Passive intermodulation elimination system suitable for 5GORAN dual-frequency radio frequency equipment

Abstract

The invention discloses a passive intermodulation elimination system suitable for 5GORAN double-frequency radio frequency equipment, which comprises a PS processing unit, a PL logic unit and a radio frequency link, the radio frequency link comprises a PIM system, the PIM system comprises a PIMI module, a PIMD module and a PIMC module, the PIMD module and the PIMI module are controlled and realized by software in the double-frequency radio frequency equipment, the PIMC module is processed by FPGA user-defined logic in real time, and the PIMC module is controlled by software in the double-frequency radio frequency equipment. The PIMD module and the PIMI module are used for determining input and configuration, and the PIMD module and the PIMI module are used for capturing required data of the Data Capture module through an FPGA (Field Programmable Gate Array); the PS processing unit is responsible for configuration of RRU multi-band passive intermodulation elimination algorithm software based on a sequence-to-sequence learning model. The system has portability, and the overall performance of the double-frequency radio frequency equipment is greatly optimized and improved.

Description

Technical Field

[0001] The present invention relates to the technical field of wireless communication equipment, and in particular to a passive intermodulation cancellation system suitable for 5G ORAN dual-band radio frequency equipment. Background Art

[0002] In wireless communication systems, remote radio units (RRUs) are widely used in 5G / 6G base stations to support multi-band coexistence and massive MIMO technology. However, in environments with high-power transmission and multi-band signal coexistence, the nonlinear characteristics of RF devices can cause passive intermodulation (PIM) interference, resulting in a degradation of received signal quality and impacting base station communication performance.

[0003] Traditional PIM interference elimination methods mainly rely on physical optimization (such as antenna layout adjustment and RF device optimization) or interference compensation methods based on mathematical models (such as polynomial modeling and regression analysis). However, these methods have the following problems: Static modeling limitations: Methods based on fixed mathematical models are difficult to adapt to the complex nonlinear intermodulation characteristics of multiple frequency bands and cannot dynamically adjust interference suppression strategies; Real-time performance and computational complexity: Traditional signal processing methods have high computational complexity and are difficult to achieve real-time PIM suppression in high-speed, low-latency 5G / 6G scenarios; Insufficient environmental adaptability: Existing PIM suppression solutions lack adaptive mechanisms and are difficult to cope with changes in intermodulation characteristics caused by environmental changes (such as temperature drift and equipment aging), resulting in poor long-term stability. Summary of the Invention

[0004] The object of the present invention is to provide a passive intermodulation cancellation system suitable for 5GORAN dual-band radio frequency equipment to solve the problems raised in the above background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a passive intermodulation cancellation system suitable for 5G ORAN dual-band radio frequency equipment, comprising a PS processing unit, a PL logic unit, and a radio frequency link. The radio frequency link includes a PIM system, which includes a PIMI module, a PIMD module, and a PIMC module. The PIMD module and PIMI module are implemented by software control in the dual-band device. The PIMC module is processed in real time by FPGA custom logic, and the input and configuration are determined by the PIMD module and PIMI module. The PIMD module and PIMI module capture the required Data Capture module data through the FPGA. The PS processing unit is responsible for configuring the RRU multi-band passive intermodulation cancellation algorithm software based on the sequence-to-sequence learning model.

[0006] Preferably, the Data Capture module: acquires the IQ data frames of the CFR output data nodes X1(t) and X1(t), acquires the IQ data frames of the ADC input node r(t), acquires the multi-band RRU transmission signals and reception signals, and performs feature extraction on the signals by using adaptive transformation to obtain multi-band intermodulation interference feature data; and realizes the DMA transmission of the captured data of each node to the PIMD module of the PS for processing.

[0007] Preferably, the PIMD module: acquires the data frames of each node captured by the Data Capture module. The PIMD module is responsible for starting data capture and processing the captured data frames by using a detection algorithm. Based on the multi-band intermodulation interference feature data, deep learning is performed on the PIM signal features by using an autoencoder to generate a non-linear intermodulation feature vector.

[0008] Preferably, the PIMI module: is responsible for processing the detected data and performing identification by using an identification algorithm. Based on the non-linear intermodulation feature vector and the multi-band transmission signal, training is performed by using a sequence-to-sequence learning model to obtain a PIM signal prediction model; the multi-band transmission signal is input into the PIM signal prediction model to generate a predicted PIM interference signal, and a residual error compensation is performed on the prediction result by using an error optimization mechanism.

