A terminal cooperative digital pre-distortion strategy system
By using a terminal-coordinated digital predistortion strategy system, which employs state monitoring, parameter evaluation, and memory polynomial models, the problems of signal distortion and high bit error rate in wireless communication are solved, achieving more efficient spectrum utilization and improved signal quality.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- THE 54TH RESEARCH INSTITUTE OF CHINA ELECTRONICS TECHNOLOGY GROUP CORPORATION
- Filing Date
- 2022-10-21
- Publication Date
- 2026-06-05
AI Technical Summary
In wireless communication, the high peak-to-average power ratio (PAPR) of signals caused by orthogonal amplitude modulation (OAM) and orthogonal frequency division multiplexing (OFDM) techniques leads to enhanced nonlinearity of the final stage power amplifier in the communication link, increased signal distortion, increased system bit error rate, and spectrum spread affecting the quality of adjacent channels. Furthermore, existing digital predistortion techniques are difficult to effectively coordinate with terminal modules, resulting in inaccurate models.
Design a terminal-coordinated digital predistortion strategy system. Through the collaborative work of the state monitoring module, digital predistortion module, parameter evaluation module, terminal module and front-end module, the predistortion model coefficients are calculated using the memory multinomial model and the least squares method. Training sequences are inserted for data filtering and model building to ensure data accuracy and model accuracy.
It achieves efficient collaboration between the terminal module and the digital predistortion module, improves the accuracy and linearization effect of the digital predistortion model, reduces signal distortion and bit error rate, and optimizes spectrum utilization.
Smart Images
Figure CN115664355B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of wireless communication technology, and in particular to a terminal-coordinated digital predistortion strategy system. Background Technology
[0002] In the field of wireless communication, orthogonal amplitude modulation (QAM) and orthogonal frequency division multiplexing (OFDM) technologies have been widely used to achieve higher spectrum utilization. Signals with complex digital modulation schemes also exhibit high peak-to-average power ratios (PAPR), which causes the final stage power amplifier in the communication link to be in a compressed state, resulting in enhanced nonlinearity. This leads to increased signal amplitude and phase distortion, increasing the system bit error rate, and also causes spectrum spread, affecting the communication quality of adjacent channels. To increase the number of users within a limited spectrum, it is necessary to improve the nonlinear distortion of the power amplifier.
[0003] Digital predistortion technology, as one of the power amplifier linearization technologies, has advantages such as digitalization, adaptability, and integration. It is easier to cooperate with terminal modules and can flexibly configure strategies according to different application scenarios.
[0004] Terminal collaborative design mainly refers to the need for the entire digital predistortion system to work collaboratively with the terminal module to make the digital predistortion model more accurate and improve the linearization effect of digital predistortion. Summary of the Invention
[0005] The purpose of this invention is to provide a terminal-coordinated digital predistortion strategy system, in which the digital predistortion module works in collaboration with the terminal module to provide more effective data for the predistortion module, making the modeling of the predistortion module more accurate and making more appropriate strategy decisions for rapidly changing scenarios.
[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:
[0007] A terminal-coordinated digital predistortion strategy system includes a status monitoring module, a digital predistortion module, a parameter evaluation module, a terminal module, and a front-end module;
[0008] The status monitoring module is used to detect the system's operating status. The parameters of the operating status include temperature, frequency, and power. When any of the above three parameters exceeds the threshold, the digital predistortion module is activated.
[0009] The digital predistortion module is used to calculate the predistortion coefficients and generate the lookup table;
[0010] The parameter evaluation module is used to evaluate the gain value and lookup table coefficient value of the predistortion, and calculate the normalized mean square error (NMSE) of the forward and feedback signals of each predistortion to determine whether the predistortion effect has been improved.
[0011] The terminal module generates waveforms and inserts specific sequences at fixed positions within the waveforms. These sequences contain waveform characteristics including high average power data, maximum value data, and minimum value data. At the same position, it generates a pulse to signal the predistortion module to begin predistortion.
