Broadband wireless signal processing method and system and control equipment

Abstract

The invention provides a broadband wireless signal processing method, a broadband wireless signal processing system and control equipment. The method is applied to a signal processing system, and the signal processing system comprises a forward path (comprising a digital pre-distortion processing module and a power amplifier) and a feedback loop, the feedback loop comprises a band-pass filter (constraining the bandwidth of the output signal of the power amplifier), a signal reconstruction module (used for reconstructing the output signal of the band-pass filter according to the output signal of the digital pre-distortion processing module) and a parameter adjustment module. A digital pre-distortion processing module applies distortion operation opposite to non-linear distortion of a power amplifier to an initial baseband signal to obtain a first signal; amplifying the power of the first signal through a power amplifier to obtain a second signal; and the parameter adjustment module updates the digital pre-distortion processing module according to the output signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module. According to the invention, the problems of a high-sampling-rate ADC can be solved, and the dynamic updating of the digital pre-distortion processing module can be realized.

Description

Technical Field

[0001] This application relates to the technical field of wireless signal processing, and particularly to a broadband wireless signal processing method, system, and control device. Background Art

[0002] With the development of mobile communication technology, in order to meet the requirements of high-rate communication, the modulation bandwidth of signals has evolved from 5 MHz in the 3G era to 100 MHz in the Sub-6 GHz band and 400 MHz in the millimeter-wave band in the 5G era. The significant increase in modulation bandwidth has brought new challenges to digital predistortion technology.

[0003] Traditional broadband wireless signal processing solutions are implemented using Figure 1 the architecture shown. In Figure 1 , the analog-to-digital converter (ADC) in the feedback loop needs to collect the baseband output signal of the power amplifier, and then, combined with the original baseband input signal x(n) of the power amplifier before the digital-to-analog converter (DAC), perform coefficient identification of the digital predistortion (DPD) device. Since the power amplifier is a non-linear device, the spectrum of its output signal will be broadened to 3 to 5 times the bandwidth of the original signal. According to the Nyquist sampling theorem, in order to completely collect the output signal of the power amplifier, the sampling rate of the ADC in the feedback loop needs to reach 3 to 5 times the bandwidth of the original signal. However, such a high-sampling-rate ADC will significantly increase the hardware cost, thus restricting the wide application of digital predistortion technology.

[0004] Secondly, the characteristics of the power amplifier will drift over time and with changes in the working environment, such as due to temperature changes, aging effects, or fluctuations in working conditions, etc., resulting in dynamic changes in its non-linear behavior. This characteristic drift will cause the pre-trained DPD model to gradually become invalid and unable to accurately compensate for the non-linear distortion of the power amplifier, thus affecting the linearity and overall performance of the system. Summary of the Invention

[0005] This application provides a broadband wireless signal processing method, system, and control device to solve the defects of the high-sampling-rate ADC in the feedback loop in the prior art, and the defect that the pre-trained DPD model cannot accurately compensate for the non-linear distortion of the power amplifier due to the drift characteristics of the power amplifier.

[0006] The present application provides a broadband wireless signal processing method, which is applied to a signal processing system. The signal processing system includes a forward path and a feedback loop. The forward path at least includes a digital pre-distortion processing module and a power amplifier connected in sequence. The feedback loop at least includes a band-pass filter, a signal reconstruction module, and a parameter adjustment module connected in sequence. The band-pass filter is connected to the output end of the power amplifier, and the parameter adjustment module is also connected to the output end of the digital pre-distortion processing module; the signal processing method includes: Receiving, by the digital pre-distortion processing module, an initial baseband signal sent by a sending end, and applying a distortion operation opposite to the non-linear distortion of the power amplifier to the initial baseband signal to obtain a first signal; Amplifying, by the power amplifier, the power of the first signal to obtain a second signal; Wherein, the parameter adjustment module is configured to adjust the parameters of the digital pre-distortion processing module according to the output signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module. The band-pass filter is configured to constrain the bandwidth of the output signal of the power amplifier. The signal reconstruction module is configured to reconstruct the output signal of the band-pass filter into a signal meeting preset conditions according to the output signal of the digital pre-distortion processing module.

[0007] According to the broadband wireless signal processing method provided by the present application, the parameter adjustment module is further connected to the input end of the digital pre-distortion processing module. Adjusting the parameters of the digital pre-distortion processing module according to the output signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module includes: Determining the similarity between the input signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module; If the similarity is less than a preset threshold, determining a new parameter value of the digital pre-distortion processing module according to the output signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module; Inputting the new parameter value into the digital pre-distortion processing module, so that the digital pre-distortion processing module processes the signal sent by the sending end according to the new parameter value.

[0008] According to the broadband wireless signal processing method provided by the present application, determining the similarity between the input signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module includes: Determining the normalized mean square error between the input signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module; Determining the similarity according to the normalized mean square error.

