A multi-parameter adjustable wireless transmission signal simulation system and method

Abstract

The present invention provides a multi-parameter adjustable wireless transmission signal simulation system and method, which relates to the field of wireless communication technologies. The present invention includes a main control module, a signal source module, a signal processing module, a combining module, a digital pre-distortion module and an adaptive adjustment module; by controlling the signal source module to generate a carrier wave and a modulation signal, processing the carrier wave signal and the modulation signal through the signal processing module to output a direct signal and a multipath signal, compensating for the non-linear distortion of the signal through the digital pre-distortion module, synthesizing a combined signal through the combining module, collecting the characteristics of the combined signal through the adaptive adjustment module, and dynamically optimizing the multipath parameters by using the RLS algorithm and updating the pre-distortion coefficient; it further includes a machine learning prediction module for predicting channel characteristic parameters and performing real-time parameter adjustment. This application solves the problems of redundant discrete devices, low parameter adjustment accuracy and signal distortion, and is applicable to dynamic channel simulation and complex scenario modeling.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technologies, and more particularly, to a multi-parameter adjustable wireless transmission signal simulation system and method. Background Art

[0002] In the field of wireless communication technologies, the transmission characteristics of signals have an important impact on the performance of communication systems. In an actual wireless communication environment, the received signal is usually formed by the superposition of a direct signal and multipath signals. In order to test and optimize the performance of communication systems, it is necessary to accurately simulate these signals and their interactions. However, the systems used to simulate wireless transmission signals in the prior art have many deficiencies. Traditional solutions often use multiple discrete devices (such as direct digital frequency synthesizer DDS modules) to implement signal generation and processing, resulting in high system complexity, serious waste of hardware resources, and cumulative synchronization errors. In addition, the existing systems are insufficient in terms of parameter adjustment accuracy. For example, the delay error usually exceeds 20 ns, and the phase error exceeds 15°, making it difficult to meet the high-precision test requirements. At the same time, problems such as unstable waveforms, frequency drift exceeding ±5 MHz, and high output distortion often occur during the signal synthesis process. Especially in high-frequency band applications, the frequency band of the combined signal is limited and the high-frequency harmonic suppression is insufficient, and the waveform distortion degree is often greater than 3%. The prior art also has the problem of poor dynamic adjustment ability, with rough parameter adjustment steps (such as a delay step of 50 ns), which cannot meet the fine test requirements.

[0003] In view of this, after studying the existing technologies, the applicant, aiming at the problems of redundant discrete devices, insufficient parameter adjustment accuracy, and poor output stability in the existing wireless communication signal simulation systems, specifically proposes this application. Summary of the Invention

[0004] The present invention aims to provide a multi-parameter adjustable wireless transmission signal simulation system and method to solve the problems of redundant discrete devices, insufficient parameter adjustment accuracy, and signal distortion in the existing wireless transmission signal simulation systems in the prior art.

[0005] To solve the above technical problems, the present invention is achieved through the following technical solutions:

[0006] A multi-parameter adjustable wireless transmission signal simulation system includes:

[0007] A main control module, connected to a signal source module through SPI and AXI buses, for controlling system parameter settings, error compensation, and real-time parameter adjustment;

[0008] A signal source module, for generating a carrier signal and a modulation signal;

[0009] A signal processing module, adaptively connected to the signal source module and the digital pre-distortion module, is configured to process a carrier signal and a modulation signal to output a direct signal and a multipath signal, and compensate for the non-linear distortion of the direct signal and the multipath signal through the digital pre-distortion module;

[0010] A combining module, adaptively connected to the signal processing module, is configured to combine the compensated direct signal and multipath signal into a combined signal;

[0011] An adaptive adjustment module, adaptively connected to the combining module, is configured to collect the characteristics of the combined signal, dynamically optimize the multipath parameters of the combined signal by using a recursive least squares adaptive algorithm, and update the pre-distortion coefficient of the digital pre-distortion module; the optimized combined signal is used for wireless transmission.

[0012] Preferably, the digital pre-distortion module compensates for the non-linear distortion of the signal through a high-order digital pre-distortion function, and the high-order digital pre-distortion function is expressed as:

[0013] ；

[0014] Wherein, is the baseband signal after pre-distortion processing; is the original baseband signal; is the order control parameter for odd-order non-linear compensation; is the order control parameter for even-order non-linear terms; is the odd-power term of the original baseband signal; is the even-power term of the original baseband signal; is the th-order pre-distortion coefficient of the odd-power term, optimized by a genetic algorithm; is the th-order pre-distortion coefficient of the even-power term; is the number of kernels of the Volterra series; is the th weight coefficient of the th Volterra kernel; is the th Volterra kernel function, used to describe the non-linear response of the original baseband signal at time delay is the convolution integral, representing the time-domain interaction between the original baseband signal and the kernel function; is the signal time delay value; t represents the signal time variable; k is the order variable.

