The simulation tool may be a QuaDriGa platform, a vienna platform, a WINNER model, a 3GPP model, or the like, or may be another simulation tool in the prior art, which is not limited in this embodiment.

Specifically, the channel data may be processed based on SAGE or other algorithms in the prior art to obtain simulated CIR data.

From the above, the sample CIR data used in the embodiment of the present invention may be calculated based on the actually collected channel data, or may be calculated based on the channel data obtained by simulation, and by combining the actually collected channel data and the channel data obtained by simulation, a larger amount of sample CIR data may be obtained, and further, the initial channel recognition model may be trained based on a larger amount of sample CIR data, so that the accuracy of the channel recognition model obtained by training may be improved.

S402: and calculating sample channel characteristic parameters of the sample wireless channel based on the sample CIR data.

Wherein the sample channel characteristic parameter comprises at least one of the following information: PL, SF, PDP, RMS delay spread, AS, DPS, K factor.

Specifically, each sample channel characteristic parameter and the wireless channel scene corresponding to each sample channel characteristic parameter can be recorded in the form of training data pairs.

Y _e ＝{ξ _e |l _e }

Specifically, e denotes the number of the training data pair, Y _e Training number for group eData pair, xi _e For the e-th sample channel characteristic parameter, l _e And a wireless channel scene label for representing the wireless channel scene corresponding to the e-th sample channel characteristic parameter.

S403: and inputting the sample channel characteristic parameters into an initial channel scene recognition model to obtain an output result.

Before executing step S403, the model parameters in the initial channel scene recognition model may be initialized, and then the sample fading data may be input into the initial channel scene recognition model.

S404: and determining a wireless channel scene to which the sample wireless channel belongs based on the output result.

Specifically, the steps S401 to S404 are similar to the steps S101 to S104, and the description thereof is omitted in this embodiment.

S405: based on the determined wireless channel scene and the sample wireless channel scene, a model loss of the initial channel scene recognition model is calculated.

The sample wireless channel scene is as follows: a predetermined radio channel scenario to which the sample radio channel belongs.

Specifically, the more the wireless channel scene determined based on the initial channel scene recognition model is theoretically close to the predetermined and real sample wireless channel scene, the smaller the model loss of the initial channel scene recognition model is, the more accurate the output result of the initial channel scene recognition model is, and the main objective of the model training process is to improve the data result of the initial channel scene model and reduce the model loss.

Wherein, the model loss can be expressed as:

e＝O-Y

specifically, e is model loss, O is a label of a sample wireless channel scene, and Y is a determined wireless channel scene.

S406: and adjusting model parameters of the initial channel scene recognition model based on the model loss to obtain the channel scene recognition model.

Specifically, if the calculated model loss is not less than the preset model loss, the accuracy of the current initial channel scene recognition model does not reach the expected accuracy, steps S401-S406 may be re-executed, training the initial channel scene recognition model is continued until the model loss is less than the preset model loss, and the initial channel scene recognition model after the model parameters are adjusted is determined as the channel scene recognition model.

Or if the number of times of adjustment of the model parameters of the initial channel scene recognition model does not reach the preset number of times, re-executing the steps S401-S406, continuing to train the initial channel scene recognition model until the number of times of adjustment of the model parameters of the initial channel scene recognition model reaches the preset number of times, and determining the initial channel scene recognition model after the model parameters are adjusted as the channel scene recognition model.

In the case where the initial channel scene recognition model is BPNN, the first weight included in the model parameter may be adjusted according to the following formula:

wherein, the aboveFor the first weight between the adjusted input layer node i and the hidden layer node t +.>In order to adjust the first weight between the input layer node i and the hidden layer node t before adjustment, η is a preset learning rate, and H is as described above _t To hide the output result of layer node t, X _n Identifying channel characteristic parameters of a model for an input channel scene, m being the number of output layer nodes,for the second weight between the output layer node k and the hidden layer node t, e is the model loss.

In addition, the second weight included in the model parameters may be adjusted according to the following formula:

Wherein,for the second weight between the adjusted output layer node k and the hidden layer node t +.>Before adjustment, the second weight between the output layer node k and the hidden layer node t is the preset learning rate, and eta is the above H _t And e is the model loss, which is the output result of the hidden layer node t.

