Radio frequency path configuration model training method and radio frequency path detection method

Abstract

The invention relates to a radio frequency path configuration model training method and a radio frequency path detection method. The radio frequency path configuration model training method comprises the steps that a training data set is acquired, and the training data set comprises feature matrixes of multiple pieces of channel state information and tags of radio frequency paths with the lowest error rate corresponding to the feature matrixes; and taking the characteristic matrix of the channel state information as the input of a first neural network model, taking the tag of the radio frequency path with the lowest error rate corresponding to the channel state information as the output of the first neural network model, and training the first neural network model to obtain a radio frequency path configuration model. According to the invention, the technical problem of low test efficiency in the research and development test process of the mobile phone radio frequency module is solved.

Description

Technical Field

[0001] The present application relates to the field of radio frequency engineering, and in particular to a training method for a radio frequency path configuration model and a radio frequency path detection method. Background Art

[0002] In the current mobile communications field, the research and development of mobile phone RF modules involves numerous testing and calibration tasks, especially in the context of the popularization of Multiple Input Multiple Output (MIMO) technology. MIMO technology uses multiple antennas at the transmitter and receiver to achieve multipath transmission and reception of signals, thereby significantly improving communication quality. The implementation of this technology is manifested in the demand for a large number of RF carriers on the RF side. Especially in the testing phase, cellular modules need to perform throughput tests of Long Term Evolution (LTE) carrier aggregation (CA), 5G NR carrier aggregation (NR-CA), E-UTRA (Evolved Universal Terrestrial Radio Access) and NR dual connectivity (E-UTRA-NR Dual Connectivity, EN-DC). In order to achieve maximum throughput, all antennas usually need to be connected, and at least four antennas are usually required for medium and high frequency bands.

[0003] For example, the typical combinations of EN-DC are N78A[4]; A[1]+B1A[2]+B3A[4]. Among them, A[1] involves dual CA connection of 5GNR band N78 and 4GLTE bands B1 and B3, which requires connecting up to 4+2+4 RF channels at the same time. To conduct throughput tests on multiple EN-DC combinations, rewiring is required every time a different combination is switched, which is not only complicated but also extremely cumbersome, seriously affecting test efficiency. Summary of the invention

[0004] The embodiments of the present application provide a method for training a radio frequency path configuration model and a method for detecting a radio frequency path, which at least solve the problem of low testing efficiency during the development and testing of mobile phone radio frequency modules.

[0005] According to one aspect of an embodiment of the present application, a training method for a radio frequency path configuration model is provided, comprising: obtaining a training data set, wherein the training data set comprises: a plurality of feature matrices of channel state information and a label of a radio frequency path with the lowest bit error rate corresponding to each feature matrix; using the feature matrix of the channel state information as the input of a first neural network model, using the label of the radio frequency path with the lowest bit error rate corresponding to the channel state information as the output of the first neural network model, training the first neural network model, and obtaining a radio frequency path configuration model.

[0006] According to another aspect of an embodiment of the present application, a RF path detection method is also provided, including: obtaining a characteristic matrix of channel state information of a channel combination to be detected, the channel combination to be detected including one or more channels; using a trained RF path configuration model to process the characteristic matrix of channel state information of the channel combination to be detected to obtain a label of a first RF path combination, wherein the first RF path combination is the RF path combination with the lowest bit error rate calculated by the RF path configuration model according to the characteristic matrix, and the first RF path combination includes one or more RF paths; opening the RF path in the first RF path combination between the test mainboard and the test mainboard; and detecting the transmission capability of the test mainboard to the channel combination to be detected through the first RF path combination.

[0007] According to another aspect of an embodiment of the present application, there is further provided an electronic device, including: a processor, and a memory storing a program, wherein the program includes instructions, and when the instructions are executed by the processor, the processor executes the above method.

[0008] According to another aspect of the embodiments of the present application, a non-transitory machine-readable medium storing computer instructions is also provided, where the computer instructions are used to enable a computer to execute the above method.

[0009] According to another aspect of the embodiments of the present application, a computer program product is provided, including a computer program, wherein the computer program is used to enable the computer to execute the above method when executed by a processor of the computer.

[0010] Beneficial effects of the embodiments of the present application:

[0011] In an embodiment of the present application, a neural network model for RF path selection is trained and the optimal RF path is automatically selected using the neural network model, thereby reducing the tedious manual testing operations during the mobile phone RF module testing process and achieving the technical effect of improving the automation level and test efficiency of the mobile phone RF module testing process.

[0012] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present application, and for ordinary technicians in this field, other embodiments can be obtained based on these drawings without creative work.

[0014] Figure 1 is a flow chart of a method for training a radio frequency path configuration model according to an embodiment of the present application;

[0015] Figure 2 is a flow chart of a radio frequency path detection method according to an embodiment of the present application;

[0016] Figure 3 is a schematic structural diagram of an electronic device of this embodiment;

[0017] Figure 4 It is a structural diagram of a radio frequency path configuration system according to an embodiment of the present application. DETAILED DESCRIPTION

[0018] Embodiments of the present embodiment will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present embodiment are shown in the accompanying drawings, it should be understood that the present embodiment can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein, which are instead provided for a more thorough and complete understanding of the present embodiment. It should be understood that the drawings and embodiments of the present embodiment are only for exemplary purposes and are not intended to limit the scope of protection of the present embodiment.