[0009] According to the broadband wireless signal processing method provided by the present application, a digital-to-analog converter and a first frequency conversion module are further included in the forward path. The input end of the digital-to-analog converter is connected to the output end of the digital predistortion processing module. The input end of the first frequency conversion module is connected to the output end of the digital-to-analog converter. The output end of the first frequency conversion module is connected to the input end of the power amplifier. The digital-to-analog converter is used to convert the output signal of the digital predistortion processing module into an analog signal. The first frequency conversion module is used to boost the spectrum of the output signal of the digital-to-analog converter to the target radio frequency band; An analog-to-digital converter and a second frequency conversion module are further included in the feedback loop. The input end of the second frequency conversion module is connected to the output end of the band-pass filter. The input end of the analog-to-digital converter is connected to the output end of the second frequency conversion module. The output end of the analog-to-digital converter is connected to the input end of the signal reconstruction module. The analog-to-digital converter is used to convert the output signal of the second frequency conversion module into a digital signal. The second frequency conversion module is used to restore the spectrum of the output signal of the band-pass filter to the same radio frequency band as the input signal of the first frequency conversion module.

[0010] According to the broadband wireless signal processing method provided by the present application, the signal reconstruction module is obtained in advance through the following steps: Obtain the baseband signal sent by the sending end and the output signal of the analog-to-digital converter; Using the output signal of the analog-to-digital converter as input sample data, using the baseband signal sent by the sending end as output sample data, and aiming to minimize the difference between the output signal of the analog-to-digital converter and the baseband signal sent by the sending end, train the first neural network model to obtain the signal reconstruction module.

[0011] According to the broadband wireless signal processing method provided by the present application, the first neural network model is an autoencoder. The autoencoder includes an encoder and a decoder. Using the output signal of the analog-to-digital converter as input sample data, using the baseband signal sent by the sending end as output sample data, and aiming to minimize the difference between the output signal of the analog-to-digital converter and the baseband signal sent by the sending end, training the first neural network model includes: Randomly initialize the weights and biases in the neural network layer of the autoencoder; Input the output signal of the analog-to-digital converter into the encoder to obtain encoded features; Input the encoded features into the decoder to obtain reconstructed features after decoding; Determine the difference value between the reconstructed features and the features of the baseband signal sent by the sending end through a first loss function; Based on the first difference value, determine the gradients corresponding to the weights and biases in the autoencoder through the backpropagation algorithm; Update the weights and biases in the autoencoder through the stochastic gradient descent algorithm; Taking minimizing the first difference value as the goal, repeat the above steps until the first preset stopping condition is met. The autoencoder when the first preset stopping condition is met is the signal reconstruction module.

[0012] According to the broadband wireless signal processing method provided by the present application, the digital predistortion processing module is obtained in advance through the following steps: Obtain the baseband signal sent by the transmitting end and the output signal of the signal reconstruction module; Taking the output signal of the signal reconstruction module as the input sample data, taking the baseband signal sent by the transmitting end as the output sample data, and taking minimizing the difference between the output signal of the signal reconstruction module and the baseband signal sent by the transmitting end as the goal, train the second neural network model to obtain the digital predistortion processing module.

[0013] According to the broadband wireless signal processing method provided by the present application, the second neural network model is a multi-layer perceptron network. The multi-layer perceptron network includes an input layer, multiple hidden layers, and an output layer. Taking the output signal of the signal reconstruction module as the input sample data, taking the baseband signal sent by the transmitting end as the output sample data, and taking minimizing the difference between the output signal of the signal reconstruction module and the baseband signal sent by the transmitting end as the goal, training the second neural network model includes: Randomly initialize the weights and biases in the multi-layer perceptron network; Input the output signal of the signal reconstruction module into the multi-layer perceptron network to obtain a predicted baseband signal; Determine the second difference value between the predicted baseband signal and the baseband signal sent by the transmitting end through a second loss function; Based on the second difference value, determine the gradients corresponding to the weights and biases in the multi-layer perceptron network through the backpropagation algorithm; Update the weights and biases in the multi-layer perceptron network through the stochastic gradient descent algorithm; Taking minimizing the second difference value as the goal, repeat the above steps until the second preset stopping condition is met. The multi-layer perceptron network when the second preset stopping condition is met is the digital predistortion processing module.

[0014] The present application also provides a signal processing system, including a forward path and a feedback loop. At least a digital predistortion processing module and a power amplifier connected in sequence are included in the forward path. At least a band-pass filter, a signal reconstruction module, and a parameter adjustment module connected in sequence are included in the feedback loop. The band-pass filter is connected to the output end of the power amplifier, and the parameter adjustment module is also connected to the output end of the digital predistortion processing module; The digital predistortion processing module is configured to receive an initial baseband signal sent by a transmitting end through the digital predistortion processing module, and apply a distortion operation opposite to the non-linear distortion of the power amplifier to the initial baseband signal to obtain a first signal; The power amplifier is configured to amplify the power of the first signal through the power amplifier to obtain a second signal; The parameter adjustment module is configured to adjust the parameters of the digital predistortion processing module according to the output signal of the digital predistortion processing module and the output signal of the signal reconstruction module; The band-pass filter is configured to constrain the bandwidth of the output signal of the power amplifier; The signal reconstruction module is configured to reconstruct the output signal of the band-pass filter into a signal meeting preset conditions according to the output signal of the digital predistortion processing module.

