IQ imbalance compensation system based on quadrature modulation

Abstract

This invention relates to the field of wireless communication technology, and more particularly to an IQ imbalance compensation system based on orthogonal modulation. The system first analyzes the transmitter system circuit topology to identify key imbalance regions and deploys acquisition units. Next, it acquires baseband signals from each region for feature identification, extracting region-level imbalance parameters. Then, based on these parameters, it performs multi-bandwidth interference analysis on the output signal to determine interference parameters. Subsequently, it establishes an adaptive compensation channel, dynamically setting compensation coefficients based on communication scenario requirements, device sensitivity, and channel environment. Finally, it uses these coefficients to perform compensation analysis on the interference parameters, generating adaptive compensation parameters and executing dynamic control. This method, through refined region monitoring and multi-bandwidth analysis, achieves precise compensation that dynamically adjusts according to the scenario, effectively solving the problems of inaccurate positioning and poor adaptability of traditional methods, and significantly improving signal quality and system reliability.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and in particular to an IQ imbalance compensation system based on quadrature modulation. Background Technology

[0002] In modern wireless communication systems, quadrature modulation (QM) technology is widely used to achieve high spectral efficiency data transmission. This technology modulates and synthesizes signals independently by splitting them into in-phase (I) and quadrature (Q) paths, requiring extremely high amplitude consistency and phase orthogonality between the I and Q signals. However, in actual hardware implementation, due to the non-ideal characteristics of analog front-end devices, such as local oscillator phase noise, mixer mismatch, power amplifier nonlinearity, and differences in I / Q path gain and delay, IQ imbalance is inevitably introduced. IQ imbalance leads to signal constellation diagram distortion, enhanced image frequency interference, and increased bit error rate, severely restricting the performance and reliability of the communication system, especially in broadband and high-speed communication scenarios.

[0003] Traditional IQ imbalance compensation methods are mostly based on the overall measurement of the transmitter's output signal, estimating and correcting the imbalance parameters through feedforward or back-feedback approaches. These methods typically treat the transmitter link as a black box model, making it difficult to accurately pinpoint the source of the imbalance. Furthermore, the compensation process is poorly adaptable to changes in system bandwidth and dynamic environments, especially in multi-band, multi-scenario applications where the compensation effect is limited. In addition, existing technologies often employ fixed or static compensation strategies, failing to fully consider the real-time requirements of the communication scenario, the response characteristics of user equipment, and changes in the channel environment, resulting in insufficient compensation accuracy or wasted resources.

[0004] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this invention is to provide an IQ imbalance compensation system based on quadrature modulation, which aims to solve the technical problems of existing IQ imbalance compensation methods in quadrature modulation systems, such as difficulty in accurately locating the source of imbalance, poor adaptability to multi-bandwidth dynamic environments, and lack of adaptive mechanisms that combine real-time scenarios, resulting in insufficient compensation accuracy and limiting system performance.

[0006] To achieve the above objectives, the present invention provides an IQ imbalance compensation system based on quadrature modulation, the system comprising: The regional deployment module is used to analyze key modules based on the circuit topology and design parameters of the target quadrature modulation transmission system, determine M key IQ imbalance influence regions, and deploy M quadrature signal acquisition units within the M key IQ imbalance influence regions. The feature recognition module is used to acquire M regional orthogonal baseband signals through the M orthogonal signal acquisition units, perform IQ imbalance feature recognition on the M regional orthogonal baseband signals, and obtain IQ imbalance parameter data information for the M regions. The interference analysis module is used to perform multi-bandwidth interference analysis on the quadrature modulation output signal based on the IQ imbalance parameter data information of the M regions, and to determine the multi-bandwidth IQ imbalance interference parameters. The coefficient setting module is used to establish a multi-bandwidth IQ imbalance adaptive compensation channel, and at the same time set the dynamic compensation coefficient of IQ imbalance according to the transmission requirements of the target communication scenario, the response sensitivity of user equipment and channel environment information. The dynamic compensation module is used to perform compensation analysis on the multi-bandwidth IQ imbalance interference parameters based on the IQ imbalance dynamic compensation coefficient of the multi-bandwidth IQ imbalance adaptive compensation channel, determine the multi-bandwidth IQ adaptive imbalance compensation parameters, and perform dynamic compensation control of the quadrature modulation signal IQ imbalance through the multi-bandwidth IQ adaptive imbalance compensation parameters.