[0015] Preferably, it includes at least one of time delay, phase and attenuation value; the recursive least squares adaptive algorithm is a recursive least squares method combined with a dynamic step size and sparse regularization, and its expression is:

[0016] ；

[0017] ；

[0018] Among them, is the filter weight vector at the th iteration; is the filter weight vector at the th iteration; is the Kalman gain matrix; is the th iteration error signal; is the dynamic step size factor; is the maximum value of the step size; is the step size attenuation coefficient, ; is the sparse regularization coefficient, ; is the sign function of the weight vector.

[0019] Preferably, it further includes a machine learning prediction module, integrating the TensorFlow Lite framework, adaptively connected to the adaptive adjustment module and the main control module, for inputting the signal after adaptive multipath adjustment into a pre-trained CNN model to output predicted channel characteristic parameters, and performing real-time adjustment through the main control module.

[0020] Preferably, the channel characteristic parameters at least include one of time delay, phase offset, and predistortion coefficient; the pre-trained CNN model is a convolutional neural network combining a spatio-temporal fusion network and a long short-term memory network and its expression is:

[0021] ;

[0022] ;

[0023] Among them, is the predicted time delay parameter at time ; is the spatio-temporal fusion network; represents the output signal matrix at time ; is the prediction time window; is the channel historical state; is the long short-term memory network, is the number of historical time steps, is the time step variable.

[0024] Preferably, it further includes: optimizing and updating the parameters of the high-order digital predistortion function by minimizing the predistortion coefficient, so that the finally output combined signal meets the set requirements, and its expression is:

[0025] ;

[0026] Wherein, is the error vector magnitude; is the amplitude of the th harmonic; N represents the harmonic order control parameter.

[0027] Preferably, the system parameters include carrier frequency, signal amplitude, modulation degree, time delay and phase offset; the main control module adopts an STM32H743 single-chip microcomputer and a Xilinx Zynq UltraScale+ MPSoC chip, and sets the carrier frequency, modulation parameters and phase offset through the SPI interface; runs the CNN model of the machine learning prediction module through the Xilinx Zynq UltraScale+ MPSoC chip; communicates with the STM32H743 single-chip microcomputer through the AXI bus to realize error compensation and parameter real-time adjustment.

[0028] Preferably, the signal source module generates a carrier signal and a modulation signal based on an ADRV9009 broadband radio frequency transceiver, and outputs them after being processed by the signal processing module.

[0029] Preferably, the signal processing module includes an ADL5380 quadrature demodulator, an ADL5375 modulator and a MAX20343 broadband amplifier;

[0030] The ADL5380 quadrature demodulator is used to demodulate the input signal into an I / Q baseband signal;

[0031] The I / Q baseband signal is subjected to nonlinear distortion compensation through the digital predistortion module;

[0032] The ADL5375 modulator is used to modulate the I / Q baseband signal after nonlinear distortion compensation into an intermediate frequency signal;

[0033] The MAX20343 broadband amplifier is used to amplify the intermediate frequency signal and then output it to compensate for the link loss.

[0034] Preferably, the adaptive adjustment module acquires the characteristics of the combined signal through a high-speed ADC; the combining module uses an ADAR1000 beamforming chip to realize the dynamic synthesis of multiple signals.

[0035] The present invention also provides a multi-parameter adjustable wireless transmission signal simulation method, which is applied to a multi-parameter adjustable wireless transmission signal simulation system as described above, and includes:

[0036] S1, the main control module controls the signal source module to generate a carrier signal and a modulation signal;

[0037] S2, the signal processing module performs signal modulation processing on the carrier signal and the modulation signal, and uses the digital pre-distortion module to compensate for the non-linear distortion of the signal, and outputs a direct signal and a multipath signal;

[0038] S3, the combining module dynamically combines the direct signal and the multipath signal, and outputs a combined signal;

[0039] S4, the adaptive adjustment module collects the characteristics of the combined signal, and dynamically optimizes the multipath parameters of the combined signal based on the recursive least squares adaptive algorithm, and the optimized combined signal is used for wireless transmission.

[0040] Preferably, it further includes: S5, inputting the optimized combined signal into the pre-trained CNN model of the machine learning prediction module for prediction, and outputting the predicted channel characteristic parameters for real-time adjustment through the main control module.

[0041] In summary, compared with the prior art, the present invention has the following beneficial effects:

[0042] The present invention can realize the full-process integration of signal generation, processing and synthesis, adopts a high-integration design, reduces the redundancy of discrete devices, reduces the deployment cost (for example, compared with the traditional multi-DDS chip solution, the deployment cost is reduced by 50%), and adapts to a compact test environment.