[0009] Preferably, the PIMC module: is responsible for updating the PIM cancellation function of the FPGA, performing adaptive filtering processing on the predicted PIM interference signal to generate an optimized interference compensation signal, and superimposing the compensation signal and the reception signal to suppress PIM interference; based on the signal data after interference suppression, the PIM signal prediction model is dynamically updated by using an online learning mechanism; combining the model update parameters of the online learning and the long-term environmental feedback data, the hyperparameters of the PIM signal prediction model are optimized by using a reinforcement learning method, and the feature extraction strategy is adjusted based on the feedback.

[0010] Preferably, the PIMC module includes a PIMC algorithm. The PIMC algorithm: a method for eliminating multi-band passive intermodulation of an RRU based on a sequence-to-sequence learning model, the method includes:

[0011] Acquires the multi-band RRU transmission signals and reception signals, and performs feature extraction on the signals by using adaptive transformation to obtain multi-band intermodulation interference feature data;

[0012] Based on the multi-band intermodulation interference feature data, generates a non-linear intermodulation feature vector;

[0013] Trains a PIM signal prediction model by using a sequence-to-sequence learning model, and performs residual error compensation on the prediction result based on an error optimization mechanism;

[0014] Perform adaptive filtering on the predicted PIM interference signal to generate an optimized interference compensation signal;

[0015] Based on the signal data after interference suppression, use the online learning mechanism to dynamically update the PIM signal prediction model;

[0016] Combine the model update parameters of the online learning with the long-term environmental feedback data, and use the reinforcement learning method to optimize the hyperparameters and adjust the feature extraction strategy.

[0017] Preferably, the startup steps of the PIMC module are as follows:

[0018] S1: Find all interference non-linear terms in IM3 and IM5;

[0019] S2: Find the interference non-linear terms from the input DL data capture;

[0020] S3: Calculate the carrier frequency of each non-linear term and perform DDC / DUC to move the center frequency to UL fc;

[0021] S4: Filter each non-inertial term and the PIM data;

[0022] S5: PIMI equalization and recognition of N samples;

[0023] S6: PIM Cancellation. Select different sampling points and cancellation lengths;

[0024] S7: Obtain the PIMC result.

[0025] Compared with the prior art, the beneficial effects of the present invention are:

[0026] The system modular FPGA design framework is convenient for transplanting different hardware solutions, provides MATLAB algorithm module analysis and verification, an adaptive error optimization mechanism, enhances the robustness of PIM signal prediction, multi-band joint signal processing, improves the interference suppression ability, adaptive filtering processing, improves the real-time performance of PIM interference suppression, the online learning mechanism, improves the model environmental adaptability, the reinforcement learning optimizes the model hyperparameters, improves the long-term stability, directly docks with ORAN Split7.2x Category A / B devices of different manufacturers, reserves upgrade interfaces, and supports three frequencies and function expansion. Brief Description of the Drawings

[0027] Figure 1 It is the system block diagram of the present invention;

[0028] Figure 2It is a schematic flow diagram of the RRU multi-band passive intermodulation cancellation method based on the sequence-to-sequence learning model;

[0029] Figure 3 It is a processing result diagram of the B1 and B3 frequency bands before and after PIMC;

[0030] Figure 4 It is a detection result diagram of the PIMC signal correlation in different coordinate systems. Specific implementation manner

[0031] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0032] Please refer to Figures 1-4 , the present invention provides a technical solution: a passive intermodulation cancellation system applicable to 5G ORAN dual-band radio frequency devices, including a PS processing unit, a PL logic unit, and a radio frequency link. The radio frequency link includes a PIM system, and the PIM system includes a PIMI module, a PIMD module, and a PIMC module. The PIMD module and the PIMI module are implemented by software control in the dual-frequency device, and the PIMC module is processed in real time by the custom logic of the FPGA. The input and configuration are determined by the PIMD module and the PIMI module. The PIMD module and the PIMI module capture the data of the required Data Capture module through the FPGA; the PS processing unit is responsible for the configuration of the RRU multi-band passive intermodulation cancellation algorithm software based on the sequence-to-sequence learning model.

[0033] In the present invention, the Data Capture module: realizes the acquisition of the IQ data frames of the CFR output data nodes X1(t) and X1(t), realizes the acquisition of the IQ data frames of the ADC input node r(t), acquires the multi-band RRU transmission signals and reception signals, and uses adaptive transformation to extract the features of the signals to obtain multi-band intermodulation interference feature data; and realizes the DMA transmission of the captured data of each node to the PIMD module of the PS for processing.

[0034] In the present invention, the PIMD module: acquires the data frames of each node captured by the Data Capture module. The PIMD module is responsible for starting data capture and processing the captured data frames using a detection algorithm. Based on the multi-band intermodulation interference feature data, a deep learning of the PIM signal features is performed using an autoencoder to generate a non-linear intermodulation feature vector.