Further, the first threshold included in the model parameters may be adjusted according to the following formula:

/>

wherein a is _t ' is the first threshold, a, of the adjusted hidden layer node t _t For the first threshold value of the hidden layer node t, η is a preset learning rate, and H is the above _t To hide the output result of layer node t, X _n Identifying channel characteristic parameters of a model for an input channel scene, m being the number of output layer nodes,for the second weight between the output layer node k and the hidden layer node t, e is the model loss.

Further, the second threshold included in the model parameters may be adjusted according to the following formula:

b′ _k ＝b _k +e

wherein b' _k To adjust the second threshold value of the post-output layer node k, b _k To adjust the front output layer node kE is the model penalty described above.

Specifically, the larger the learning rate, the larger the adjustment range of the model parameters is each time the model parameters are adjusted. For example, the learning rate may be set to 0.05.

As can be seen from the above, in the embodiment of the present invention, the initial channel scene recognition model may be trained based on the sample CIR data, so that the wireless channel scene determined based on the output result of the channel scene recognition model obtained by training is similar to the real sample wireless channel scene, that is, the accuracy of the output result of the channel scene recognition model obtained by training is higher. Accurate wireless channel scene recognition can be realized based on the channel scene recognition model with higher accuracy.

Referring to fig. 5, a flow chart of a second model training method according to an embodiment of the present invention, in comparison with the embodiment shown in fig. 4, in the case that the sample channel characteristic parameter includes PDP, the following step S407 is further included before the step S403.

S407: and removing sample channel characteristic parameters of sample wireless channels of which the PDP is larger than the preset PDP.

Specifically, under normal conditions, considering that the interference degree of the channel environment of the wireless channel scene to which the sample wireless channel belongs to the sample wireless channel is generally in a certain range, and the transmission power of the sample wireless channel is generally in a certain range, the PDP of the sample wireless channel is not always oversized, and the sample channel characteristic parameters of the sample wireless channel with oversized PDP may be abnormal, so that the sample channel characteristic parameters of the sample wireless channel can be removed.

For example, the preset PDP may be 2000ns, 2500ns, 3000ns, etc.

From the above, the embodiment of the invention removes the sample channel characteristic parameters of the sample wireless channel with the PDP larger than the preset PDP, so as to remove the sample channel characteristic parameters of the sample wireless channel with abnormal data, that is, the embodiment of the invention does not use the abnormal sample channel characteristic parameters to train the initial channel scene recognition model, and can improve the accuracy of the channel scene recognition model obtained by training. In addition, the abnormal sample channel characteristic parameters can be removed, so that the data volume of the sample channel characteristic parameters can be reduced, and the training speed of the initial channel scene recognition model can be improved.

Referring to fig. 6, a flow chart of a third model training method according to an embodiment of the present invention, compared with the embodiment shown in fig. 4, the following step S408 is further included before the above step S403, and the above step S403 may be implemented by the following step S403A.

S408: and carrying out data preprocessing on the sample channel characteristic parameters.

Specifically, the step S408 is similar to the step S105, and only the difference is that the data to be subjected to the data preprocessing is different, which is not described in detail in this embodiment.

S403A: and inputting the sample channel characteristic parameters subjected to data preprocessing into an initial channel scene recognition model to obtain an output result.

Specifically, the step S403A is similar to the step S103A, and the description of this embodiment is omitted.

As can be seen from the above, the embodiment of the invention can remove abnormal data in the sample channel characteristic parameters by preprocessing the sample channel characteristic parameters, reduce the dimension of the sample channel characteristic parameters, unify the sample channel characteristic parameters into the preset interval, and thereby improve the model training speed in the subsequent training process of the initial channel scene recognition model by using the sample channel characteristic parameters.

Referring to fig. 7, a flow chart of a fourth model training method according to an embodiment of the present invention, compared with the embodiment shown in fig. 4, further includes the following steps S409-S413 after the step S406.

S409: test CIR data representing a communication condition of a test wireless channel is acquired.

S410: and calculating the characteristic parameters of the test channel of the test wireless channel based on the test CIR data.

Wherein the test channel characteristic parameter includes at least one of the following information: PL, SF, PDP, RMS delay spread, AS, DPS, K factor.