[0043] The present invention jointly optimizes and adjusts the multipath signal parameters through the recursive least squares adaptive algorithm (RLS algorithm) and Kalman filtering, improves the accuracy of the multipath signal, so that the multipath delay error ≤ 2 ns, the phase error ≤ 3°, and the delay adjustment step resolution is 0.1 ns, meeting the stringent requirements of 5G millimeter-wave communication for timing synchronization. At the same time, the RLS algorithm combines a dynamic step and sparse regularization for improvement, accelerating the convergence speed of the RLS algorithm, with a 50% increase in the convergence speed compared to the traditional LMS algorithm and a 30% reduction in the steady-state error.

[0044] The present invention adopts a pre-trained CNN model based on the fusion of the ST-Transformer spatio-temporal fusion network and LSTM, realizes the functions of real-time and advanced compensation for channel prediction, improves the dynamic adaptability of the system, reduces the time-delay change of real-time predicted channels, reduces the parameter advanced adjustment error, and supports dynamic channel tracking in high-speed mobile scenarios of 120 km / h (such as vehicle communication, drone links).

[0045] The closed-loop feedback of the high-speed ADC AD9208 based on the high-order digital predistortion function optimized by the genetic algorithm and the adaptive adjustment module improves the accuracy of the direct signal, making the amplitude error of the direct signal ≤5mV and the frequency resolution reach 0.1Hz.

[0046] The present invention reduces the waveform distortion degree and improves the numerical value of harmonic suppression through the multi-order Volterra series predistortion function and the genetic optimization algorithm. Brief Description of the Drawings

[0047] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0048] Figure 1 It is the overall architecture diagram of a multi-parameter adjustable wireless transmission signal simulation system provided by Embodiment 1 of the present invention.

[0049] Figure 2 It is the schematic diagram of the processing flow of the combining module provided by Embodiment 1 of the present invention.

[0050] Figure 3 It is the network structure architecture diagram of the pre-trained CNN model provided by Embodiment 1 of the present invention.

[0051] Figure 4 It is the working principle architecture diagram of the high-order predistortion module provided by Embodiment 1 of the present invention.

[0052] Figure 5 It is the schematic diagram of the process of a multi-parameter adjustable wireless transmission signal simulation method provided by Embodiment 2 of the present invention.

[0053] The present invention will be further described in detail below with reference to the drawings and specific embodiments. Detailed Embodiments

[0054] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed present invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0055] Embodiment 1

[0056] As Figure 1 shown, a multi-parameter adjustable wireless transmission signal simulation system includes a main control module, a signal source module, a signal processing module, a combining module, a digital pre-distortion module, an adaptive adjustment module, a machine learning prediction module, and a human-machine interface.

[0057] The main control module is connected to control the signal source module through the SPI and AXI buses, and is used for controlling system parameter settings, error compensation, and real-time parameter adjustment.

[0058] Specifically, the system parameters include carrier frequency, signal amplitude, modulation degree, time delay, and phase offset, etc.

[0059] In this embodiment, the main control module adopts an STM32H743 single-chip microcomputer and a Xilinx Zynq UltraScale+ MPSoC chip. The carrier frequency (such as 30 MHz to 6 GHz), modulation parameters (such as AM / FM / PM / QAM), and phase offset are set through the SPI interface; the CNN model of the machine learning prediction module is run through the Xilinx Zynq UltraScale+ MPSoC chip; error compensation and real-time parameter adjustment are realized through communication between the AXI bus and the STM32H743 single-chip microcomputer, so that the inference delay < 0.1 ms.

[0060] The signal source module is used to generate a carrier signal and a modulation signal.

[0061] In this embodiment, the signal source module generates a carrier signal and a modulation signal based on the ADRV9009 broadband radio frequency transceiver, and outputs them after being processed by the signal processing module, that is, the ADL5375 quadrature modulator of the signal processing module loads them onto the carrier to output I / Q two-way radio frequency signals. The ADRV9009 supports multiple modulation methods such as AM, FM, PM, and QAM. The carrier frequency covers 30 MHz to 6 GHz, the frequency resolution reaches 0.1 Hz, and it supports the dynamic frequency scanning mode with a frequency scanning rate of 10 MHz / μs.

[0062] This embodiment adopts a highly integrated architecture of the ADRV9009 broadband radio frequency transceiver and the Xilinx Zynq UltraScale+ MPSoC chip, which reduces the system volume by 60% compared with the traditional discrete solution, reduces the power consumption by 40%, and improves the hardware resource utilization rate by 50%.

[0063] The signal processing module is adaptively connected to the signal source module and the digital predistortion module, and is used to process the carrier signal and the modulation signal to output the direct signal and the multipath signal, and compensate for the nonlinear distortion of the direct signal and the multipath signal through the digital predistortion module.