[0015] The present application also provides a control device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the broadband wireless signal processing method described in any one of the above is implemented.

[0016] The present application also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the broadband wireless signal processing method described in any one of the above is implemented.

[0017] The present application also provides a computer program product, including a computer program. When the computer program is executed by a processor, the broadband wireless signal processing method described in any one of the above is implemented.

[0018] The broadband wireless signal processing method of the present application has at least the following technical effects: First, a digital pre-distortion processing module is provided, which can apply pre-distortion operations to the initial baseband signal, so that the signal output by the power amplifier can be restored to a linear state as much as possible after non-linear processing, improving the performance of the communication system and the signal quality; Second, a reconstruction module is provided, which can reconstruct a high-speed feedback signal using the high-speed baseband signal sent by the transmitting end, making up for the acquisition error caused by the low-speed ADC, achieving the performance of an approximate high-speed ADC, and avoiding the defect of high hardware cost when using an ADC with a high sampling rate in the related art; Third, a parameter adjustment module is provided, which can adjust the parameters of the digital pre-distortion processing module according to the output signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module, realizing the regular update of the digital pre-distortion processing module, enabling the digital pre-distortion processing module to adapt to the characteristic changes of the power amplifier, accurately compensating for the non-linear distortion of the power amplifier, and improving the linearity and overall performance of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0020] Figure 1 is a schematic structural diagram of a traditional signal processing system shown in an embodiment of the present application; Figure 2 is a flowchart of a broadband wireless signal processing method shown in an embodiment of the present application; Figure 3 is a schematic structural diagram of a signal processing system shown in an embodiment of the present application; Figure 4 is a schematic physical structure diagram of a control device shown in an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] To make the objectives, technical solutions, and advantages of the present application clearer, the following will clearly and completely describe the technical solutions in the present application in conjunction with the drawings in the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present application.

[0022] The broadband wireless signal processing method of the present application is applied to an improved signal processing system. The improved signal processing system includes a forward path and a feedback loop. The forward path at least includes a digital predistortion processing module and a power amplifier connected in sequence. The feedback loop at least includes a band-pass filter, a signal reconstruction module, and a parameter adjustment module connected in sequence. The band-pass filter is connected to the output end of the power amplifier, and the parameter adjustment module is also connected to the output end of the digital predistortion processing module.

[0023] Figure 2 is a flowchart of a broadband wireless signal processing method shown in an embodiment of the present application. Refer to Figure 2 , the broadband wireless signal processing method of the present application may specifically include: Step 101: Receive the initial baseband signal sent by the transmitting end through the digital predistortion processing module, and apply a distortion operation opposite to the nonlinear distortion of the power amplifier to the initial baseband signal to obtain a first signal; Step 102: Amplify the power of the first signal through the power amplifier to obtain a second signal.

[0024] Among them, the parameter adjustment module is used to adjust the parameters of the digital predistortion processing module according to the output signal of the digital predistortion processing module and the output signal of the signal reconstruction module. The band-pass filter is used to constrain the bandwidth of the output signal of the power amplifier. The signal reconstruction module is used to reconstruct the output signal of the band-pass filter into a signal that meets the preset conditions according to the output signal of the digital predistortion processing module.

[0025] Before the method of the present application is officially implemented, it is necessary to pre-train the signal reconstruction module and the digital predistortion processing module in advance, and then deploy these two trained modules into the improved signal processing system to obtain the system architecture as Figure 3 shown. Figure 3 is a schematic diagram of the architecture of a signal processing system shown in an embodiment of the present application.

[0026] In Figure 3 , the forward path further includes a digital-to-analog converter and a first frequency conversion module (i.e., Figure 3 the up-conversion in

[0027] The digital-to-analog converter is used to convert the output signal of the digital predistortion processing module into an analog signal.

[0028] The first frequency conversion module is used to boost the spectrum of the output signal of the digital-to-analog converter to the target radio frequency band.

[0029] The feedback loop also includes an analog-to-digital converter and a second frequency conversion module (i.e. Figure 3 The input end of the second frequency conversion module is connected to the output end of the bandpass filter, the input end of the analog-to-digital converter is connected to the output end of the second frequency conversion module, and the output end of the analog-to-digital converter is connected to the input end of the signal reconstruction module.

[0030] The analog-to-digital converter is used to convert the output signal of the second frequency conversion module into a digital signal.

[0031] The second frequency conversion module is used to restore the frequency spectrum of the output signal of the bandpass filter to the same radio frequency band as the input signal of the first frequency conversion module.

[0032] exist Figure 3 In the figure, x(n) represents the baseband signal sent by the transmitter.

[0033] When the method of the present application is formally implemented, it is assumed that the signal to be processed is the initial baseband signal sent by the transmitting end. The initial baseband signal first enters the digital predistortion processing module (i.e. Figure 3 The DPD in the digital pre-distortion processing module applies a distortion operation opposite to the nonlinear distortion of the power amplifier to the initial baseband signal to obtain a first signal. Then, the first signal enters the digital-to-analog converter (i.e. Figure 3 The DAC in the DAC is used to convert the first signal into an analog signal. Then, the converted analog signal enters the first frequency conversion module, which increases the spectrum of the analog signal to the target RF frequency band (the target RF frequency band can be set according to actual needs). Then, the signal increased to the target RF frequency band enters the power amplifier (i.e. Figure 3 The power amplifier amplifies the power of the signal and finally obtains the required second signal.