[0012] The front-end module adds a coupling channel before the output filter and adjusts the feedback power to a suitable power value through a digitally controlled attenuator. The purpose of setting the coupling channel before the filter is to ensure that the pre-distortion signal is not affected by the out-of-band suppression of the filter.
[0013] The specific process of the terminal-coordinated digital predistortion strategy method is as follows: The host computer sends a request to enable the digital predistortion state; the status monitoring module detects the request for digital predistortion, initializes the digital predistortion module, and sends information to the terminal module. The terminal module synchronously generates a pulse at the position of the training sequence, so that the digital predistortion module starts to collect forward data from the terminal module and feedback data from the terminal module.
[0014] The collected forward and feedback data first enter the parameter evaluation module, where the average power, peak power, and peak-to-average power ratio are detected. If the following situations occur: the average power of the collected data exceeds the optimal modeling power range, no peak power is detected, or the peak-to-average power ratio is severely compressed, the current group of data is discarded and re-collected; if the conditions are still not met after multiple collections, a warning is reported to the host computer, and the digital predistortion module is stopped, waiting for the digital predistortion module to restart.
[0015] Forward and feedback data are detected by average power, peak power, and peak-to-average power ratio. If the conditions are met, the data enters the calculation of phase shift, amplitude error, and normalized mean square error. The normalized mean square error calculated for the first time after the state change is stored in a global variable for subsequent comparison. After phase shift and amplitude error correction, the forward and feedback data are sent to the digital predistortion module for coefficient calculation.
[0016] The digital predistortion module models the forward and feedback data using a memory multinomial model, calculates the predistortion model coefficients using the least squares method, and then calculates the LUT table data.
[0017] The parameter evaluation module checks the lookup table data for overflow. If overflow occurs, the currently calculated lookup table data is discarded, the predistortion coefficients are not downloaded, and the forward and feedback data are re-acquired for predistortion again. If the conditions are still not met after multiple acquisitions, a warning is reported to the host computer, and the digital predistortion module is stopped and waited for the digital predistortion module to restart.
[0018] The lookup table data is checked by the parameter evaluation module. If the conditions are met, digital predistortion is performed, and the lookup table data is downloaded to the predistorter to complete one digital predistortion cycle.
[0019] Furthermore, each time the state monitoring module detects a state change, it performs digital predistortion three times consecutively. When calculating the NMSE for the second and third predistortions, if the predistortion occurs after a state change but not during the first predistortion, the calculated NMSE needs to be compared with the previous one. If the current NMSE is better than the previous NMSE, it indicates that the previous predistortion was effective, and the next step can proceed. If the current NMSE is worse than the previous NMSE, it indicates that the digital predistortion has failed, and it is then compared with the original NMSE. If the current NMSE is better than the original NMSE, the current effect is retained; if it is worse than the original NMSE, the predistortion module is switched to a pass-through state, and digital predistortion is not performed. A smaller NMSE value is better.
[0020] Furthermore, the front-end module mainly consists of a transmit branch, a receive branch, and a feedback branch. The output port of the power amplifier at the end of the transmit branch is connected to a coupler. The coupling port of the coupler is connected to a digitally controlled attenuator to feed back to the digital predistortion module. The function of the digitally controlled attenuator is to adjust the power of the feedback branch to ensure that the data used by the digital predistortion module is neither too small and inaccurate nor too large and distorted. A temperature sensor is placed near the power amplifier to measure the amplifier's temperature status in real time. The microcontroller inside the front-end module feeds back to the status monitoring module at a certain frame frequency via an asynchronous serial port.