[0064] In this embodiment, the signal processing module includes an ADL5380 quadrature demodulator, an ADL5375 modulator, and a MAX20343 broadband amplifier;

[0065] The ADL5380 quadrature demodulator is used to demodulate the input signal into I / Q baseband signals;

[0066] The I / Q baseband signals are compensated for nonlinear distortion through the digital predistortion module;

[0067] The ADL5375 modulator is used to modulate the I / Q baseband signals after nonlinear distortion compensation into intermediate frequency signals;

[0068] The MAX20343 broadband amplifier is used to amplify the intermediate frequency signals and then output them to compensate for the link loss.

[0069] The combining module is adaptively connected to the signal processing module and is used to combine the compensated direct signal and multipath signal into a combined signal.

[0070] In this embodiment, the combining module uses the ADAR1000 beamforming chip to realize the dynamic synthesis of multiple signals. As Figure 2As shown, a method of adding 8 independent multipath processing channels can be adopted during the multipath channel processing. Each channel integrates an ADAR1000 beamforming chip, which supports dynamic directivity adjustment (azimuth angle ±60°, elevation angle ±30°), expanding the time delay adjustment range to 0 - 500 ns with a step resolution of 0.1 ns, and the phase adjustment range is 0° - 360° with a resolution of 0.01°.

[0071] Eight multipath signals and the direct signal are synthesized by the ADAR1000 chip, and the output end is connected to a Mini-Circuits LFCN-8000 low-pass filter (cut-off frequency 6 GHz, in-band ripple < 0.2 dB). The synthesized signal is amplified by an OPA847 operational amplifier (gain 20 dB, bandwidth 4 GHz) and then output to the test equipment. The output frequency band of the combining module covers 30 MHz - 6 GHz, the frequency error ≤ 0.1%, the waveform distortion < 0.3%, and it can simulate 5G, LTE, and Wi-Fi hybrid network scenarios simultaneously, meeting the test requirements of heterogeneous communication systems.

[0072] An adaptive adjustment module is adaptively connected to the combining module, used to collect the characteristics of the combined signal, and adopt the recursive least squares (RLS) adaptive algorithm to dynamically optimize the multipath parameters (such as time delay, phase, attenuation value) of the combined signal, with an adjustment period < 1 ms; and update the coefficients of the digital pre-distortion module; the optimized combined signal is used for wireless transmission.

[0073] In this embodiment, the adaptive adjustment module collects the characteristics of the combined signal through a high-speed ADC.

[0074] Specifically, the recursive least squares adaptive algorithm is an improved recursive least squares algorithm, which combines dynamic step size and sparse regularization, and the expression is:

[0075] ；

[0076] ；

[0077] Among them, is the filter weight vector at the th iteration; is the filter weight vector at the th iteration; is the Kalman gain matrix; is the error signal at the th iteration; is the dynamic step size factor; is the maximum value of the step size; is the step size attenuation coefficient, ; is the sparse regularization coefficient, ; is the sign function of the weight vector.

[0078] Combined with the Kalman filter, the embedded RLS algorithm can dynamically optimize the multipath parameters, making the update period of the RLS algorithm < 1 ms and the convergence speed 50% higher than that of the traditional LMS. The Kalman filter suppresses the measurement noise, reducing the steady-state error by 30%. When updating the parameters, the data can be transmitted through the DMA channel to ensure real-time performance.

[0079] Specifically, the digital predistortion module compensates for the nonlinear distortion of the signal through a high-order digital predistortion function, and the high-order digital predistortion function is expressed as:

[0080] ；

[0081] where is the baseband signal after predistortion processing; is the original baseband signal; is the order control parameter for odd-order nonlinear compensation; is the order control parameter for even-order nonlinear terms; is the odd-power term of the original baseband signal; is the even-power term of the original baseband signal; is the order predistortion coefficient of the odd-power term, optimized by the genetic algorithm; is the order predistortion coefficient of the even-power term; is the number of kernels of the Volterra series; is the weight coefficient of the th Volterra kernel; is the th Volterra kernel function, used to describe the nonlinear response of the original baseband signal at the time delay is the convolution integral, representing the time-domain interaction between the original baseband signal and the kernel function; is the signal time delay value; t represents the signal time variable; k is the order variable.

[0082] In this embodiment, by default , , , and of course other values can also be set according to actual needs. The predistortion coefficients are optimized by the genetic algorithm, enabling the system to support 5th-order and 7th-order distortion compensation.

[0083] The machine learning prediction module integrates the TensorFlow Lite framework and is adapted to be connected to the adaptive adjustment module and the main control module. It is used to input the signal after adaptive multipath adjustment into a pre-trained CNN model to output predicted channel characteristic parameters (such as delay values, predistortion coefficients, and multipath parameters, etc.), and perform real-time parameter adjustment through the main control module.