[0034] In the present application, since the power amplifier has nonlinear characteristics, the output signal will produce nonlinear distortion, and the digital pre-distortion processing module can pre-learn the nonlinear characteristics of the power amplifier and apply a distortion operation opposite to the nonlinear distortion of the power amplifier to the received signal, so that the signal output by the power amplifier can be restored to a linear state as much as possible after nonlinear processing, thereby improving the performance and signal quality of the communication system.

[0035] In the present application, a band pass filter (BPF) is provided in the feedback loop, and the band pass filter can constrain the bandwidth of the signal output by the power amplifier, thereby ensuring that the low-speed ADC can collect the output signal of the band pass filter without aliasing.

[0036] In this application, a signal reconstruction module is provided in the feedback loop. The signal reconstruction module can reconstruct the output signal of the analog-to-digital converter into a signal that meets preset conditions (the preset conditions can be set according to actual requirements). Exemplarily, the signal reconstruction module can utilize the high-speed baseband signal transmitted by the transmitter to reconstruct the low-speed feedback signal output by the analog-to-digital converter into a high-speed signal that is the same as the high-speed baseband signal transmitted by the transmitter. Here, "the same" means the same amplitude and phase, the same data rate, the same signal characteristics (coding method, modulation method), etc. The purpose of the reconstruction is to enable the reconstructed high-speed feedback signal to be compared with the high-speed baseband signal transmitted by the transmitter, so that the distortion degree can be accurately calculated, facilitating more precise adjustment of the parameters in the digital predistortion processing module in the subsequent process, and achieving effective compensation for the nonlinear distortion of the power amplifier. Therefore, in this application, by reconstructing the high-speed feedback signal using the high-speed baseband signal transmitted by the transmitter, the acquisition error caused by the low-speed ADC can be compensated, and the performance of an approximate high-speed ADC can be achieved.

[0037] In addition, since the characteristics of the power amplifier will drift over time and with changes in the operating environment, for example, due to temperature changes, aging effects, or fluctuations in operating conditions, resulting in dynamic changes in its nonlinear behavior. This characteristic drift will gradually render the pre-trained digital predistortion processing module ineffective, unable to accurately compensate for the nonlinear distortion of the power amplifier, thus affecting the linearity and overall performance of the system. Therefore, in this application, in order to ensure that the digital predistortion processing module can continuously and effectively compensate for the nonlinear distortion of the power amplifier, a parameter adjustment module is provided to adjust the parameters of the digital predistortion processing module according to the output signal of the digital predistortion processing module and the output signal of the signal reconstruction module, thereby realizing the dynamic update of the digital predistortion processing module, enabling the digital predistortion processing module to better adapt to the characteristic changes of the power amplifier during the signal processing process, maintaining the high performance and stability of the system, and meeting the requirements of modern communication systems for high linearity and high reliability.

[0038] In summary, the broadband wireless signal processing method of the present application has at least the following technical effects: First, a digital pre-distortion processing module is provided, which can apply pre-distortion operations to the initial baseband signal, so that the signal output by the power amplifier can be restored to a linear state as much as possible after non-linear processing, improving the performance of the communication system and the signal quality; Second, a reconstruction module is provided, which can reconstruct a high-speed feedback signal from the high-speed baseband signal sent by the sending end, make up for the acquisition error caused by the low-speed ADC, achieve the performance of an approximate high-speed ADC, and avoid the defect of high hardware cost when using an ADC with a high sampling rate in the related art; Third, a parameter adjustment module is provided, which can adjust the parameters of the digital pre-distortion processing module according to the output signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module, realize the dynamic update of the digital pre-distortion processing module, enable the digital pre-distortion processing module to adapt to the characteristic changes of the power amplifier, accurately compensate for the non-linear distortion of the power amplifier, and improve the linearity and overall performance of the system.

[0039] In combination with the above embodiments, in one implementation, the parameter adjustment module is also connected to the input end of the digital pre-distortion processing module, and adjusts the parameters of the digital pre-distortion processing module according to the output signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module. Specifically, it may include: Determine the similarity between the input signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module; If the similarity is less than a preset threshold, determine a new parameter value of the digital pre-distortion processing module according to the output signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module; Input the new parameter value into the digital pre-distortion processing module, so that the digital pre-distortion processing module processes the signal sent by the sending end according to the new parameter value.

[0040] Among them, determining the similarity between the input signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module includes: Determine the normalized mean square error between the input signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module; Determine the similarity according to the normalized mean square error.

[0041] In the present application, after the digital pre-distortion processing module is officially put into use, the parameter adjustment module can monitor the similarity between the input signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module in real time. Once it is detected that the similarity is less than the preset threshold, the parameters of the digital pre-distortion processing module are updated according to the output signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module.