[0035] In the present invention, the PIMI module: is responsible for processing the detected data, performing identification using an identification algorithm, training a PIM signal prediction model based on the non-linear intermodulation feature vector and the multi-band transmission signal using a sequence-to-sequence learning model; inputting the multi-band transmission signal into the PIM signal prediction model to generate a predicted PIM interference signal, and performing residual compensation on the prediction result using an error optimization mechanism.

[0036] In the present invention, the PIMC module: is responsible for updating the PIM cancellation function of the FPGA, performing adaptive filtering on the predicted PIM interference signal to generate an optimized interference compensation signal, and superimposing the compensation signal on the received signal to suppress PIM interference; dynamically updating the PIM signal prediction model based on the signal data after interference suppression using an online learning mechanism; optimizing the hyperparameters of the PIM signal prediction model using a reinforcement learning method in combination with the model update parameters of the online learning and long-term environmental feedback data, and adjusting the feature extraction strategy based on the feedback.

[0037] In the present invention, the PIMC module includes a PIMC algorithm. The PIMC algorithm: is a method for eliminating multi-band passive intermodulation of RRU based on a sequence-to-sequence learning model, and the method includes:

[0038] Obtain the multi-band RRU transmission signal and the received signal, and perform feature extraction on the signals using an adaptive transformation to obtain multi-band intermodulation interference feature data;

[0039] Generate a non-linear intermodulation feature vector based on the multi-band intermodulation interference feature data;

[0040] Train a PIM signal prediction model using a sequence-to-sequence learning model, and perform residual compensation on the prediction result based on an error optimization mechanism;

[0041] Perform adaptive filtering on the predicted PIM interference signal to generate an optimized interference compensation signal;

[0042] Dynamically update the PIM signal prediction model based on the signal data after interference suppression using an online learning mechanism;

[0043] In combination with the model update parameters of the online learning and long-term environmental feedback data, use a reinforcement learning method to optimize the hyperparameters and adjust the feature extraction strategy.

[0044] In the present invention, Data Capture realizes the DMA transfer of data captured by each node to the PIMD module of the PS for processing; PIMD is responsible for initiating data capture and processing the captured data frames using detection algorithms; PIMI is responsible for processing the detected data and performing identification using identification algorithms; based on the non-linear intermodulation feature vectors and the multi-band transmission signals, a sequence-to-sequence learning model is used for training to obtain a PIM signal prediction model; the multi-band transmission signals are input into the PIM signal prediction model to generate predicted PIM interference signals, and an error optimization mechanism is used to perform residual compensation on the prediction results; PIMC is responsible for updating the PIM cancellation function of the FPGA, and this part of the function is completed using programmable FPGA logic functions; the predicted PIM interference signals are subjected to adaptive filtering processing to generate optimized interference compensation signals, and the compensation signals are superimposed on the received signals to suppress PIM interference.

[0045] The steps for starting the PIMC module are as follows:

[0046] S1: Find all interfering non-linear terms in IM3 and IM5;

[0047] S2: Find the interfering non-linear terms from the input DL data capture;

[0048] S3: Calculate the carrier frequencies of each non-linear term and perform DDC / DUC to shift the center frequency to UL fc;

[0049] S4: Filter each non-inert term and the PIM data;

[0050] S5: PIMI equalization and identification of N samples;

[0051] S6: PIM Cancellation. Select different sampling points and cancellation lengths;

[0052] S7: Obtain the PIMC results.

[0053] This system uses the VHDL coding language and integrates the ORAN fronthaul interface for interconnection between the O-DU and O-RU. It includes:

[0054] · 10 / 25G line rate Fronthaul Interface.

[0055] · C-Plane, U-Plane, S-Plane, M-Plane link data processing.

[0056] · Beamforming Management.

[0057] · AXI-Lite configuration interface.

[0058] · Processed by the Fronthaul Interface Package filter module.

[0059] · ORAN router module.

[0060] · Support for the 1588v2 protocol.

[0061] · Support for module simulation and system simulation.

[0062] The entire PIM elimination system solution of the present invention consists of three modules: PIM Detection (PIMD), PIM Identification (PIMI), and PIM Elimination (PIMC). Among them, the PIMD and PIMI modules are implemented by software control in the dual-frequency device, and the PIMC module needs to be processed in real time by the custom logic of the FPGA. The input and configuration are determined by the PIMD and PIMI modules; the data capture required by the PIMD and PIMI is also completed by the FPGA, and the software needs to be configured for this; this software system solution includes the RRU multi-band passive intermodulation elimination algorithm software and driver based on the sequence-to-sequence learning model, as well as the FPGA real-time processing module, without the need for additional hardware circuits; in addition, the software coding reuse rate is high, and the PIM detection and identification algorithms are reusable; the FPGA needs to provide data capture and PIM elimination functions. The application software reads the captured data, detects and identifies PIMs through the algorithm, and then calls the PIM elimination module of the FPGA to eliminate PIMs; this software system solution has portability and greatly optimizes and improves the overall performance of the dual-frequency radio frequency device.