[0084] Specifically, the pre-trained CNN model can be a convolutional neural network that combines a spatio-temporal fusion network and a long short-term memory network The delay parameter is predicted by the CNN model, and its expression is:

[0085] ;

[0086] ;

[0087] Among them, is the delay parameter predicted at time ; is the spatio-temporal fusion network; represents the output signal matrix at time ; is the prediction time window; is the channel historical state; is the long short-term memory network, is the number of historical time steps, is the time step variable.

[0088] In this embodiment, the spatio-temporal fusion network is a deep learning model architecture that combines spatial and temporal dimension information, aiming to improve the model's ability to model complex spatio-temporal data and prediction accuracy by simultaneously modeling the spatial features (such as images or geospatial information) and temporal features (such as time series or dynamic changes) of the data. The core of ST-Transformer lies in extending the Transformer architecture to the spatio-temporal dimension, and simultaneously modeling the dependencies in space and time through the self-attention mechanism. Traditional methods usually process spatial and temporal features separately, while ST-Transformer can capture spatio-temporal interaction information more efficiently through a unified framework.

[0089] Other channel characteristic parameters, such as phase offset and predistortion coefficient, can also be predicted by the pre-trained CNN model. Based on the genetic algorithm, the parameters of the high-order digital predistortion function are optimized and updated by minimizing the predistortion coefficient, so that the finally output combined signal meets the set requirements. The expression of the predistortion coefficient optimization objective function is:

[0090] ;

[0091] Wherein, is the error vector magnitude; is the magnitude of the th harmonic; N represents the harmonic order control parameter, which can be set to 7th order for testing, for example.

[0092] After testing, through the 7th order Volterra series predistortion function and genetic algorithm optimization, the waveform distortion degree < 0.3% (measured 0.15% - 0.28%), and the third harmonic suppression > 40 dBc (traditional scheme < 25 dBc).

[0093] In this embodiment, the machine learning prediction module can store the pre-trained CNN model and collect signal data in real time by carrying LPDDR4 memory (8GB, 3200MHz) and NVMe SSD (1TB); the model inference engine is deployed on the PL part of the Zynq chip, supporting parallel computing acceleration.

[0094] In another preferred embodiment, as Figure 1 shown, for convenient parameter adjustment, the system can also set a human-computer interaction interface and a voice remote control module, such as a 10-inch touch screen (resolution 1920×1200), and set system parameters including: carrier frequency, modulation type (AM / FM / PM / QAM), modulation depth (0% - 100%), and multipath parameters (delay, phase, attenuation), and the machine learning prediction window (Δt = 0.5ms - 5ms), so that the system supports voice commands (integrated with Google Speech-to-Text API) and remote API control (based on RESTful interface).

[0095] As Figure 3 shown, 100,000 groups of signal spectra (30MHz - 6GHz) and delay tags in a dynamic multipath environment are collected as the training data set for model training. The CNN model network structure adopts 5 layers of convolution (3×3 kernel) + 3 layers of fully connected, the loss function is MAE (mean absolute error), through the NVIDIA DGXA100 training platform, the training time is 12 hours, and when the final test set error < 2ns, it meets the pre-training requirements.

[0096] The pre-trained model is quantized and then loaded onto the Zynq chip, the input is the historical signal spectrum matrix (such as 128×64×2); the output is the predicted delay and the phase offset , which are used for error compensation.

[0097] In another preferred embodiment, in order to verify the predistortion situation of the system, high-order predistortion tests are carried out. As Figure 4As shown, the optimization variables of the high-order digital predistortion function are set to (the coefficient of the cubic term), (the coefficient of the fifth-order term), (the coefficient of the seventh-order term); the number of iterations of the genetic algorithm is set to 1000 generations, and the population size is 50. The harmonic components are detected in real time through an ADC (such as the AD9208 chip), and the gradient descent method is used to dynamically adjust the optimization coefficients to generate a signal with a distortion value meeting the requirements.

[0098] The conditions of this test are as follows: input a 5G NR 100 MHz OFDM signal (which can be modulated by 256QAM), and the output power is 20 dBm.

[0099] The test results are as follows:

[0100] When predistortion is not enabled, the error vector magnitude EVM is 3.2%, and the third harmonic component is -25 dBc;

[0101] After enabling predistortion, the error vector magnitude EVM drops to 0.8%, and the third harmonic suppression > 40 dBc.

[0102] It can be seen from this that compared with the traditional scheme without the high-order digital predistortion function, the error vector magnitude EVM of the system of the present invention combined with the high-order digital predistortion function is significantly reduced, and the harmonic suppression effect has obvious advantages.