[0021] The beneficial effects of this invention are as follows:
[0022] 1. This invention enables the interaction and collaborative work between the terminal module and the digital predistortion module, which is more conducive to the establishment of the digital predistortion model and results in better predistortion effect;
[0023] 2. This invention simplifies the data filtering process and saves processing resources by inserting training sequences. Attached Figure Description
[0024] Figure 1 This is a schematic diagram of the overall digital predistortion scheme according to an embodiment of the present invention;
[0025] Figure 2 This is a schematic diagram of the digital predistortion module according to an embodiment of the present invention;
[0026] Figure 3 This is a flowchart of the strategy execution in an embodiment of the present invention. Detailed Implementation
[0027] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings and examples. However, the embodiments described herein are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0028] Reference Figure 1 ,to Figure 3 A terminal-coordinated digital predistortion strategy system includes a status monitoring module, a digital predistortion module, a parameter evaluation module, a terminal module, and a front-end module.
[0029] The status monitoring module runs on ARM and monitors the device's temperature, frequency, and power status information. Temperature information is detected by the front-end device and reported to the status monitoring module in real time; frequency changes are controlled by the terminal module and simultaneously notified to the predistortion module; power information is detected by the predistortion module itself.
[0030] The digital predistortion module completes the calculation of predistortion coefficients and the generation of lookup tables.
[0031] The predistortion coefficients are calculated by constructing a linear equation using the forward and feedback signals, and then calculating the predistortion model coefficients using the least squares method. The specific steps are as follows:
[0032] 1) Based on the memory polynomial model Construct matrices X and A, and rewrite the formula as Y = A * X, where X, Y, and A are x(n), y(n), and a, respectively. k,q The matrix form of ;
[0033] 2) The transformation formula yields the coefficient matrix A = (X... H X) -1 X H Y and X have a number of rows much greater than the number of columns, and there is no correlation between the data in each column. They are full-rank matrices and can be decomposed into QR.
[0034] 3) X = Q * R, where Q is an orthogonal matrix (Q... H *Q=E), H denotes the conjugate transpose, and R is an upper triangular matrix. The least squares solution can be expressed as follows:
[0035]
[0036] 4) The formula RA = Q is obtained. H By substituting Y back into A, we can obtain the parameters of A.
[0037] The lookup table uses a memoized polynomial model:
[0038]
[0039] Where x(n) and y(n) correspond to the input and output signals of the front-end module, respectively, and a k,q Here, K represents the coefficients of the model, K represents the maximum nonlinear order of the model, and Q represents the maximum memory depth of the model.
[0040] Calculated lookup table (LUT)q The (x(nq)) data is stored in the RAM of the FPGA, using the signal amplitude as the index address, and the predistorted signal is calculated by accumulating and summing. The deeper each lookup table is, the smaller the quantization interval and the higher the calculation accuracy.
[0041] The parameter evaluation module evaluates the data characteristics:
[0042] 1) Includes forward and feedback average power detection, peak power detection, and peak-to-average power ratio (PAPR) detection. Average power, peak power, and PAPR are used as criteria for determining whether to continue with the next pre-distortion step. If average power, peak power, or PAPR are outside the expected range, it indicates an abnormal signal requiring re-acquisition or a faulty device requiring repair.
[0043] 2) Includes phase shift, amplitude error, and NMSE between the forward and feedback signals. The phase shift and amplitude error are calculated, and the feedback signal is calibrated. The calibrated data serves as the input for calculating the predistortion coefficients. NMSE is used as a criterion for evaluating the quality of predistortion. Calculating the NMSE between the predistorted signal and the feedback signal after power amplifier processing allows for an assessment of the predistortion effect.
[0044] 3) Includes lookup table data detection, checking for data overflow in the lookup table. An extreme case occurs when |x(n)| = 1, where the maximum data size in the lookup table (LUT) is MAX{(a 00 +...+a 0K ),(a q0 +...+a qK ),...,(a Q0 +...+a QK This ensures that the data written to RAM from the lookup table does not overflow after the floating-point number is converted to a fixed-point number.
[0045] The terminal module inserts a training sequence of data specifically for digital predistortion in the middle of the generated communication data. This training sequence is characterized by high average power, the presence of peak data, and the presence of trough data. This specific set of data ensures that the model established by digital predistortion covers all features of the power amplifier, making the model more accurate.