[0042] Among them, the normalized mean square error is inversely proportional to the similarity. The smaller the normalized mean square error, the higher the similarity; the larger the normalized mean square error, the lower the similarity. Therefore, in actual implementation, when it is determined that the normalized mean square error is greater than the preset error value, it can be determined that the similarity is less than the preset threshold.

[0043] This application only updates the parameters of the digital predistortion processing module when the similarity is less than the preset threshold, which can effectively save the system computing resources.

[0044] Among them, in addition to using the normalized mean square error to determine the similarity, other methods can also be used, and this embodiment does not make specific limitations on this.

[0045] Combined with the above embodiments, in one implementation manner, the signal reconstruction module is obtained in advance through the following steps: Obtain the baseband signal sent by the transmitter and the output signal of the analog-to-digital converter; Taking the output signal of the analog-to-digital converter as the input sample data, taking the baseband signal sent by the transmitter as the output sample data, and aiming to minimize the difference between the output signal of the analog-to-digital converter and the baseband signal sent by the transmitter, train the first neural network model to obtain the signal reconstruction module.

[0046] Among them, the first neural network model is an autoencoder, and the autoencoder includes an encoder and a decoder. Taking the output signal of the analog-to-digital converter as the input sample data, taking the baseband signal sent by the transmitter as the output sample data, and aiming to minimize the difference between the output signal of the analog-to-digital converter and the baseband signal sent by the transmitter, training the first neural network model includes: Randomly initialize the weights and biases in the neural network layer of the autoencoder; Input the output signal of the analog-to-digital converter into the encoder to obtain the encoded features; Input the encoded features into the decoder to obtain the decoded reconstructed features; Determine the difference value between the reconstructed features and the features of the baseband signal sent by the transmitter through the first loss function; Based on the first difference value, determine the gradients corresponding to the weights and biases in the autoencoder through the backpropagation algorithm; Update the weights and biases in the autoencoder through the stochastic gradient descent algorithm; Aiming to minimize the first difference value, repeat the above steps until the first preset stop condition is met. The autoencoder when the first preset stop condition is met is the signal reconstruction module.

[0047] In the initial stage of this application, when the digital pre-distortion processing module has not been obtained yet, the baseband signal sent by the transmitter directly passes through the DAC, the first frequency conversion module, and the power amplifier. The feedback loop feeds back the output signal of the power amplifier. The feedback signal first passes through a band-pass filter, and the band-pass filter restricts the bandwidth of the output signal of the power amplifier to ensure that the low-speed ADC can sample the output of the band-pass filter without aliasing. The filtered band-pass signal is sampled by the second frequency conversion module and the low-speed ADC to obtain a low-speed feedback signal. This application can use the low-speed feedback signal and the high-speed baseband signal sent by the transmitter to train and obtain a signal reconstruction module.

[0048] Specifically, during the training process, after obtaining the high-speed baseband signal, first sample the high-speed baseband signal to obtain a low-speed baseband signal that is consistent with the low-speed feedback signal; then use the two low-speed baseband signals as the input and output of the first neural network model for training respectively, and the successfully trained first neural network is the signal reconstruction module. When the signal reconstruction module is put into Figure 3 the system architecture shown for use, input the low-speed feedback signal into the signal reconstruction module, and a reconstructed high-speed feedback signal that is the same as the high-speed baseband signal can be obtained.

[0049] In this application, a multi-layer neural network can be constructed as an encoder, and the number of nodes in its input layer is the same as the dimension of the low-speed feedback signal. Multiple hidden layers can be set in the middle layer, and the number of nodes in the hidden layer can be adjusted according to the computing resources. The role of the encoder is to map the low-speed feedback signal to a low-dimensional feature space, that is, to learn the effective feature representation of the low-speed signal.

[0050] In this application, a multi-layer neural network can be constructed as a decoder, and the number of nodes in its input layer is the same as the feature dimension output by the encoder. The number of nodes in the output layer of the decoder is the same as the dimension of the high-speed baseband signal. The decoder is used to restore the low-dimensional feature representation output by the encoder to the high-speed baseband signal.

[0051] Among them, the first loss function can be the Mean Squared Error (MSE), or the Normalized Mean Squared Error (NMSE). The first preset stop condition can be that the first difference value is less than the preset difference value, or the number of training rounds is greater than the preset number of times.

[0052] Of course, in addition to using an autoencoder to train and obtain a signal reconstruction module, other types of neural networks can also be used to train and obtain a signal reconstruction module, and this application does not make specific restrictions on this.

[0053] In this application, since the characteristics of the power amplifier have little impact on the reconstruction performance when the sampling rate is fixed, after obtaining the signal reconstruction module, there is no need to train the signal reconstruction module again subsequently.

[0054] Combined with the above embodiments, in one implementation, the digital pre-distortion processing module is obtained in advance through the following steps: Obtain the baseband signal sent by the transmitter and the output signal of the signal reconstruction module; Using the output signal of the signal reconstruction module as the input sample data, the baseband signal sent by the transmitter as the output sample data, and aiming to minimize the difference between the output signal of the signal reconstruction module and the baseband signal sent by the transmitter, train the second neural network model to obtain the digital pre-distortion processing module.