[0063] The content not described in detail in this specification belongs to the prior art well-known to those skilled in the art. Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A passive intermodulation cancellation system applicable to 5G ORAN dual-band radio frequency devices, comprising a PS processing unit, a PL logic unit, and a radio frequency link, characterized in that: The radio frequency link includes a PIM system, which consists of a PIMI module, a PIMD module, and a PIMC module. The PIMD module and the PIMI module are controlled by software in the dual-frequency device, and the PIMC module is processed in real time by the custom logic of the FPGA. The input and configuration are determined by the PIMD module and the PIMI module, and the PIMD module and the PIMI module capture the data of the required Data Capture module through the FPGA. The PS processing unit is responsible for the configuration of the RRU multi-band passive intermodulation cancellation algorithm software based on the sequence-to-sequence learning model.

2. The passive intermodulation cancellation system for a 5G ORAN dual-band radio frequency device according to claim 1, wherein: The Data Capture module: realizes the acquisition of the IQ data frames of the CFR output data nodes X1(t) and X1(t), realizes the acquisition of the IQ data frames of the ADC input node r(t), acquires the multi-band RRU transmission signal and reception signal, and uses adaptive transformation to extract the features of the signal to obtain the multi-band intermodulation interference feature data; and realizes the DMA transmission of the captured data of each node to the PIMD module of the PS for processing.

3. The passive intermodulation cancellation system for a 5G ORAN dual-band radio frequency device according to claim 1, characterized in that: The PIMD module: acquires the data frames of each node captured by the Data Capture module. The PIMD module is responsible for starting data capture and processing the captured data frames using a detection algorithm. Based on the multi-band intermodulation interference feature data, it uses an autoencoder to perform deep learning on the PIM signal features and generates a non-linear intermodulation feature vector.

4. A passive intermodulation cancellation system for a 5G ORAN dual-band radio frequency device according to claim 1, characterized in that: The PIMI module: is responsible for processing the detected data and using an identification algorithm for identification. Based on the non-linear intermodulation feature vector and the multi-band transmission signal, it uses a sequence-to-sequence learning model for training to obtain a PIM signal prediction model; inputs the multi-band transmission signal into the PIM signal prediction model to generate a predicted PIM interference signal, and uses an error optimization mechanism to perform residual compensation on the prediction result.

5. A passive intermodulation cancellation system applicable to a 5G ORAN dual-band radio frequency device according to claim 1, characterized in that: The PIMC module: is responsible for updating the PIM cancellation function of the FPGA, performing adaptive filtering on the predicted PIM interference signal to generate an optimized interference compensation signal, and superimposing the compensation signal on the reception signal to suppress PIM interference; Based on the signal data after interference suppression, the PIM signal prediction model is dynamically updated using an online learning mechanism; combining the model update parameters of the online learning with the long-term environmental feedback data, the hyperparameters of the PIM signal prediction model are optimized using a reinforcement learning method, and the feature extraction strategy is adjusted based on the feedback.

6. The passive intermodulation cancellation system for 5GORAN dual-band radio frequency equipment according to claim 1, characterized in that: The PIMC module includes a PIMC algorithm. The PIMC algorithm: an RRU multi-band passive intermodulation cancellation method based on a sequence-to-sequence learning model, and the method includes: Acquiring the multi-band RRU transmission signal and reception signal, and using adaptive transformation to extract the features of the signal to obtain the multi-band intermodulation interference feature data; Generating a non-linear intermodulation feature vector based on the multi-band intermodulation interference feature data; Training the PIM signal prediction model using a sequence-to-sequence learning model and performing residual compensation on the prediction result based on an error optimization mechanism; Perform adaptive filtering on the predicted PIM interference signal to generate an optimized interference compensation signal; Based on the signal data after interference suppression, dynamically update the PIM signal prediction model using an online learning mechanism; Combining the model update parameters of the online learning with the long-term environmental feedback data, optimize the hyperparameters using a reinforcement learning method and adjust the feature extraction strategy.

7. A passive intermodulation cancellation system for a 5G ORAN dual-band radio frequency device according to claim 1, characterized in that: The startup steps of the PIMC module are as follows: S1: Find all interference non-linear terms in IM3 and IM5; S2: Find the interference non-linear terms from the input DL data capture; S3: Calculate the carrier frequency of each non-linear term and perform DDC / DUC to shift the center frequency to UL fc; S4: Filter each non-inertial term and the PIM data; S5: PIMI equalization and identification of N samples; S6: PIM Cancellation. Select different sampling points and cancellation lengths; S7: Obtain the PIMC result.

Images