[0046] Terminal collaboration refers to the information exchange and collaborative work between the terminal module, the status monitoring module, and the digital predistortion module. First, the terminal module provides a dedicated training sequence as the modeling signal for digital predistortion. When the status monitoring module detects a change in the device status, it notifies the terminal module to begin predistortion. The terminal module generates a pulse at the position of the training sequence, and the digital predistortion module begins collecting data and calculating the predistortion coefficients.
[0047] The specific process of the digital predistortion strategy method for terminal collaboration is as follows:
[0048] 1) The host computer sends a request to enable digital predistortion mode;
[0049] 2) The status monitoring module detects the need for digital predistortion, initializes each predistortion module, and sends information to the terminal module. The terminal module synchronously generates a pulse at the position of the training sequence, so that the digital predistortion module starts to collect forward and feedback data.
[0050] 3) The collected forward and feedback data first enter the parameter evaluation module, where its average power, peak power, and peak-to-average power ratio (PAPR) are checked. If any of the following conditions occur: the average power of the collected data exceeds the optimal modeling power range, no peak power is detected, or the PAPR is severely compressed, the current group of data is discarded and re-collected. If the conditions are still not met after multiple collections, a warning is reported to the host computer, and the predistortion module is stopped, waiting for the predistortion module to restart.
[0051] 4) Forward and feedback data are detected by average power, peak power, and peak-to-average power ratio (PAPR). If the conditions are met, the calculation of phase shift, amplitude error, and normalized mean square error (NMSE) is performed. The first NMSE calculated after the state change is stored in a global variable for subsequent comparisons. After phase shift and amplitude error correction, the forward and feedback data are sent to the digital predistortion module for coefficient calculation.
[0052] 5) Model the forward and feedback data using a memory polynomial model, calculate the predistortion model coefficients using the least squares method, and then calculate the LUT table data;
[0053] 6) Re-enter the parameter evaluation module to check the LUT data for overflow in the lookup table. If overflow occurs, discard the currently calculated LUT data, do not download the predistortion coefficients, and re-collect the forward and feedback data for predistortion again. If the conditions are still not met after multiple collections, report a warning to the host computer and stop the predistortion module, waiting for it to restart.
[0054] 7) The LUT table data is tested by the parameter evaluation module. If the conditions are met, the digital predistortion LUT data is downloaded to the predistorter to complete one digital predistortion.
[0055] 8) The status monitoring module performs digital predistortion three times consecutively each time it detects a status change. The execution steps for the second and third predistortions are basically the same as the first, except that when calculating the NMSE, if it is not the first predistortion after the status change, the current calculated NMSE needs to be compared with the previous one. If the current NMSE is better than the previous NMSE, it means that the previous predistortion was effective and the next step can be performed; if the current NMSE is worse than the previous NMSE, it means that the digital predistortion failed, and it is compared with the original NMSE again. If it is better than the original NMSE, the current effect is retained; if it is worse than the original NMSE, the predistortion module is switched to pass-through mode and no digital predistortion is performed.
[0056] The front-end module consists of a transmit branch, a receive branch, and a feedback branch. The output port of the power amplifier at the end of the transmit branch is connected to a coupler. The coupler's coupling port is connected to a digitally controlled attenuator (DCA) to feed back to the digital predistortion module. The DCA's function is to adjust the power of the feedback branch, ensuring that the data used by the digital predistortion module is neither too small (inaccurate) nor too large (distorted). A temperature sensor is placed near the power amplifier to measure its temperature in real time. The microcontroller inside the front-end module feeds this data back to the status monitoring module at a certain frame rate via asynchronous serial port.