[0055] Among them, the second neural network model is a Multilayer Perceptron (MLP). The Multilayer Perceptron includes an input layer, multiple hidden layers, and an output layer. Using the output signal of the signal reconstruction module as the input sample data, the baseband signal sent by the transmitter as the output sample data, and aiming to minimize the difference between the output signal of the signal reconstruction module and the baseband signal sent by the transmitter, training the second neural network model includes: Randomly initialize the weights and biases in the Multilayer Perceptron; Input the output signal of the signal reconstruction module into the Multilayer Perceptron to obtain the predicted baseband signal; Determine the second difference value between the predicted baseband signal and the baseband signal sent by the transmitter through the second loss function; Based on the second difference value, determine the gradients corresponding to the weights and biases in the Multilayer Perceptron through the backpropagation algorithm; Update the weights and biases in the Multilayer Perceptron through the stochastic gradient descent algorithm; Aiming to minimize the second difference value, repeat the above steps until the second preset stop condition is met. The Multilayer Perceptron when the second preset stop condition is met is the digital pre-distortion processing module.

[0056] In this application, the number of nodes in the input layer of the Multilayer Perceptron is the same as the dimension of the feedback signal and is used to receive the feedback signal. The number of hidden layers and the number of nodes in each layer can be set according to requirements. The number of nodes in the output layer is the same as the dimension of the baseband signal and is used to output the predicted baseband signal.

[0057] Among them, the second loss function can be the mean square error or the normalized mean square error. The second preset stop condition can be that the second difference value is less than the preset difference value, or the number of training epochs is greater than the preset number.

[0058] In addition to using a multi-layer perceptron network to train the digital pre-distortion processing module, other neural networks can also be used in this application, such as a Convolutional Neural Network (CNN), a Recurrent Neural Network (RNN), etc., to train the digital pre-distortion processing module, which can be specifically set according to actual requirements.

[0059] In summary, the broadband wireless signal processing method of this application has at least the following technical effects: (1) The larger the bandwidth of the input signal of the power amplifier, the higher the requirement for the ADC sampling rate in the feedback loop. The price of the ADC increases sharply with the increase of the sampling rate, resulting in a sharp increase in hardware costs. Secondly, for ultra-wideband signals, existing ADC products are difficult to meet the requirements of the Nyquist sampling rate. In this application, a signal reconstruction module is added to the feedback loop to reconstruct a high-speed feedback signal from the high-speed baseband signal at the transmitting end, which can compensate for the acquisition error caused by the low-speed ADC and achieve the performance of an approximate high-speed ADC, greatly alleviating the demand for high-speed ADCs and providing strong support for the application of digital pre-distortion technology in large-bandwidth signals.

[0060] (2) Since the offline training method cannot adapt to the dynamic characteristics of the power amplifier. In this application, the digital pre-distortion processing module is iteratively trained in the feedback loop. According to the dynamic changes of the non-linear behavior of the power amplifier, the parameters in the digital pre-distortion processing module are dynamically adjusted to maintain the high performance and stability of the system, thus meeting the requirements of modern communication systems for high linearity and high reliability.

[0061] (3) When dynamically updating the parameters of the digital pre-distortion processing module, a measurement index is introduced. By calculating the similarity between the baseband signal transmitted at the transmitting end and the reconstructed signal, the digital pre-distortion processing module is updated only when the similarity is less than a preset threshold, which can effectively save resources.

[0062] This application also provides a signal processing system, including a forward path and a feedback loop. The forward path at least includes a digital pre-distortion processing module and a power amplifier connected in sequence. The feedback loop at least includes a band-pass filter, a signal reconstruction module, and a parameter adjustment module connected in sequence. The band-pass filter is connected to the output end of the power amplifier, and the parameter adjustment module is also connected to the output end of the digital pre-distortion processing module; The digital pre-distortion processing module is configured to receive an initial baseband signal transmitted by a transmitting end through the digital pre-distortion processing module, and apply a distortion operation opposite to the non-linear distortion of the power amplifier to the initial baseband signal to obtain a first signal; The power amplifier is used to amplify the power of the first signal through the power amplifier to obtain a second signal; The parameter adjustment module is used to adjust the parameters of the digital predistortion processing module according to the output signal of the digital predistortion processing module and the output signal of the signal reconstruction module; The band-pass filter is used to constrain the bandwidth of the output signal of the power amplifier; The signal reconstruction module is used to reconstruct the output signal of the band-pass filter into a signal that meets the preset conditions according to the output signal of the digital predistortion processing module.

[0063] The description of the use of this signal processing system can be referred to the foregoing, and this application will not elaborate herein.