Specifically, the steps S409-S410 are similar to the steps S401-S402, the obtained test CIR data is similar to the sample CIR data, the test channel characteristic parameters are similar to the sample channel characteristic parameters, and only the sample CIR data and the sample channel characteristic parameters are used for training the initial channel scene recognition model, and the test CIR data and the test channel characteristic parameters are used for analyzing the accuracy of the channel scene recognition model obtained by training. In this embodiment, the CIR data of a radio channel environment to which a certain data amount is known may be obtained, and a part of the obtained CIR data is used as sample CIR data, and the other part is used as test CIR data, which will not be described in detail in this embodiment.

S411: and inputting the characteristic parameters of the test channel into the channel scene recognition model to obtain an output result.

S412: and determining a wireless channel scene to which the test wireless channel belongs based on the output result.

Specifically, the steps S411-S412 are similar to the steps S103-S104, and the description thereof is omitted in this embodiment.

S413: and calculating the accuracy of the channel scene recognition model based on the determined wireless channel scene and the test wireless channel scene.

The test wireless channel scene is as follows: and a predetermined wireless channel scene to which the test wireless channel belongs.

In one embodiment of the present invention, the accuracy of the channel scene recognition model may be calculated based on the confusion matrix and the ROC curve, or the accuracy of the channel scene recognition model may be calculated based on other manners, which will not be described in detail in this embodiment.

In addition, the accuracy obtained by calculation can be compared with the preset accuracy, and when the accuracy reaches the preset accuracy, the trained channel scene recognition model is used to realize the embodiment shown in fig. 1, so as to ensure the accuracy of the wireless channel scene recognition result. If the accuracy does not reach the preset accuracy, training the channel scene recognition model can be continued.

From the above, the embodiment of the invention can also test the accuracy of the trained channel scene recognition model by adopting the test CIR data after training the initial channel scene recognition model to obtain the trained channel scene recognition model, and can judge whether the channel scene recognition model needs to be trained continuously or not based on the determined accuracy so as to ensure the accuracy of the channel scene recognition model.

Referring to fig. 8, a schematic flow chart of model training and wireless channel scene recognition is provided in an embodiment of the present invention.

AS can be seen from the figure, firstly, in the model training stage, a wireless channel scene tag corresponding to sample CIR data is obtained AS training data, the sample CIR data is processed to obtain sample channel characteristic parameters including PL, SF, PDP, RMS delay spread, AS, DPS and K factors, the sample channel characteristic parameters are input into an initial channel scene recognition model, specifically, the structure of the initial channel scene recognition model is the same AS that shown in fig. 2, and the embodiment is omitted for brevity, and the trained channel scene recognition model is obtained after training the initial channel scene recognition model.

And then, in a wireless channel scene recognition stage, CIR data of a target wireless channel are obtained, the CIR data are processed to obtain channel characteristic parameters comprising PL, SF, PDP, RMS time delay expansion, AS, DPS and K factors, and the channel characteristic parameters are input into a trained channel scene recognition model to obtain a wireless channel scene corresponding to the target wireless channel.

Corresponding to the wireless channel scene recognition method, the embodiment of the invention also provides a wireless channel scene recognition device.

Referring to fig. 9, a schematic structural diagram of a wireless channel scene recognition device according to an embodiment of the present invention is provided, where the device includes:

a CIR acquisition module 901, configured to acquire CIR data representing a communication condition of a target wireless channel;

a characteristic parameter calculating module 902, configured to calculate a channel characteristic parameter of the target wireless channel based on the CIR data, where the channel characteristic parameter includes at least one of the following information: path loss PL, shadow fading SF, power delay spectrum PDP, root mean square RMS delay spread, angle spread AS, doppler power spectrum DPS, K factor;

a first result obtaining module 903, configured to input the channel characteristic parameter into a pre-trained channel scene recognition model, to obtain an output result;

a first scenario determining module 904, configured to determine, based on the output result, a wireless channel scenario to which the target wireless channel belongs.