[0007] Optionally, determining the M key IQ imbalance impact areas includes: Based on the circuit topology and design parameters of the target quadrature modulation transmission system, correlation data mining is performed to obtain historical IQ imbalance data of the transmitter and the characteristics of the system power amplifier and mixer module. By combining the historical IQ imbalance data of the transmitter with the characteristics of the system power amplifier and mixer module, a signal offset transmission simulation model is established to simulate the IQ imbalance transmission. Based on the transmitter’s different operating bandwidth and power conditions, the set of IQ imbalance sources is determined. The IQ imbalance propagation simulation model is used to simulate the imbalance propagation of the set of IQ imbalance sources to determine the set of imbalance impact propagation paths. The transmission contribution of the imbalance impact transmission path set is statistically analyzed and key nodes are extracted to obtain the key impact node set of IQ imbalance. Based on the key impact node set of IQ imbalance, the regions are divided and marked to determine the M key IQ imbalance impact regions.

[0008] Optionally, obtaining the IQ imbalance parameter data information for M regions includes: Based on the distribution characteristics of the orthogonal baseband signals in the M regions and the IQ imbalance analysis requirements, the sliding acquisition window length and sliding step size of the orthogonal signals are determined. The orthogonal baseband signals of the M regions are divided into blocks according to the acquisition window length and sliding step size to obtain a set of orthogonal signal blocks of the M regions. Amplitude and phase characteristics are calculated for each signal block in the set of M orthogonal signal blocks in the region to obtain the amplitude difference and phase deviation sequence of the I and Q paths; Imbalance features are extracted and quantified from the amplitude difference and phase deviation sequences of the I and Q paths to obtain the IQ imbalance parameter data information of the M regions.

[0009] Optionally, determining the multi-bandwidth IQ imbalance interference parameters includes: Based on the transmitter's operating mode and the operating bandwidth range of the quadrature modulation system, the bandwidth division interval for IQ imbalance analysis is set. According to the bandwidth division interval, the IQ imbalance parameter data information of the M regions is divided into M regional sub-bandwidth IQ imbalance sets. Based on the IQ imbalance propagation simulation model, the M regional bandwidth IQ imbalance sets and the final modulation output port are integrated and propagated to obtain the multi-bandwidth IQ imbalance propagation path. Output interference analysis is performed based on the M regional bandwidth IQ imbalance sets and the multi-bandwidth IQ imbalance propagation paths to determine the multi-bandwidth IQ imbalance interference parameters.

[0010] Optionally, determining the multi-bandwidth IQ imbalance interference parameters includes: According to the bandwidth division interval, the historical IQ imbalance data of the transmitter is divided into bandwidth and the transfer deviation is fitted to generate a multi-bandwidth imbalance transfer function. Based on the multi-bandwidth imbalance transfer function, the cumulative deviation of the M regional sub-bandwidth IQ imbalance sets and the multi-bandwidth IQ imbalance transfer paths is calculated to obtain the M regional sub-bandwidth cumulative IQ imbalance sets. Based on the bit error rate performance requirements of the quadrature modulation system, a multi-bandwidth IQ imbalance threshold is set. According to the multi-bandwidth IQ imbalance threshold, the deviation comparison and screening of the M regional multi-bandwidth cumulative IQ imbalance sets and the cumulative deviation sum calculation are performed to obtain the total intensity of multi-bandwidth IQ imbalance. Obtain the IQ imbalance assessment index set, perform interference analysis on the total intensity of the multi-bandwidth IQ imbalance according to the IQ imbalance assessment index set, and determine the interference parameters of the multi-bandwidth IQ imbalance.