[0064] This application also provides a control device for controlling Figure 3 the operation of the signal processing system shown. Figure 4 It is a schematic diagram of the physical structure of a control device shown in an embodiment of this application. As Figure 4 shown, the control device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440. Among them, the processor 410, the communication interface 420, and the memory 430 communicate with each other through the communication bus 440. The processor 410 can call the logical instructions in the memory 430 to execute the broadband wireless signal processing method of this application. The method includes: Receiving an initial baseband signal sent by a sending end through the digital predistortion processing module, and applying a distortion operation opposite to the non-linear distortion of the power amplifier to the initial baseband signal to obtain a first signal; Amplifying the power of the first signal through the power amplifier to obtain a second signal; Among them, the parameter adjustment module is used to adjust the parameters of the digital predistortion processing module according to the output signal of the digital predistortion processing module and the output signal of the signal reconstruction module. The band-pass filter is used to constrain the bandwidth of the output signal of the power amplifier. The signal reconstruction module is used to reconstruct the output signal of the band-pass filter into a signal that meets the preset conditions according to the output signal of the digital predistortion processing module.

[0065] In addition, when the logical instructions in the above-mentioned memory 430 can be implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0066] On the other hand, this application also provides a computer program product. The computer program product includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute a broadband wireless signal processing method provided by the above-mentioned various methods. This method includes: Receiving, by the digital pre-distortion processing module, an initial baseband signal sent by a sending end, and applying a distortion operation opposite to the non-linear distortion of the power amplifier to the initial baseband signal to obtain a first signal; Amplifying, by the power amplifier, the power of the first signal to obtain a second signal; Wherein, the parameter adjustment module is used to adjust the parameters of the digital pre-distortion processing module according to the output signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module. The band-pass filter is used to constrain the bandwidth of the output signal of the power amplifier. The signal reconstruction module is used to reconstruct the output signal of the band-pass filter into a signal that meets the preset conditions according to the output signal of the digital pre-distortion processing module.

[0067] In yet another aspect, this application also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it realizes the execution of a broadband wireless signal processing method provided by the above-mentioned various methods. This method includes: Receiving, by the digital pre-distortion processing module, an initial baseband signal sent by a sending end, and applying a distortion operation opposite to the non-linear distortion of the power amplifier to the initial baseband signal to obtain a first signal; Amplifying, by the power amplifier, the power of the first signal to obtain a second signal; Among them, the parameter adjustment module is used to adjust the parameters of the digital predistortion processing module according to the output signal of the digital predistortion processing module and the output signal of the signal reconstruction module. The band-pass filter is used to constrain the bandwidth of the output signal of the power amplifier. The signal reconstruction module is used to reconstruct the output signal of the band-pass filter into a signal that meets the preset conditions according to the output signal of the digital predistortion processing module.

[0068] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0069] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. And these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present application.

Claims

1. A broadband wireless signal processing method, characterized in that: Applied to a signal processing system, the signal processing system comprises a forward path and a feedback loop, the forward path comprises at least a digital pre-distortion processing module and a power amplifier connected in sequence, the feedback loop comprises at least a band-pass filter, a signal reconstruction module and a parameter adjustment module connected in sequence, the band-pass filter is connected to the output end of the power amplifier, and the parameter adjustment module is also connected to the output end of the digital pre-distortion processing module; the signal processing method comprises: Receiving an initial baseband signal sent by a transmitting end through the digital predistortion processing module, applying a distortion operation opposite to the nonlinear distortion of the power amplifier to the initial baseband signal, to obtain a first signal; amplifying the power of the first signal by the power amplifier to obtain a second signal; Among them, the parameter adjustment module is used to adjust the parameters of the digital pre-distortion processing module according to the output signal of the digital pre-distortion processing module and the output signal of the signal reconstruction module, the bandpass filter is used to constrain the bandwidth of the output signal of the power amplifier, and the signal reconstruction module is used to reconstruct the output signal of the bandpass filter into a signal that meets preset conditions according to the output signal of the digital pre-distortion processing module.

2. The broadband wireless signal processing method according to claim 1, characterized in that: The parameter adjustment module is also connected to the input end of the digital predistortion processing module, and the parameters of the digital predistortion processing module are adjusted according to the output signal of the digital predistortion processing module and the output signal of the signal reconstruction module, including: Determining a similarity between an input signal of the digital predistortion processing module and an output signal of the signal reconstruction module; If the similarity is less than a preset threshold, determining a new parameter value of the digital predistortion processing module according to the output signal of the digital predistortion processing module and the output signal of the signal reconstruction module; The new parameter value is input into the digital predistortion processing module, so that the digital predistortion processing module processes the signal sent by the sending end according to the new parameter value.

3. The broadband wireless signal processing method according to claim 1, characterized in that: The determining the similarity between the input signal of the digital predistortion processing module and the output signal of the signal reconstruction module comprises: Determining a normalized mean square error between an input signal of the digital predistortion processing module and an output signal of the signal reconstruction module; The similarity is determined according to the normalized mean square error.