In one embodiment of the invention, the apparatus further comprises:

the first result obtaining module 903 is specifically configured to:

Referring to fig. 10, a schematic structural diagram of a model training device according to an embodiment of the present invention is provided, where the device includes:

a sample CIR acquisition module 1001, configured to acquire sample CIR data representing a communication condition of a sample wireless channel;

a sample parameter calculation module 1002, configured to calculate a sample channel feature parameter of the sample wireless channel based on the sample CIR data, where the sample channel feature parameter includes at least one of the following information: path loss PL, shadow fading SF, power delay spectrum PDP, root mean square RMS delay spread, angle spread AS, doppler power spectrum DPS, K factor;

A second result obtaining module 1003, configured to input the sample channel characteristic parameter into an initial channel scene identification model, to obtain an output result;

a second scene determining module 1004, configured to determine, based on the output result, a wireless channel scene to which the sample wireless channel belongs;

a loss calculation module 1005, configured to calculate a model loss of the initial channel scene recognition model based on the determined wireless channel scene and a sample wireless channel scene, where the sample wireless channel scene is: a predetermined wireless channel scene to which the sample wireless channel belongs;

and a parameter adjustment module 1006, configured to adjust model parameters of the initial channel scene recognition model based on the model loss, so as to obtain a channel scene recognition model.

In one embodiment of the present invention, the sample CIR obtaining module 1001 is specifically configured to:

and/or

In one embodiment of the invention, the apparatus further comprises:

the second result obtaining module 1003 is specifically configured to:

In one embodiment of the invention, the apparatus further comprises:

The embodiment of the present invention further provides an electronic device, as shown in fig. 11, including a processor 1101, a communication interface 1102, a memory 1103 and a communication bus 1104, where the processor 1101, the communication interface 1102 and the memory 1103 complete communication with each other through the communication bus 1104,

A memory 1103 for storing a computer program;

the processor 1101 is configured to implement the above-mentioned wireless channel scene recognition method when executing the program stored in the memory 1103.

When the electronic equipment provided by the embodiment of the invention is used for identifying the wireless channel scene, the channel characteristic parameters can represent the interfered condition of the target wireless channel, and the interference caused by the channel environments of different wireless channel scenes on the wireless channel is different, so that the identification of the wireless channel scene to which the target wireless channel belongs can be realized based on the channel characteristic parameters in the embodiment of the invention. And the channel scene recognition model is obtained by pre-training, and the trained channel scene recognition model can accurately mine the characteristics of the channel characteristic parameters and realize accurate recognition of the wireless channel scene based on the mined characteristics.

The embodiment of the present invention also provides another electronic device, as shown in fig. 12, including a processor 1201, a communication interface 1202, a memory 1203, and a communication bus 1204, where the processor 1201, the communication interface 1202, and the memory 1203 perform communication with each other through the communication bus 1204,

a memory 1203 for storing a computer program;

The processor 1201 is configured to implement the above-described model training method when executing the program stored in the memory 1203.

When the electronic equipment provided by the embodiment of the invention is used for model training, the initial channel scene recognition model can be trained based on the sample CIR data, so that the wireless channel scene determined based on the output result of the channel scene recognition model obtained through training is similar to the real sample wireless channel scene, namely the accuracy of the output result of the channel scene recognition model obtained through training is higher. Accurate wireless channel scene recognition can be realized based on the channel scene recognition model with higher accuracy.

The communication bus mentioned above for the electronic devices may be a peripheral component interconnect standard (Peripheral Component Interconnect, PCI) bus or an extended industry standard architecture (Extended Industry Standard Architecture, EISA) bus, etc. The communication bus may be classified as an address bus, a data bus, a control bus, or the like. For ease of illustration, the figures are shown with only one bold line, but not with only one bus or one type of bus.

The communication interface is used for communication between the electronic device and other devices.

The Memory may include random access Memory (Random Access Memory, RAM) or may include Non-Volatile Memory (NVM), such as at least one disk Memory. Optionally, the memory may also be at least one memory device located remotely from the aforementioned processor.

The processor may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU), a network processor (Network Processor, NP), etc.; but also digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), field programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

In yet another embodiment of the present invention, there is also provided a computer readable storage medium having stored therein a computer program which, when executed by a processor, implements the steps of any of the wireless channel scene recognition methods described above.

When the computer program stored in the computer readable storage medium provided by the embodiment of the invention is executed to identify the wireless channel scene, the channel characteristic parameters can represent the interfered condition of the target wireless channel, and the interference caused by the channel environments of different wireless channel scenes to the wireless channel is different, so that the identification of the wireless channel scene to which the target wireless channel belongs can be realized based on the channel characteristic parameters. And the channel scene recognition model is obtained by pre-training, and the trained channel scene recognition model can accurately mine the characteristics of the channel characteristic parameters and realize accurate recognition of the wireless channel scene based on the mined characteristics.