[0011] Optionally, establishing a multi-bandwidth IQ imbalance adaptive compensation channel includes: Collect historical IQ compensation adjustment dataset, divide the historical IQ compensation adjustment dataset into bandwidth according to the bandwidth division interval, and obtain multi-bandwidth IQ compensation dataset; The multi-bandwidth IQ compensation dataset is divided and labeled to obtain multi-bandwidth IQ imbalance data, multi-bandwidth IQ adaptive compensation parameters, and corresponding compensation effect data. Based on the compensation effect data, the multi-bandwidth IQ imbalance data and multi-bandwidth IQ adaptive compensation parameters are optimized to obtain an available multi-bandwidth IQ compensation sample set. The available multi-bandwidth IQ compensation sample set is trained in parallel using a deep neural network structure to establish the multi-bandwidth IQ imbalance adaptive compensation channel.

[0012] Optionally, setting the dynamic compensation coefficient for IQ imbalance includes: The transmission requirements of the target communication scenario are quantitatively evaluated according to the bandwidth division intervals to determine the multi-bandwidth requirement weighting factors. Based on the bandwidth division interval, the response sensitivity of the user equipment and the channel environment information are classified and constrained to obtain the multi-bandwidth sensitivity threshold and the multi-bandwidth channel interference ratio. The IQ imbalance dynamic compensation coefficient is determined based on the empirical weighted average of the multi-bandwidth demand weighting factor, the multi-bandwidth sensitivity threshold, and the multi-bandwidth channel interference ratio.

[0013] Optionally, determining the multi-bandwidth IQ adaptive imbalance compensation parameters includes: Based on the multi-bandwidth IQ imbalance adaptive compensation channel, the multi-bandwidth IQ imbalance interference parameters are compensated and analyzed, and the basic multi-bandwidth IQ compensation parameters are output. The IQ imbalance dynamic compensation coefficient is used to dynamically correct the basic multi-bandwidth IQ compensation parameters to determine the multi-bandwidth IQ adaptive imbalance compensation parameters.

[0014] Optionally, after performing dynamic compensation control of the quadrature modulation signal IQ imbalance through the multi-bandwidth IQ adaptive imbalance compensation parameters, the method further includes: Set IQ abnormal jump conditions, and perform abnormal judgment analysis on the IQ imbalance parameter data information of the M regions based on the IQ abnormal jump conditions to obtain IQ abnormal judgment results; If the IQ anomaly determination result is yes, then the adaptive adjustment of the IQ imbalance dynamic compensation coefficient is paused during IQ imbalance dynamic compensation.

[0015] Optionally, setting the IQ abnormal jump condition includes: Acquire IQ imbalance parameter data for the M regions over N consecutive acquisition cycles, and calculate the cycle-by-cycle change rate of the IQ imbalance parameter for each region. A change rate threshold is set. If the change rate of any region exceeds the change rate threshold for K consecutive periods, or the change amount of a single change exceeds the preset maximum allowable deviation, then the IQ anomaly determination result is determined to be yes, where N and K are preset positive integers greater than 1.