[0103] In another preferred embodiment, in order to verify the dynamic multipath test situation of the present system, a test scenario of simulating a 28 GHz millimeter-wave channel, a time delay change rate of 50 ns / ms, and a Doppler frequency shift of 1 kHz is tested. The dynamic multipath test results are as follows: the time delay tracking error < 2 ns (the traditional scheme > 10 ns); the phase error < 3°, and the amplitude fluctuation < 0.5 dB.

[0104] It can be seen from this that for the signal simulated and generated by the present system, compared with the traditional scheme, the error is significantly lower, especially the time delay tracking error is significantly reduced.

[0105] In another preferred embodiment, in order to verify the machine learning prediction situation of the present system, in a fast time-varying channel (such as a moving speed of 120 km / h), the system performance between the prediction models with and without the machine learning prediction module enabled is compared. The results are as follows:

[0106] After enabling the prediction model, the bit error rate (BER) of the system drops from to , a reduction of 60%; the parameter adjustment delay is shortened from 1.2 ms to 0.4 ms.

[0107] It can be seen from this that by adopting the scheme of the present invention combined with the machine learning prediction module, the system bit error rate and the parameter adjustment delay time can be significantly reduced.

[0108] In summary, compared with the prior art, the present invention has the following beneficial effects:

[0109] The present invention can achieve the full-process integration of signal generation, processing and synthesis. It adopts a highly integrated design, reduces the redundancy of discrete devices, and lowers the deployment cost (for example, compared with the traditional multi-DDS chip solution, the deployment cost is reduced by 50%), and is suitable for compact test environments.

[0110] The present invention jointly optimizes and adjusts the multipath signal parameters through the RLS algorithm and Kalman filtering, improves the accuracy of the multipath signal, making the multipath delay error ≤ 2 ns (measured as ±1.2 ns to ±2 ns), the phase error ≤ 3° (measured as ±1.5° to ±3°), and the delay adjustment step resolution is 0.1 ns, meeting the stringent requirements of 5G millimeter-wave communication for timing synchronization. At the same time, the RLS algorithm combines dynamic step with sparse regularization, accelerating the convergence speed of the RLS algorithm. Compared with the traditional LMS algorithm, the convergence speed is increased by 50%, and the steady-state error is reduced by 30%.

[0111] The present invention adopts a pre-trained CNN model based on the fusion of the ST-Transformer spatio-temporal fusion network and LSTM, realizing the functions of real-time and advanced compensation for channel prediction, improving the dynamic adaptability of the system, reducing the delay variation of real-time predicted channels (prediction delay < 0.5 ms), reducing the parameter advanced adjustment error (< 2 ns), and supporting dynamic channel tracking in high-speed mobile scenarios of 120 km / h (such as vehicle communication, drone links).

[0112] The present invention is based on the closed-loop feedback of the high-speed ADC AD9208 of the high-order digital predistortion function optimized by the genetic algorithm and the adaptive adjustment module, improving the accuracy of the direct signal, making the amplitude error of the direct signal ≤ 5 mV (measured as ±3 mV to ±5 mV), and the frequency resolution reaches 0.1 Hz.

[0113] The present invention reduces the waveform distortion degree (can achieve less than 0.3%, the measured value is 0.15% to 0.28%) and improves the value of third-harmonic suppression (greater than 40 dBc, the traditional solution < 25 dBc) through the multi-order Volterra series predistortion function and genetic optimization algorithm.

[0114] The present invention supports the continuous frequency band of 30 MHz to 6 GHz, is compatible with Sub-6G and millimeter-wave communication standards (such as 5G NR FR1 / FR2, Wi-Fi 6E), and meets the requirements of multi-standard hybrid testing.

[0115] The present invention supports 8 independent multipath channels, can simulate the characteristics of 8×8 MIMO channels, and provides a high-precision test platform for verifying the beamforming and interference suppression performance of large-scale antenna systems (Massive MIMO).

[0116] Embodiment 2

[0117] As Figure 5 shown, the second embodiment of the present invention further provides a method for simulating a multi-parameter adjustable wireless transmission signal, including steps S1 to S4.

[0118] S1, the main control module controls the signal source module to generate a carrier signal and a modulation signal;

[0119] S2, the signal processing module performs signal modulation processing on the carrier signal and the modulation signal, and compensates for the non-linear distortion of the signal by using the digital pre-distortion module, and outputs a direct signal and a multipath signal;

[0120] S3, the combining module dynamically combines the direct signal and the multipath signal, and outputs a combined signal;

[0121] S4, the adaptive adjustment module collects the characteristics of the combined signal, and dynamically optimizes the multipath parameters of the combined signal based on the recursive least squares adaptive algorithm, and the optimized combined signal is used for wireless transmission.

[0122] It further includes: S5, inputting the optimized combined signal into the pre-trained CNN model of the machine learning prediction module for prediction, and outputting the predicted channel characteristic parameters for real-time adjustment by the main control module.