[0019] In order to better understand the embodiments of the present application, the technical terms involved in the embodiments of the present application are explained as follows:

[0020] MIMO, in order to greatly increase the channel capacity, uses multiple antennas at both the transmitting and receiving ends, forming an antenna system with multiple channels between the transmitting and receiving ends. An obvious feature of the MIMO system is its extremely high spectrum utilization efficiency. On the basis of making full use of existing spectrum resources, it obtains reliability and effectiveness gains by utilizing spatial resources, but at the cost of increasing the processing complexity of the transmitting and receiving ends. Massive MIMO technology uses a large number of antennas to serve a relatively small number of users, which can effectively improve spectrum efficiency.

[0021] LTE is the long-term evolution of the UMTS (Universal Mobile Telecommunications System) technical standard developed by 3GPP (The 3rd Generation Partnership Project). The LTE system introduces key technologies such as OFDM (Orthogonal Frequency Division Multiplexing) and MIMO, which significantly increases spectrum efficiency and data transmission rate.

[0022] CA, or carrier aggregation, is a technology that increases transmission bandwidth in LTE-Advanced systems to meet the requirements of single-user peak rate and system capacity improvement.

[0023] E-UTRA is the core network architecture of the LTE system, a high-speed wireless access technology that can provide faster data transmission rates, higher capacity, lower latency and better coverage.

[0024] 5GNR, the abbreviation of 5G New Radio, is a global 5G standard based on the new air interface design of OFDM. It is also a very important foundation for the next generation of cellular mobile technology. 5G technology will achieve ultra-low latency and high reliability.

[0025] In the process of developing mobile phone RF modules, when performing the enhanced LTE and NR dual connection (EN-DC) test, the test motherboard (a development board or evaluation board that can be used to simulate the actual mobile phone hardware environment, usually provided by chip suppliers or equipment manufacturers, and the test motherboard can support multiple RF paths, and different RF paths can support one or more channels) and the test equipment (wireless communication test equipment, which can simulate LTE and 5GNR network environments and perform various tests and measurements) need to connect multiple RF channels at the same time, which is not only complicated to operate, but also extremely cumbersome. Every time you switch to a different combination, you need to rewire, which seriously affects the test efficiency. In view of this, there is an urgent need for a new method to reduce the complexity of wiring to improve work efficiency.

[0026] To solve the above technical problems, the technical solution proposed in this application can effectively reduce the complexity of wiring selection by training a neural network model for RF path selection, and automatically select the optimal RF path using the neural network model. Through intelligent RF channel selection, the efficiency and operation problems in the RF module test mentioned above can be solved, thereby improving the performance and reliability of the entire RF test process. The details are described below.

[0027] Figure 1is a flow chart of a method for training a radio frequency path configuration model according to an embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps:

[0028] Step S102, obtaining a training data set, wherein the training data set includes: a plurality of channel state information feature matrices and labels of radio frequency path combinations with the lowest bit error rate corresponding to each feature matrix.

[0029] In the embodiments of the present application, the RF path refers to the physical or logical path for signal transmission from the transmitter to the receiver. It includes antennas, transmission lines, filters, amplifiers and other RF components. These components work together to ensure that the signal can be effectively transmitted from the transmitter to the receiver.

[0030] A frequency band refers to a continuous frequency range, and a channel refers to an information transmission channel divided within this frequency band for communication. The channel state refers to the current state and characteristics of a channel in a wireless communication system. The channel state describes the various influencing factors that a signal experiences during transmission.

[0031] Channel state information refers to the specific description and measurement results of the channel state. The characteristic matrix of the channel state information is a mathematical representation of the characteristics of the wireless communication channel. In a test scenario where a combination of M RF paths is tested against a channel combination of N channels, the characteristic matrix H of the channel state information is an N×M matrix, where each element h ij It indicates the channel state information of the jth channel transmitted by the i-th radio frequency path, i ranges from 1 to N, and j ranges from 1 to M.

[0032] The bit error rate (BER) of a radio frequency path refers to the ratio of the number of erroneously received bits among the bits transmitted through the radio frequency path to the total number of received bits in a radio frequency communication system. In an embodiment of the present application, each feature matrix also corresponds to a radio frequency path combination with the lowest bit error rate. Specifically, each feature matrix includes channel state information of a channel combination for each of a plurality of radio frequency path combinations, wherein the bit error rate of each radio frequency path combination when transmitting the channel combination is different, and the label of the radio frequency path combination with the lowest bit error rate corresponding to the feature matrix of each channel state information in the training data set indicates the radio frequency path combination with the lowest bit error rate for the channel combination corresponding to the feature matrix among the plurality of radio frequency path combinations included in the feature matrix.

[0033] In the specific implementation, assuming BER k is the transmission bit error rate of the kth RF channel combination, then the RF channel combination with the lowest bit error rate k *It can be selected by the following formula: * =argmin k BER k .

[0034] In an embodiment of the present application, the characteristic matrices of the channel state information of multiple channels and the labels of the radio frequency paths with the lowest bit error rates corresponding to each characteristic matrix are used as training data sets to train the neural network model.