In yet another embodiment of the present invention, a computer readable storage medium is provided, in which a computer program is stored, which when executed by a processor, implements the steps of any of the model training methods described above.

When the computer program stored in the computer readable storage medium provided by the embodiment of the invention is executed to perform model training, the initial channel scene recognition model can be trained based on the sample CIR data, so that the wireless channel scene determined based on the output result of the channel scene recognition model obtained by training is similar to the real sample wireless channel scene, that is, the accuracy of the output result of the channel scene recognition model obtained by training is higher. Accurate wireless channel scene recognition can be realized based on the channel scene recognition model with higher accuracy.

In yet another embodiment of the present invention, there is also provided a computer program product containing instructions that, when run on a computer, cause the computer to perform any of the wireless channel scene recognition methods of the above embodiments.

When the computer program provided by the embodiment of the invention is executed to identify the wireless channel scene, the channel characteristic parameters can represent the interfered condition of the target wireless channel, and the interference caused by the channel environments of different wireless channel scenes on the wireless channel is different, so that the identification of the wireless channel scene to which the target wireless channel belongs can be realized based on the channel characteristic parameters in the embodiment of the invention. And the channel scene recognition model is obtained by pre-training, and the trained channel scene recognition model can accurately mine the characteristics of the channel characteristic parameters and realize accurate recognition of the wireless channel scene based on the mined characteristics.

In yet another embodiment of the present invention, there is also provided a computer program product containing instructions that, when run on a computer, cause the computer to perform the model training method of any of the above embodiments.

When the computer program provided by the embodiment of the invention is executed to perform model training, the initial channel scene recognition model can be trained based on the sample CIR data, so that the wireless channel scene determined based on the output result of the channel scene recognition model obtained by training is similar to the real sample wireless channel scene, namely the accuracy of the output result of the channel scene recognition model obtained by training is higher. Accurate wireless channel scene recognition can be realized based on the channel scene recognition model with higher accuracy.

In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, produces a flow or function in accordance with embodiments of the present invention, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions may be stored in or transmitted from one computer-readable storage medium to another, for example, by wired (e.g., coaxial cable, optical fiber, digital Subscriber Line (DSL)), or wireless (e.g., infrared, wireless, microwave, etc.). The computer readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains an integration of one or more available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid State Disk (SSD)), etc.

It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

In this specification, each embodiment is described in a related manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. In particular, for apparatus, electronic devices, storage media, computer program product embodiments, the description is relatively simple, as it is substantially similar to method embodiments, with reference to the description of method embodiments in part.

The foregoing description is only of the preferred embodiments of the present invention and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. A method for identifying a wireless channel scene, the method comprising:

2. The method of claim 1, further comprising, prior to said inputting the channel characteristic parameters into a pre-trained wireless channel scene recognition model to obtain an output result:

3. A method of model training, the method comprising:

4. A method according to claim 3, wherein said obtaining sample CIR data representative of a communication condition of a sample radio channel comprises:

and/or

5. A method according to claim 3, wherein in case PDP is included in the sample channel characteristic parameters, before the inputting the sample channel characteristic parameters into the initial channel scene recognition model, obtaining an output result, further comprising:

6. A method according to claim 3, further comprising, prior to said inputting said sample channel characteristic parameters into an initial channel scene recognition model to obtain an output result:

7. The method according to any of claims 3-6, further comprising, after said adjusting model parameters of said initial channel scene recognition model based on said model loss, a channel scene recognition model:

8. A wireless channel scene recognition apparatus, the apparatus comprising:

9. The apparatus of claim 8, wherein the apparatus further comprises:

the first result obtaining module is specifically configured to:

10. A model training apparatus, the apparatus comprising:

11. The apparatus of claim 10, wherein the sample CIR acquisition module is configured to:

and/or

12. The apparatus of claim 10, wherein the apparatus further comprises:

13. The apparatus of claim 10, wherein the apparatus further comprises:

the second result obtaining module is specifically configured to:

14. The apparatus according to any one of claims 10-13, wherein the apparatus further comprises:

15. The electronic equipment is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory are communicated with each other through the communication bus;

a memory for storing a computer program;

a processor for carrying out the method steps of any one of claims 1-2 or 3-7 when executing a program stored on a memory.

16. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored therein a computer program which, when executed by a processor, implements the method steps of any of claims 1-2 or 3-7.