[0016] In this invention, an IQ imbalance compensation system based on quadrature modulation, the circuit topology is analyzed and the system is divided into multiple key influencing regions. Combined with a distributed acquisition unit, this breaks through the limitations of traditional black-box overall measurement. This method can accurately locate specific imbalance-causing modules, such as mixers and amplifiers, thereby obtaining finer-grained regional imbalance parameters, laying a data foundation for subsequent high-precision compensation. A multi-bandwidth interference analysis mechanism is introduced, moving beyond single-frequency point correction to comprehensively evaluate imbalance characteristics across different frequency bands. This enables the method to effectively suppress image interference caused by uneven frequency response in broadband signal or multi-carrier aggregation scenarios, significantly improving the system's robustness in complex spectrum environments. Furthermore, the method innovatively combines transmission requirements, user equipment sensitivity, and real-time channel environment to set dynamic compensation coefficients. This mechanism enables the compensation strategy to be adjusted in real time according to changes in the communication scenario (such as high-speed movement, channel fading, and service type switching), avoiding both overcompensation and undercompensation, and achieving an optimal balance between performance and power consumption. By establishing an adaptive compensation channel and forming a complete closed loop from parameter identification to interference analysis to dynamic compensation control, amplitude imbalance and phase quadrature error can be corrected in real time. This effectively restores the normality of the signal constellation diagram, reduces the bit error rate and error vector amplitude, thereby significantly improving the overall communication quality and reliability of the quadrature modulation system. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the structure of the first embodiment of the IQ imbalance compensation system based on quadrature modulation of the present invention; Figure 2 This is a flowchart illustrating the specific steps involved in determining M key IQ imbalance influence regions in the IQ imbalance compensation system based on orthogonal modulation of this invention. Figure 3 This is a flowchart illustrating the specific steps involved in obtaining IQ imbalance parameter data information for M regions in the IQ imbalance compensation system based on quadrature modulation according to the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] In one embodiment, such as Figure 1As shown, an IQ imbalance compensation system based on quadrature modulation is provided, the system comprising: The area deployment module 10 is used to analyze key modules based on the circuit topology and design parameters of the target quadrature modulation transmission system, determine M key IQ imbalance influence areas, and deploy M quadrature signal acquisition units within the M key IQ imbalance influence areas. The circuit topology of the target quadrature modulation (QMC) transmitter system can be a structural diagram describing the functional modules and signal flow of the system, providing a structural basis for the analysis of key modules. Furthermore, the circuit topology can be obtained from system design documents or through reverse modeling. Design parameters can be a set of engineering indicators characterizing the performance and structural features of the QMC transmitter system, used to assist in identifying key modules sensitive to IQ balance. For example, design parameters may include, but are not limited to, one or more of mixer conversion gain, amplifier linearity, and filter group delay characteristics. Key modules can be functional units in the circuit topology that have a decisive impact on the orthogonality or amplitude consistency of IQ signals, serving as a logical basis for delineating key IQ imbalance influence regions. In an exemplary embodiment, key modules may include, but are not limited to, quadrature mixers, I / Q baseband amplifiers, and local oscillator distribution networks.

[0021] The critical IQ imbalance influence region can be a physical or functional sub-region in a quadrature modulation transmission system that significantly affects the amplitude consistency or phase orthogonality of IQ signals due to the non-ideal characteristics of the analog front-end devices. It can serve as a basic unit for IQ imbalance parameter acquisition and analysis, supporting precise location of the imbalance source. In this embodiment, the critical IQ imbalance influence region can be identified by structurally analyzing the circuit topology and design parameters of the target quadrature modulation transmission system, thus identifying the signal path regions corresponding to key modules sensitive to IQ balance. For example, the critical IQ imbalance influence region may include, but is not limited to, the mixer input pre-stage region, the local oscillator feed path region, and the power amplifier driver stage region. The M critical IQ imbalance influence regions can be a set of M independent and non-overlapping IQ imbalance sensitive regions determined after analysis, which can be used to form the spatial division basis for distributed sensing. The quadrature signal acquisition unit can be a dedicated sensing device deployed within the critical IQ imbalance influence region, used to simultaneously acquire the I and Q baseband signals within that region, providing regional-level IQ signal samples to support fine-grained imbalance feature extraction. Furthermore, the orthogonal signal acquisition unit can perform real-time digital acquisition of local I / Q signals through a high-precision analog-to-digital converter and a synchronous sampling control circuit. In one specific embodiment, the orthogonal signal acquisition unit is spatially bound to the key I / Q imbalance area, and its acquired data directly reflects the signal state of that area; simultaneously, it outputs raw signal data to the feature recognition module. The M orthogonal signal acquisition units can be a set of signal acquisition devices corresponding one-to-one with the M key I / Q imbalance areas, which can be used to achieve regional-level signal monitoring with full-link coverage.