[0035] Step S104, taking the characteristic matrix of the channel state information as the input of the first neural network model, taking the label of the RF path with the lowest bit error rate corresponding to the channel state information as the output of the first neural network model, training the first neural network model to obtain the RF path configuration model.

[0036] In an embodiment of the present application, an RNN model can be used as the above-mentioned first neural network model, the characteristic matrix of the channel state information is used as the input of the first neural network model, and the label of the RF path with the lowest bit error rate corresponding to the channel state information is used as the output of the first neural network model to train the first neural network model.

[0037] The RNN model, or recurrent neural network model, is a neural network model used to process sequence data. It is particularly suitable for processing data with time series characteristics, such as text, speech, and time series prediction. The main feature of the RNN model is that there are connections between neurons in its hidden layer, allowing information to be passed between multiple time steps, thereby capturing the temporal dependencies in sequence data.

[0038] Exemplarily, in an embodiment of the present application, the first neural network model is trained, and a cross entropy loss function can be used to calculate the gap between the predicted value and the true value of the first neural network model, and then the loss function is minimized by the gradient descent method to update the weights and parameters of the RNN model.

[0039] In the training process, the RNN model is used and the input is the feature matrix x i , the output is the label of the best RF path:

[0040]

[0041] This loss function will minimize the cross entropy between the model's output and the actual label. The loss function is then minimized through the gradient descent algorithm to update the weights and parameters of the RNN model.

[0042] Through the above technical scheme, by training the neural network model for RF path selection, a RF path configuration model that can be applied to RF path testing can be obtained, so that the RF path configuration model can be used to automatically select the optimal RF path, reducing the tedious manual testing operations during the mobile phone RF module testing process, and achieving the technical effect of improving the automation level and test efficiency of the mobile phone RF module testing process.

[0043] Current terminal devices need to support multiple channel combinations to cope with complex communication scenarios. Therefore, in the embodiment of the present application, training data sets corresponding to multiple channel combinations can be obtained, and these training data sets can be used to train a radio frequency path configuration model, so that the radio frequency path configuration model can output the corresponding radio frequency path labels under multiple channel combination scenarios.

[0044] The method can be implemented on a test device. The test device is connected to each RF socket of the mainboard in communication. After the test device obtains the RF path tags based on the method, it can conduct the connection between the RF sockets corresponding to these RF path tags in the mainboard, so that the device can automatically control the wiring status between the mainboard and the mainboard, without the need to rewire each time switching to a different combination, eliminating the reliance on physical connections in traditional testing.

[0045] According to an optional embodiment of the present application, step S102 is executed to obtain a training data set, including the following steps: obtaining training data sets corresponding to multiple channel combinations, wherein the training data set corresponding to each channel combination includes test data of one or more RF path combinations for the channel combination, wherein the test data includes at least the signal strength and bit error rate of each RF path in each RF path combination; according to the training data sets corresponding to the multiple channel combinations, obtaining a characteristic matrix of channel state information corresponding to the multiple channel combinations and a label of the RF path combination with the lowest bit error rate.

[0046] In this optional embodiment, the multiple channel combinations may be multiple channel combinations supported by common terminal devices. Each channel combination may include one or more channels, and different channels may be located in the same or different frequency bands. For example, a channel combination may include channels on frequency band 1 and channels on frequency band 2. A radio frequency path combination may include one or more radio frequency paths.

[0047] It should be noted that the test data of each RF path combination to the channel combination includes the signal strength and bit error rate used to calculate each RF path in each RF path combination. For example, RF path combination 1 includes RF paths RF1, RF2 and RF3, and channel combination 1 includes CH1, CH2 and CH3. Then the test data of RF path combination 1 to channel combination 1 includes at least the signal strength and bit error rate of RF1 transmitting CH1, the signal strength and bit error rate of RF2 transmitting CH2, and the signal strength and bit error rate of RF3 transmitting CH3.

[0048] It can be understood that the test data may also include one or more aspects of channel quality, signal-to-noise ratio, modulation and coding scheme, delay, bandwidth, throughput, RF parameters, interference level, path loss and antenna parameters, etc., which can be selected according to actual needs, and the embodiments of the present application do not specifically limit this.

[0049] It should be noted that the signal strength of the above-mentioned radio frequency path is the strength of the received signal, and all signal strengths are measured under the same transmission strength.

[0050] After collecting the data, data preprocessing is performed, including: data cleaning. The data cleaning stage is used to deal with missing values, outliers and noise to ensure the accuracy of the data. Next, feature extraction is performed. The feature extraction step extracts channel state information from the cleaned data, such as signal strength, bit error rate, signal-to-noise ratio and interference level. Finally, the data partitioning stage divides the extracted channel state information into a training set, a validation set and a test set, where the training set is used to train the neural network model. Exemplarily, the data partitioning stage usually uses 70% of the data for training the model, 15% for verifying the performance of the model, and the remaining 15% for the final test. This proportional allocation helps to evaluate the generalization ability and accuracy of the model.

[0051] Data collection and sophisticated data preprocessing can help improve the quality of the data set, accelerate the model training process, improve the performance and stability of the model, and enhance the generalization and robustness of the model. Ultimately, it ensures that the model can effectively learn from high-quality data.