4. The broadband wireless signal processing method according to claim 1, characterized in that: The forward path also includes a digital-to-analog converter and a first frequency conversion module, the input end of the digital-to-analog converter is connected to the output end of the digital pre-distortion processing module, the input end of the first frequency conversion module is connected to the output end of the digital-to-analog converter, the output end of the first frequency conversion module is connected to the input end of the power amplifier, the digital-to-analog converter is used to convert the output signal of the digital pre-distortion processing module into an analog signal, and the first frequency conversion module is used to increase the frequency spectrum of the output signal of the digital-to-analog converter to a target radio frequency band; The feedback loop also includes an analog-to-digital converter and a second frequency conversion module, the input end of the second frequency conversion module is connected to the output end of the bandpass filter, the input end of the analog-to-digital converter is connected to the output end of the second frequency conversion module, the output end of the analog-to-digital converter is connected to the input end of the signal reconstruction module, the analog-to-digital converter is used to convert the output signal of the second frequency conversion module into a digital signal, and the second frequency conversion module is used to restore the spectrum of the output signal of the bandpass filter to the same radio frequency band as the input signal of the first frequency conversion module.

5. The broadband wireless signal processing method according to claim 4, characterized in that: The signal reconstruction module is obtained in advance through the following steps: Acquire a baseband signal sent by the transmitting end and an output signal of the analog-to-digital converter; The output signal of the analog-to-digital converter is used as input sample data, the baseband signal sent by the transmitting end is used as output sample data, and the first neural network model is trained with the goal of minimizing the difference between the output signal of the analog-to-digital converter and the baseband signal sent by the transmitting end to obtain the signal reconstruction module.

6. The broadband wireless signal processing method according to claim 5, characterized in that: The first neural network model is an autoencoder, the autoencoder includes an encoder and a decoder, the output signal of the analog-to-digital converter is used as input sample data, the baseband signal sent by the transmitting end is used as output sample data, and the difference between the output signal of the analog-to-digital converter and the baseband signal sent by the transmitting end is minimized, and the first neural network model is trained, including: Randomly initializing weights and biases in a neural network layer of the autoencoder; Inputting the output signal of the analog-to-digital converter into the encoder to obtain a coding feature; Inputting the encoded features into the decoder to obtain decoded reconstructed features; Determining a difference value between the reconstructed feature and the feature of the baseband signal sent by the transmitting end through a first loss function; Based on the first difference value, determining the gradients corresponding to the weights and biases in the autoencoder by a back propagation algorithm; Updating the weights and biases in the autoencoder by a stochastic gradient descent algorithm; With the goal of minimizing the first difference value, the above steps are repeated until a first preset stop condition is met, and the autoencoder when the first preset stop condition is met is the signal reconstruction module.

7. The broadband wireless signal processing method according to claim 4, characterized in that: The digital pre-distortion processing module is obtained in advance through the following steps: Acquire a baseband signal sent by the transmitting end and an output signal of the signal reconstruction module; The output signal of the signal reconstruction module is used as input sample data, the baseband signal sent by the transmitting end is used as output sample data, and the second neural network model is trained with the goal of minimizing the difference between the output signal of the signal reconstruction module and the baseband signal sent by the transmitting end to obtain the digital predistortion processing module.

8. The broadband wireless signal processing method according to claim 7, characterized in that: The second neural network model is a multi-layer perceptron network, which includes an input layer, multiple hidden layers, and an output layer. The output signal of the signal reconstruction module is used as input sample data, the baseband signal sent by the transmitting end is used as output sample data, and the difference between the output signal of the signal reconstruction module and the baseband signal sent by the transmitting end is minimized. The second neural network model is trained, including: Randomly initializing weights and biases in the multilayer perceptron network; Inputting the output signal of the signal reconstruction module into the multi-layer perceptron network to obtain a predicted baseband signal; Determining a second difference value between the predicted baseband signal and the baseband signal sent by the transmitting end by using a second loss function; Based on the second difference value, determining the gradients corresponding to the weights and biases in the multilayer perceptron network by a back propagation algorithm; Updating weights and biases in the multilayer perceptron network by a stochastic gradient descent algorithm; With the goal of minimizing the second difference value, the above steps are repeated until a second preset stop condition is met, and the multilayer perceptron network when the second preset stop condition is met is the digital predistortion processing module.

9. A signal processing system, characterized in that: The method comprises a forward path and a feedback loop, wherein the forward path at least comprises a digital pre-distortion processing module and a power amplifier connected in sequence, and the feedback loop at least comprises a band-pass filter, a signal reconstruction module and a parameter adjustment module connected in sequence, wherein the band-pass filter is connected to the output end of the power amplifier, and the parameter adjustment module is also connected to the output end of the digital pre-distortion processing module; The digital pre-distortion processing module is used to receive an initial baseband signal sent by a transmitting end through the digital pre-distortion processing module, apply a distortion operation opposite to the nonlinear distortion of the power amplifier to the initial baseband signal, and obtain a first signal; The power amplifier is used to amplify the power of the first signal through the power amplifier to obtain a second signal; The parameter adjustment module is used to adjust the parameters of the digital predistortion processing module according to the output signal of the digital predistortion processing module and the output signal of the signal reconstruction module; The bandpass filter is used to constrain the bandwidth of the output signal of the power amplifier; The signal reconstruction module is used to reconstruct the output signal of the bandpass filter into a signal that meets preset conditions according to the output signal of the digital predistortion processing module.

10. A control device, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the broadband wireless signal processing method according to any one of claims 1 to 8 is implemented.

Images