[0123] Specifically, the carrier signal and the modulation signal can be generated by the ADRV9009 broadband radio frequency transceiver of the signal source module, which supports multiple modulation methods such as AM, FM, PM, and QAM, the carrier frequency covers 30 MHz to 6 GHz, and the frequency resolution is 0.1 Hz.

[0124] The direct signal and the multipath signal are subjected to amplitude modulation, time delay addition and phase shift processing by the signal processing module, and the non-linear distortion is compensated by using the digital pre-distortion module. Further, the multipath signal processing includes that the time delay adjustment range is 0 to 500 ns, the step resolution is 0.1 ns; the phase adjustment range is 0° to 360°, the resolution is 0.01°; the amplitude attenuation adjustment range is 0 dB to 20 dB, and the step is 0.1 dB.

[0125] The ADAR1000 beamforming chip is used to dynamically combine multiple signals, and the output frequency band covers 30 MHz to 6 GHz, the frequency error is ≤0.1%, and the waveform distortion is <0.3%.

[0126] Collect the combined signal characteristics through a high-speed ADC, dynamically optimize the multipath parameters (delay, phase, attenuation) based on the RLS algorithm, and the adjustment period < 1 ms.

[0127] Predict parameters such as channel delay change through a convolutional neural network (CNN). Input the historical signal spectrum matrix (128×64×2) into the CNN, and output the delay prediction value to achieve lead compensation.

[0128] Use the genetic algorithm to optimize the predistortion coefficients (cubic term, fifth term, seventh term) offline, and fine-tune them online through the gradient descent method to make the EVM of the output signal < 0.8% and the harmonic suppression > 40 dBc.

[0129] Calculate the repeatability index through multiple groups of following error waveforms, so that the delay tracking error of the output signal < 2 ns and the phase error ≤ 3°.

[0130] In several embodiments provided by the embodiments of the present invention, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device and method embodiments described above are only illustrative. For example, the flowcharts in the drawings show the possible architectures, functions, and operations of devices, methods, and computer program products according to multiple embodiments of the present invention. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of the code, and the module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order from that marked in the drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and they can sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0131] In addition, the functional modules in each embodiment of the present invention can be integrated together to form an independent part, or each module can exist alone, or two or more modules can be integrated to form an independent part.

[0132] When the above-mentioned function is implemented in the form of a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, an electronic device, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media that can store program codes such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs. It should be noted that in this article, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such a process, method, article or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, article or device including the said element.

[0133] The terms used in the embodiments of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The singular forms "a", "the" and "said" used in the embodiments of the present invention and the appended claims are also intended to include the plural forms unless the context clearly dictates otherwise.

[0134] It should be understood that the term "and / or" used herein is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this article generally represents an "or" relationship between the associated objects before and after.

[0135] Depending on the context, the word "if" as used herein can be interpreted as "when" or "while" or "in response to determining" or "in response to detecting". Similarly, depending on the context, the phrase "if determined" or "if detected (stated condition or event)" can be interpreted as "when determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)".

[0136] The "first / second" mentioned in the embodiments is only used to distinguish similar objects and does not represent a specific order for the objects. It can be understood that the "first / second" can be interchanged with a specific order or sequence when permitted. It should be understood that the objects distinguished by the "first / second" can be interchanged under appropriate circumstances so that the embodiments described herein can be implemented in an order other than those illustrated or described herein.

[0137] The foregoing are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A multi-parameter adjustable wireless transmission signal simulation system, characterized in that Including: A main control module, connected to a control signal source module through SPI and AXI buses, for controlling system parameter settings, error compensation, and real-time parameter adjustment; A signal source module for generating a carrier signal and a modulation signal; A signal processing module, adaptively connected to the signal source module and the digital predistortion module, for processing the carrier signal and the modulation signal to output a direct signal and a multipath signal, and compensating for the nonlinear distortion of the direct signal and the multipath signal through the digital predistortion module; A combining module, adaptively connected to the signal processing module, for combining the compensated direct signal and multipath signal into a combined signal; An adaptive adjustment module, adaptively connected to the combining module, for collecting the characteristics of the combined signal, dynamically optimizing the multipath parameters of the combined signal using a recursive least squares adaptive algorithm, and updating the predistortion coefficient of the digital predistortion module; the optimized combined signal is used for wireless transmission; A machine learning prediction module, integrating the TensorFlow Lite framework, adaptively connected to the adaptive adjustment module and the main control module, for inputting the signal after adaptive multipath adjustment into a pre-trained CNN model to output predicted channel characteristic parameters, and performing real-time adjustment through the main control module; Wherein, the multipath parameters at least include one of delay, phase, and attenuation value; the recursive least squares adaptive algorithm is a recursive least squares method combined with a dynamic step size and sparse regularization, and its expression is: ； ； Among them, is the filter weight vector at the -th iteration; is the filter weight vector at the -th iteration; is the Kalman gain matrix; is the error signal at the -th iteration; is the dynamic step size factor; is the maximum value of the step size; is the step size decay coefficient, ; is the sparse regularization coefficient, ; is the sign function of the weight vector; The channel characteristic parameters at least include one of a time delay, a phase offset, and a pre-distortion coefficient; the pre-trained CNN model is a convolutional neural network combining a spatio-temporal fusion network and a long short-term memory network and its expression is: ； ； Among them, is the delay parameter predicted at time ; is the spatio-temporal fusion network; represents the output signal matrix at time ; is the prediction time window; is the historical state of the channel; is the long short-term memory network, is the number of historical time steps, is the time step variable.