[0052] According to the training data sets corresponding to the multiple channel combinations, characteristic matrices of channel state information corresponding to the multiple channel combinations are obtained, which may specifically include the following steps: for each channel combination, the channel state information of the channel combination is obtained from the training data set, the channel state information including the center frequency, bandwidth and energy of each frequency band in the channel combination on each RF path combination, and the characteristic matrix of the channel state information is calculated based on the state information of the channel combination.

[0053] For example, channel combination 1 includes CH1, CH2 and CH3, and both RF path combinations 1 and 2 can transmit the channel combination, wherein RF path combination 1 includes RF paths RF1, RF2 and RF3, and RF path combination 2 includes RF2, RF1 and RF3. Then, the channel state information of channel combination 1 is obtained from the training data set, and the channel state information includes channel state information inf11 of CH1 on RF path RF1, channel state information inf12 of CH2 on RF path RF2, channel state information inf13 of CH3 on RF path RF3, and channel state information inf21 of CH1 on RF path RF2, channel state information inf12 of CH2 on RF path RF1, and channel state information inf33 of CH3 on RF path RF3.

[0054] Afterwards, a characteristic matrix of the channel state information may be calculated based on the above channel state information. In this example, the obtained channel state information may be a 3×3 matrix, with rows RF1, RF2, and RF3, and columns CH1, CH2, and CH3, expressed as:

[0055] (Feature Matrix 1)

[0056] It can be understood that the value of each inf in the matrix can be the eigenvalue after normalization or the like, or it can be the original value directly calculated from the test data, and the embodiments of the present application do not impose any restrictions on this.

[0057] According to another optional embodiment of the present application, executing step S102 to obtain a training data set also includes the following steps: for each channel combination, calculating the average bit error rate of each RF path combination according to the bit error rate of each RF path in each RF path combination in the test data, and taking the RF path combination with the lowest average bit error rate as the RF path combination with the lowest bit error rate.

[0058] In this optional embodiment, the channel combination includes one or more channels. One radio frequency path can be used to test the transmission capability of one or more channels. Each channel combination can be tested with one or more radio frequency path combinations.

[0059] Taking the above-mentioned channel combination as an example, both RF path combinations 1 and 2 can transmit the channel combination, then the average bit error rate of RF path combination 1 and the average bit error rate of RF path combination 2 are calculated respectively. Among them, the bit error rate of RF path combination 1 is specifically the average value between the bit error rate of RF1 transmitting CH1, the bit error rate of RF2 transmitting CH2, and the bit error rate of RF3 transmitting CH3. The bit error rate of RF path combination 2 is specifically the average value between the bit error rate of RF1 transmitting CH2, the bit error rate of RF2 transmitting CH1, and the bit error rate of RF3 transmitting CH3. By comparing the relative size between the bit error rate of RF path combination 1 and the bit error rate of RF path combination 2, the RF path combination with the smaller bit error rate is selected as the RF path with the lowest bit error rate corresponding to the above-mentioned feature matrix 1.

[0060] In some optional embodiments of the present application, characteristic matrices of channel state information corresponding to multiple channel combinations are obtained according to training data sets corresponding to multiple channel combinations, which is achieved by the following method: for each channel combination, the channel state information of the channel combination is obtained from the training data set, the channel state information including the center frequency, bandwidth and energy of each frequency band in the channel combination on each RF path combination; and the channel state information characteristic matrix is ​​calculated based on the state information of the channel combination.

[0061] In this optional embodiment, for each channel combination, the number of columns or rows of the characteristic matrix is ​​determined based on the channel state information. In this example, the channel state information includes the center frequency, bandwidth and energy of each frequency band in the channel combination on each RF path combination.

[0062] In this possible implementation manner, the center frequency, bandwidth and energy of each frequency band in the channel combination on each RF path combination in the channel state information can be obtained by extracting features from the data after data cleaning in the following manner:

[0063] Before calculating the energy of the frequency band, the cleaned data is transformed into a time-frequency transform, usually using the Fast Fourier Transform (FFT). FFT converts the signal from the time domain to the frequency domain and generates a spectrum graph, where the horizontal axis represents the frequency and the vertical axis represents the amplitude of the signal. The formula for the Fast Fourier Transform is as follows:

[0064]

[0065] Among them, X(f) is the frequency band component with frequency f, x(n) is the sampling value of the time domain signal, N is the length of the signal, and n is the sampling point index of the time domain.

[0066] Energy: Energy is usually obtained by calculating the square of the amplitude of each frequency band. For a frequency band f, its energy E(f) can be calculated using the following formula: E(f) = |π(f)| 2Where π(f) is the complex amplitude of the frequency band component at frequency f. The energy level of the frequency band can be estimated by calculating the relative size of the energy E(f). The energy level of each frequency band can be normalized to the range [0,1] for comparison.