Claims
1. A terminal-coordinated digital predistortion strategy system, characterized in that, It includes a status monitoring module, a digital predistortion module, a parameter evaluation module, a terminal module, and a front-end module; The status monitoring module is used to detect the system's operating status. The parameters of the operating status include temperature, frequency, and power. When any of the above three parameters exceeds the threshold, the digital predistortion module is activated. The digital predistortion module is used to calculate the predistortion coefficients and generate the lookup table; The parameter evaluation module is used to evaluate the gain value and lookup table coefficient value of the predistortion, and calculate the normalized mean square error of the forward and feedback signals of each predistortion to determine whether the predistortion effect has been improved. The terminal module generates waveforms and inserts specific sequences at fixed positions within the waveforms. These sequences contain waveform characteristics including high average power data, maximum value data, and minimum value data. At the same position, it generates a pulse to signal the predistortion module to begin predistortion. The front-end module adds a coupling channel before the output filter and adjusts the feedback power to a suitable power value through a digitally controlled attenuator. The purpose of setting the coupling channel before the filter is to ensure that the pre-distortion signal is not affected by the out-of-band suppression of the filter. The specific process of the terminal-coordinated digital predistortion strategy method is as follows: The host computer sends a request to enable the digital predistortion state; the status monitoring module detects the request for digital predistortion, initializes the digital predistortion module, and sends information to the terminal module. The terminal module synchronously generates a pulse at the position of the training sequence, so that the digital predistortion module starts to collect forward data from the terminal module and feedback data from the terminal module. The collected forward and feedback data first enter the parameter evaluation module, where the average power, peak power, and peak-to-average power ratio are detected. If the following situations occur: the average power of the collected data exceeds the optimal modeling power range, no peak power is detected, or the peak-to-average power ratio is severely compressed, the current group of data is discarded and re-collected; if the conditions are still not met after multiple collections, a warning is reported to the host computer, and the digital predistortion module is stopped, waiting for the digital predistortion module to restart. Forward and feedback data are detected by average power, peak power, and peak-to-average power ratio. If the conditions are met, the data enters the calculation of phase shift, amplitude error, and normalized mean square error. The normalized mean square error calculated for the first time after the state change is stored in a global variable for subsequent comparison. After phase shift and amplitude error correction, the forward and feedback data are sent to the digital predistortion module for coefficient calculation. The digital predistortion module models the forward and feedback data using a memory multinomial model, calculates the predistortion model coefficients using the least squares method, and then calculates the LUT table data. The parameter evaluation module checks the lookup table data for overflow. If overflow occurs, the currently calculated lookup table data is discarded, the predistortion coefficients are not downloaded, and the forward and feedback data are re-acquired and predistortion is performed again. If the conditions are still not met after multiple data collections, a warning is reported to the host computer, and the digital predistortion module is stopped, waiting for the digital predistortion module to restart; The lookup table data is checked by the parameter evaluation module. If the conditions are met, digital predistortion is performed, and the lookup table data is downloaded to the predistorter to complete one digital predistortion cycle.
2. The terminal-coordinated digital predistortion strategy system according to claim 1, characterized in that, Each time the state monitoring module detects a state change, it performs digital predistortion three times consecutively. When calculating the normalized mean square error for the second and third predistortions, if the state change occurs after the first predistortion, the normalized mean square error calculated this time needs to be compared with the previous one. If the current normalized mean square error (MSE) is better than the previous one, it means the previous predistortion was effective, and we can proceed to the next step. If the current MSE is worse than the previous one, it means the digital predistortion failed. In this case, we compare it with the original MSE again. If it is better than the original MSE, we retain the current effect. If it is worse than the original MSE, we switch the predistortion module to pass-through mode and do not perform digital predistortion. The smaller the value of the normalized mean square error, the better.
3. The terminal-coordinated digital predistortion strategy system according to claim 1, characterized in that, The front-end module mainly consists of a transmit branch, a receive branch, and a feedback branch. The output port of the power amplifier at the end of the transmit branch is connected to a coupler. The coupling port of the coupler is connected to a digitally controlled attenuator to feed back to the digital predistortion module. The function of the digitally controlled attenuator is to adjust the power of the feedback branch to ensure that the data used by the digital predistortion module is neither too small and inaccurate nor too large and distorted. A temperature sensor is placed near the power amplifier to measure the amplifier's temperature status in real time. The microcontroller inside the front-end module feeds back to the status monitoring module at a certain frame frequency via an asynchronous serial port.