2. The multi-parameter adjustable wireless transmission signal simulation system according to claim 1, wherein, The digital predistortion module compensates for the nonlinear distortion of the signal through a high-order digital predistortion function, and the high-order digital predistortion function is expressed as: ； Among them, is the baseband signal after predistortion processing; is the original baseband signal; is the order control parameter for odd-order nonlinear compensation; is the order control parameter for even-order nonlinear terms; is the odd-order power term of the original baseband signal; is the even-order power term of the original baseband signal; is the th-order predistortion coefficient of the odd-order power term, optimized by the genetic algorithm; is the th-order predistortion coefficient of the even-order power term; is the number of kernels of the Volterra series; is the weight coefficient of the th Volterra kernel; is the th Volterra kernel function, used to describe the nonlinear response of the original baseband signal at the time delay ; is the convolution integral, representing the time-domain interaction between the original baseband signal and the kernel function; is the signal time delay value; t represents the signal time variable; k is the order variable.

3. A multi-parameter adjustable wireless transmission signal simulation system according to claim 2, characterized in that Also including: The parameters of the high-order digital predistortion function are optimized and updated by minimizing the predistortion coefficient, so that the finally output combined signal meets the set requirements, and its expression is: ； Among them, is the error vector magnitude; is the magnitude of the th-order harmonic; N represents the harmonic order control parameter.

4. A multi-parameter adjustable wireless transmission signal simulation system according to claim 1, wherein The system parameters at least include carrier frequency, signal amplitude, modulation parameters, delay, and phase offset; the main control module uses an STM32H743 single-chip microcomputer and a Xilinx Zynq UltraScale+ MPSoC chip, and sets the carrier frequency, modulation parameters, and phase offset through the SPI interface; runs the CNN model of the machine learning prediction module through the Xilinx Zynq UltraScale+ MPSoC chip; communicates with the STM32H743 single-chip microcomputer through the AXI bus to achieve error compensation and real-time parameter adjustment.

5. A multi-parameter adjustable wireless transmission signal simulation system according to claim 1, characterized in that, The signal source module generates a carrier signal and a modulation signal based on an ADRV9009 broadband radio frequency transceiver, and outputs them after being processed by the signal processing module; The signal processing module includes an ADL5380 quadrature demodulator, an ADL5375 modulator, and a MAX20343 broadband amplifier; The ADL5380 quadrature demodulator is used to demodulate the input signal into an I / Q baseband signal; The I / Q baseband signal is compensated for nonlinear distortion through the digital predistortion module; The ADL5375 modulator is used to modulate the I / Q baseband signal after nonlinear distortion compensation into an intermediate frequency signal; The MAX20343 broadband amplifier is used to output the intermediate frequency signal after gain to compensate for the link loss; The adaptive adjustment module collects the characteristics of the combined signal through a high-speed ADC; The combining module uses the ADAR1000 beamforming chip to achieve dynamic synthesis of multiple signals.

6. A multi-parameter adjustable wireless transmission signal simulation method, applied to a multi-parameter adjustable wireless transmission signal simulation system as described in any one of claims 1-5, characterized in that, It includes: S1, the main control module controls the signal source module to generate a carrier signal and a modulation signal; S2, the signal processing module performs signal modulation processing on the carrier signal and the modulation signal, and uses the digital pre-distortion module to compensate for the non-linear distortion of the signal, and outputs the direct signal and the multipath signal; S3, the combining module dynamically synthesizes the direct signal and the multipath signal, and outputs the combined signal; S4, the adaptive adjustment module collects the characteristics of the combined signal, and dynamically optimizes the multipath parameters of the combined signal based on the recursive least squares adaptive algorithm. The optimized combined signal is used for wireless transmission.

7. A multi-parameter adjustable wireless transmission signal simulation method according to claim 6, characterized in that It also includes: S5, input the optimized combined signal into the pre-trained CNN model of the machine learning prediction module for prediction, and output the predicted channel characteristic parameters for real-time adjustment through the main control module.

Images