[0067] Center frequency: You can select the peak frequency of the frequency band as the center frequency. Center frequency f c It can be obtained by finding the frequency component f with the highest energy E(f): c =argmaxE(f). Specifically, the following steps may be included: Input: spectrum data E(f), where f represents frequency and E(f) represents the energy of the frequency band. Initialization: Set a threshold θ to select the peak. Find all frequency points in the spectrum that satisfy E(f)>θ and use them as candidate peaks. For the candidate peak points, select the frequency with the highest energy as the center frequency of the signal. Output: The determined signal center frequency f c , through the above algorithm, the center frequency f of the signal can be estimated c , this frequency can be used for further frequency band selection. The specific value of the threshold θ can be adjusted according to system performance and signal characteristics.

[0068] Bandwidth calculation: The bandwidth of a frequency band can be estimated by measuring the width of the spectrum. A common method is to calculate the Full Width at Half Maximum (FWHM) of the spectrum, which is the frequency range where the spectrum drops to half of its peak amplitude.

[0069] In conjunction with the RF path configuration model, as an optional embodiment of the present application, the test data of the channel combination of one or more RF path combinations is used as input, and the characteristic matrix of the channel state information of the channel combination is used as output to train the second neural network model to obtain the channel feature extraction model. Therefore, the channel state information characteristic matrix of the channel combination to be tested can be quickly extracted through the channel feature extraction model during the RF path test.

[0070] In this optional embodiment, test data of the channel combination using one or more RF path combinations are used as training data and input into a second neural network model. The second neural network model may be an RNN model. The output obtained is a feature matrix of the channel state information of the channel combination. The feature matrix can be used to characterize the channel state of the channel combination corresponding to the input test data.

[0071] In this step, the loss function is the mean square error (MSE):

[0072] Among them, N t is the sample size, N ris the output dimension of each sample, H(i,j) is the actual feature matrix of the channel state information, is the estimated feature matrix of the channel state information. The actual feature matrix is ​​the channel state information feature matrix calculated based on the state information of the channel combination, and the estimated feature matrix is ​​the feature matrix output by the channel feature extraction model.

[0073] This loss function measures the difference between the estimated feature matrix of the channel state information and the actual feature matrix of the channel state information.

[0074] In a possible implementation, the test data may be obtained by traversing each RF path and each channel. For example, there are 10 RF paths and 10 channel combinations, and the 10 channel combinations include 20 channels in total. Then, the test data of each RF path for the 20 channels may be tested separately. When calculating the characteristic matrix of the channel state information of each channel combination, the channel state information of the channel combination under each RF path may be extracted from the test data, thereby constructing the channel state characteristic matrix.

[0075] In the above scheme, since a traversal sampling method is adopted, the interference between different channels is not considered when collecting data. Moreover, there is often a large amount of scattering in the actual use environment. In order to further improve the accuracy of model training, in a possible implementation method, after the characteristic matrix of the channel state information is calculated based on the channel state information, the Rayleigh fading channel model can be used to further process the characteristic matrix of the channel state information.

[0076] The Rayleigh fading channel model is a statistical model of the radio signal propagation environment. This model assumes that after the signal passes through the wireless channel, its signal amplitude is random, that is, "fading", and its envelope follows the Rayleigh distribution. This channel model can describe shortwave channels reflected by the ionosphere and troposphere, as well as urban environments with dense buildings.

[0077] After further processing the channel characteristic matrix using a Rayleigh fading channel model, the processed characteristic matrix of the channel state information is used for training the first neural network and the second neural network.

[0078] The present application also provides a method for detecting a radio frequency path. Figure 2 As shown, the method comprises the following steps:

[0079] Step S202: Acquire a characteristic matrix of channel state information of a channel combination to be detected, where the channel combination to be detected includes one or more channels.

[0080] Step S204, using the trained RF path configuration model to process the characteristic matrix of the channel state information of the channel combination to be detected to obtain a label of the first RF path, wherein the first RF path combination is the RF path combination with the lowest bit error rate calculated by the RF path configuration model according to the characteristic matrix, and the first RF path combination includes one or more RF paths.

[0081] The RF path configuration model in step S204 is: Figure 1 The model obtained by training the method in the illustrated embodiment is used to select a radio frequency path combination with the lowest bit error rate from multiple radio frequency path combinations using the radio frequency path configuration model.

[0082] Step S206, opening a radio frequency path in the first radio frequency path combination between the test mainboard and the test mainboard.

[0083] Step S208, testing the transmission capability of the mainboard for the channel combination to be tested by the first radio frequency path combination test.

[0084] As an optional embodiment of the present application, a radio frequency test instrument is used to verify the performance of the channel combination to be tested in combination with the first radio frequency path to ensure that it meets the expected communication quality standard.

[0085] In a possible implementation, once it is determined that the expected communication quality standard is met, the label of the first RF path combination and part or all of the channel state information of the first RF path combination for the channel combination to be detected may also be saved, such as the center frequency, bandwidth, etc., for subsequent use.

[0086] The above method provided in the embodiment of the present application uses a deep learning algorithm to optimize the switch operation. Based on the above method, detection can quickly locate the RF path combination used for testing, which can significantly improve the work efficiency of the test process.

[0087] According to an optional embodiment of the present application, executing step S202 to obtain a characteristic matrix of the channel state information of the channel combination to be detected can also be achieved by the following method: connecting any second RF path combination that supports the channel combination to be detected between the test main board; transmitting a signal of the channel combination to be detected through the second RF path combination to obtain test data of the received signal; inputting the test data into a pre-trained feature extraction model to obtain a characteristic matrix of the channel state information of the channel combination to be detected.

[0088] The second RF path combination is any RF path combination between the test mainboards that supports the channel combination to be detected. The feature extraction model is an RNN model. In this embodiment, after the second RF path combination between the test mainboards is turned on, the signal of the channel combination to be detected is transmitted through the second RF path, the test data of the signal is received, and then the test data is input into the feature extraction model trained above, so that the feature matrix of the channel state information of the channel combination to be detected can be obtained.

[0089] The embodiment of the present application also provides an electronic device, comprising: at least one processor; and a memory connected to the at least one processor in communication. The memory stores a computer program executable by the at least one processor, and the computer program is used to enable the electronic device to perform the method of the embodiment of the present application when executed by the at least one processor.

[0090] In an embodiment of the present application, the processor is used to execute a program for executing the following functions: obtaining a training data set, wherein the training data set includes: a plurality of characteristic matrices of channel state information and a label of the RF path with the lowest bit error rate corresponding to each characteristic matrix; using the characteristic matrix of the channel state information as the input of a first neural network model, using the label of the RF path with the lowest bit error rate corresponding to the channel state information as the output of the first neural network model, training the first neural network model, and obtaining a RF path configuration model.

[0091] Optionally, the processor is also used to execute a program for the following functions: obtaining training data sets corresponding to multiple channel combinations, wherein the training data sets corresponding to each channel combination include test data of the channel combination by multiple RF path combinations, wherein the test data includes at least the signal strength and bit error rate of each RF path in each RF path combination; and obtaining, based on the training data sets corresponding to the multiple channel combinations, a feature matrix of the channel state information corresponding to the multiple channel combinations and a label of the RF path combination with the lowest bit error rate.

[0092] Optionally, the processor is also used to execute a program having the following functions: for each channel combination, the average bit error rate of each RF path combination is calculated based on the bit error rate of each RF path in each RF path combination in the test data, and the RF path combination with the lowest average bit error rate is taken as the RF path combination with the lowest bit error rate.

[0093] Optionally, the processor is also used to execute a program having the following functions: for each channel combination, obtaining channel state information of the channel combination from a training data set, the channel state information including the center frequency, bandwidth and energy of each frequency band in the channel combination on each RF path combination; and obtaining a characteristic matrix of the channel state information based on the state information of the channel combination.

[0094] Optionally, the processor is also used to execute a program having the following functions: taking test data of channel combinations of multiple RF path combinations as input, and a characteristic matrix of channel state information of the channel combination as output, training a second neural network model, and obtaining a channel feature extraction model.

[0095] In an embodiment of the present application, the processor is used to execute a program for the following functions: obtaining a characteristic matrix of channel state information of a channel combination to be detected, the channel combination to be detected including one or more channels; processing the characteristic information of the channel combination to be detected using a trained RF path configuration model to obtain a label of a first RF path combination, wherein the first RF path combination is a RF path combination with the lowest bit error rate calculated by the RF path configuration model according to the characteristic matrix, and the first RF path combination includes one or more RF paths; opening the RF path in the first RF path combination between the test mainboard and the test mainboard; and detecting the transmission capability of the test mainboard to the channel combination to be detected through the first RF path combination.

[0096] Optionally, the processor is also used to execute a program for the following functions: connecting any second RF path combination that supports the channel combination to be detected between the test mainboard and the test mainboard; transmitting the signal of the channel combination to be detected through the second RF path combination to obtain test data of the received signal; inputting the test data into a pre-trained channel feature extraction model to obtain a feature matrix of the channel state information of the channel combination to be detected.

[0097] Optionally, the processor is further configured to execute a program having the following functions: storing the to-be-detected channel combination and the first radio frequency path combination in an associated manner.

[0098] In an embodiment of the present application, the above-mentioned memory is used to store a program for the following functions: obtaining a training data set, wherein the training data set includes: a plurality of characteristic matrices of channel state information and a label of the RF path with the lowest bit error rate corresponding to each characteristic matrix; using the characteristic matrix of the channel state information as the input of the first neural network model, using the label of the RF path with the lowest bit error rate corresponding to the channel state information as the output of the first neural network model, training the first neural network model, and obtaining a RF path configuration model.

[0099] Optionally, the memory is also used to store a program for the following functions: obtaining training data sets corresponding to multiple channel combinations, wherein the training data sets corresponding to each channel combination include test data of the channel combination by multiple RF path combinations, wherein the test data includes at least the signal strength and bit error rate of each RF path in each RF path combination; and obtaining, based on the training data sets corresponding to the multiple channel combinations, a feature matrix of the channel state information corresponding to the multiple channel combinations and a label of the RF path combination with the lowest bit error rate.

[0100] Optionally, the above-mentioned memory is also used to store a program for the following functions: for each channel combination, the average bit error rate of each RF path combination is calculated according to the bit error rate of each RF path in each RF path combination in the test data, and the RF path combination with the lowest average bit error rate is taken as the RF path combination with the lowest bit error rate.

[0101] Optionally, the above-mentioned memory is also used to store programs with the following functions: for each channel combination, obtaining channel state information of the channel combination from a training data set, the channel state information including the center frequency, bandwidth and energy of each frequency band in the channel combination on each RF path combination; and calculating a characteristic matrix of the channel state information based on the state information of the channel combination.

[0102] Optionally, the memory is also used to store a program with the following functions: taking the test data of the channel combination of multiple RF path combinations as input and the characteristic matrix of the channel state information of the channel combination as output, training the second neural network model and obtaining a channel feature extraction model.

[0103] Optionally, the above-mentioned memory is also used to store programs for the following functions: inputting the center frequency into the second neural network model for processing, obtaining the RF path matrix output by the second neural network model; using the fading channel model to generate a channel state information matrix corresponding to the RF path matrix; determining the amplitude of each channel state information in the channel state information matrix, and determining the characteristic matrix of the channel state information based on the amplitude.

[0104] In an embodiment of the present application, the above-mentioned memory is used to store a program for the following functions: obtaining a characteristic matrix of channel state information of a channel combination to be detected, the channel combination to be detected including one or more channels; using a trained RF path configuration model to process the characteristic information of the channel combination to be detected to obtain a label of a first RF path combination, wherein the first RF path combination is a RF path combination with the lowest bit error rate calculated by the RF path configuration model according to the characteristic matrix, and the first RF path combination includes one or more RF paths; opening the RF path in the first RF path combination between the test mainboard and the test mainboard; and detecting the transmission capability of the test mainboard to the channel combination to be detected through the first RF path combination.

[0105] Optionally, the memory is also used to store programs for the following functions: connecting any second RF path combination between the test mainboard and the test mainboard that supports the channel combination to be detected; transmitting the signal of the channel combination to be detected through the second RF path combination to obtain test data of the received signal; inputting the test data into a pre-trained channel feature extraction model to obtain a feature matrix of the channel state information of the channel combination to be detected.

[0106] Optionally, the memory is further used to store a program for the following function: associatively storing a channel combination to be detected and a first radio frequency path combination.

[0107] An embodiment of the present application also provides a non-transitory machine-readable medium storing a computer program, wherein the computer program, when executed by a processor of a computer, is used to cause the computer to execute the method of the embodiment of the present application.

[0108] The embodiment of the present application also provides a computer program product, including a computer program, wherein the computer program, when executed by a processor of a computer, is used to cause the computer to execute the method of the embodiment of the present application.

[0109] refer to Figure 3 , the structural block diagram of the electronic device that can be used as the server or client of the embodiment of the present application will now be described, which is an example of the hardware device that can be applied to various aspects of the present application. The electronic device is intended to represent various forms of digital electronic computer equipment, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices and other similar computing devices. The components shown herein, their connections and relationships, and their functions are only examples, and are not intended to limit the implementation of the present application described and / or required herein.

[0110] like Figure 3 As shown, the electronic device includes a computing unit 301, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 302 or a computer program loaded from a storage unit 308 into a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic device can also be stored. The computing unit 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0111] Multiple components in the electronic device are connected to the I / O interface 305, including: an input unit 306, an output unit 307, a storage unit 308, and a communication unit 309. The input unit 306 can be any type of device that can input information to the electronic device, and the input unit 306 can receive input digital or character information, and generate key signal input related to user settings and / or function control of the electronic device. The output unit 307 can be any type of device that can present information, and can include but is not limited to a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. The storage unit 308 can include but is not limited to a disk, an optical disk. The communication unit 309 allows the electronic device to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks, and can include but is not limited to a modem, a network card, an infrared communication device, a wireless communication transceiver, and / or a chipset, such as a Bluetooth device, a WiFi device, a WiMax device, a cellular communication device, and / or the like.

[0112] The computing unit 301 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 301 include, but are not limited to, a CPU, a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 301 performs the various methods and processes described above. For example, in some embodiments, the method embodiments of the present application may be implemented as a computer program, which is tangibly contained in a machine-readable medium, such as a storage unit 308. In some embodiments, part or all of the computer program may be loaded and / or installed on an electronic device via a ROM 302 and / or a communication unit 309. In some embodiments, the computing unit 301 may be configured to perform the above method in any other appropriate manner (e.g., by means of firmware).

[0113] Figure 4 is a structural diagram of a radio frequency path configuration system according to an embodiment of the present application, such as Figure 4 As shown, the system includes a mobile phone mainboard antenna interface (T1), a radio frequency signal input module (T2), a frequency band switching selection module (T3), a data acquisition module (T4), a deep learning module (T5), a frequency band selection output module (T6), and a radio frequency test instrument (T7).

[0114] Among them, the mobile phone mainboard antenna interface (T1) is the RF socket on the mobile phone mainboard.

[0115] The RF signal input module (T2) includes a multi-way switch, an amplifier, a mixer and a filter. Its main function is to receive the signal from the device under test (DUT). It transmits the signal to the frequency band switching selection module (T3) for processing.

[0116] The frequency band switching selection module (T3) performs signal acquisition and preprocessing, involving data cleaning, feature extraction and data partitioning, ensuring data quality and preparing for model training.

[0117] The data acquisition module (T4) uses frequency-time modulation and multi-channel technology to process signals through FFT and other algorithms to extract frequency band characteristics.

[0118] The deep learning module (T5) is mainly used to train the RF path configuration model.

[0119] The frequency band selection output module (T6) is used to select one from all possible radio frequency path combinations to minimize the bit error rate and optimize the signal quality, and update the database.

[0120] The RF test instrument (T7) is used to verify the performance of the selected RF path to ensure that it meets the expected communication quality standards.

[0121] The computer program for implementing the method of the embodiment of the present application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0122] In the context of the present application embodiment, machine-readable medium can be a tangible medium that can contain or store a program for use by an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable signal medium can include but is not limited to electronic, magnetic, optical, electromagnetic or infrared systems, devices or equipment, or any suitable combination of the above. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

[0123] It should be noted that the term "including" and its variations used in the embodiments of the present application are open inclusions, that is, "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments". The modifications of "one" and "multiple" mentioned in the embodiments of the present application are illustrative and not restrictive. Those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0124] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the embodiments of the present application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0125] The various steps described in the method implementation methods provided in the embodiments of the present application can be performed in different orders and / or in parallel. In addition, the method implementation methods may include additional steps and / or omit the steps shown. The scope of protection of the present application is not limited in this respect.

[0126] The term "embodiment" in this specification refers to specific features, structures or characteristics described in conjunction with the embodiment that can be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. The various embodiments in this specification are described in a related manner, and the same or similar parts between the various embodiments refer to each other. In particular, for the device, equipment, and system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts refer to the partial description of the method embodiment.

[0127] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of patent protection. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the scope of protection of the present application. Therefore, the scope of protection of the present application shall be subject to the attached claims.

Claims

1. A training method for a radio frequency path configuration model, include: Acquire a training data set, wherein the training data set includes: a plurality of channel state information feature matrices and labels of radio frequency paths with the lowest bit error rates corresponding to each feature matrix; The characteristic matrix of the channel state information is used as the input of the first neural network model, the label of the RF path combination with the lowest bit error rate corresponding to the channel state information is used as the output of the first neural network model, and the first neural network model is trained to obtain the RF path configuration model.

2. The method according to claim 1, in, Get the training dataset, including: Acquire training data sets corresponding to a plurality of channel combinations, respectively, wherein the training data set corresponding to each channel combination includes test data of the channel combination respectively obtained by a plurality of radio frequency path combinations, wherein the test data includes at least a signal strength and a bit error rate of each radio frequency path in each radio frequency path combination; According to the training data sets respectively corresponding to the multiple channel combinations, characteristic matrices of the channel state information respectively corresponding to the multiple channel combinations and labels of the radio frequency path combination with the lowest bit error rate are obtained.

3. The method according to claim 2, in, Get the training data set, also include: For each channel combination, the average bit error rate of each RF path combination is calculated according to the bit error rate of each RF path in each RF path combination in the test data, and the RF path combination with the lowest average bit error rate is used as the RF path combination with the lowest bit error rate.

4. The method according to claim 2, in, Acquiring, according to the training data sets respectively corresponding to the multiple channel combinations, characteristic matrices of channel state information respectively corresponding to the multiple channel combinations, including: For each channel combination, acquiring channel state information of the channel combination from the training data set, the channel state information including a center frequency, a bandwidth, and an energy of each frequency band in the channel combination on each radio frequency path combination; A characteristic matrix of the channel state information is calculated based on the state information of the channel combination.

5. The method according to claim 4, in, The method further comprises: The channel state information of the channel combination of the multiple radio frequency path combinations is respectively used as input, and the characteristic matrix of the channel state information of the channel combination is used as output, and a second neural network model is trained to obtain a channel feature extraction model.

6. A radio frequency path detection method, in, include: Acquire a characteristic matrix of channel state information of a channel combination to be detected, where the channel combination to be detected includes one or more channels; Processing the characteristic matrix of the channel state information of the channel combination to be detected by using a trained radio frequency path configuration model to obtain a label of a first radio frequency path combination, wherein the first radio frequency path combination is a radio frequency path combination with the lowest bit error rate calculated by the radio frequency path configuration model according to the characteristic matrix, and the first radio frequency path combination includes one or more radio frequency paths; Opening a radio frequency path in the first radio frequency path combination between the test mainboard and the test mainboard; The transmission capability of the test mainboard to the channel combination to be tested is tested through the first radio frequency path combination.

7. The method according to claim 6, in, Obtaining a characteristic matrix of channel state information of a channel combination to be detected, including: Conducting any second RF path combination supporting the channel combination to be detected between the test mainboard and the test mainboard; Transmitting a signal of the channel combination to be detected through the second radio frequency path combination to obtain channel state information of the channel combination to be detected by the second radio frequency path combination; The channel state information is input into a pre-trained channel feature extraction model to obtain a feature matrix of the channel state information of the channel combination to be detected.

8. An electronic device, include: A processor, and a memory storing a program, wherein the program comprises instructions which, when executed by the processor, cause the processor to perform the method according to any one of claims 1 to 7.

9. A non-transitory machine-readable medium storing computer instructions, in, The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 7.

10. A computer program product comprising a computer program, in, When the computer program is executed by a processor of a computer, the computer is used to cause the computer to execute the method according to any one of claims 1 to 7.

Images