[0022] Key module analysis based on the circuit topology and design parameters of the target quadrature modulation transmission system can be achieved by inputting the circuit topology diagram and design parameters into an analytical algorithm to identify functional units sensitive to I / Q balance. Furthermore, key module analysis based on the circuit topology and design parameters of the target quadrature modulation transmission system can be achieved by using graph theory methods to perform node sensitivity analysis on the signal flow graph, or by simulating and ranking the I / Q imbalance contribution of each module based on a device model library. This provides a logical basis for delineating key I / Q imbalance influence regions. Determining M key I / Q imbalance influence regions can be done by defining M physical or logical regions based on the location of the key modules and their signal path range. Further, determining the M key I / Q imbalance influence regions can be achieved by defining the region boundaries by extending a fixed electrical length outward from the key modules, or by dynamically delineating the influence range based on signal correlation attenuation curves. This achieves a refined perception granularity from the system level to the region level. Deploying M orthogonal signal acquisition units within the M key I / Q imbalance influence regions can be achieved by installing sensing nodes with I / Q synchronous sampling capabilities in each region. Furthermore, deploying M orthogonal signal acquisition units within the M key IQ imbalance areas can be achieved through embedded sampling probes integrated into the chip or coupled acquisition modules connected to the PCB traces, thereby achieving the technical effect of establishing a distributed signal monitoring network.

[0023] The feature recognition module 20 is used to acquire M regional orthogonal baseband signals through M orthogonal signal acquisition units, perform IQ imbalance feature recognition on the M regional orthogonal baseband signals, and obtain IQ imbalance parameter data information for the M regions. The M quadrature baseband signals can be I-channel and Q-channel baseband signal samples acquired by M quadrature signal acquisition units in each region, serving as the raw input data for IQ imbalance feature identification. IQ imbalance features can be quantifiable attributes reflecting the degree of amplitude and phase deviation between I / Q signals, characterizing the imbalance state within a specific region. In an exemplary embodiment, IQ imbalance features may include, but are not limited to, amplitude imbalance factor, phase quadrature error angle, and cross-coupling leakage component. The M region IQ imbalance parameter data can be a structured set of imbalance parameters obtained after feature identification of the M region quadrature baseband signals, supporting subsequent multi-bandwidth interference analysis and compensation decisions.

[0024] Acquiring M orthogonal baseband signals from M regions using M orthogonal signal acquisition units can be achieved by synchronously triggering all acquisition units to digitally sample and upload I / Q signals. Furthermore, this acquisition can be accomplished by using a unified clock source to drive all ADCs for hard synchronization, or by using timestamp marking and post-alignment for soft synchronization, thus achieving the technical effect of obtaining spatiotemporally aligned regional signal snapshots. IQ imbalance feature identification of the M regional orthogonal baseband signals can be performed by calculating statistical quantities such as amplitude ratio and phase difference for each group of I / Q signals. Further, IQ imbalance feature identification can be achieved by estimating IQ imbalance parameters based on least squares fitting, or by directly mapping signal waveforms to imbalance features using a trained neural network, thus achieving the technical effect of extracting standardized imbalance features for each region. The obtained IQ imbalance parameter data for the M regions can be structured and stored as parameter vectors or matrices, thus achieving the technical effect of forming a standardized data format for subsequent analysis.

[0025] Interference analysis module 30 is used to perform multi-bandwidth interference analysis on the quadrature modulation output signal based on the IQ imbalance parameter data information of M regions, and to determine the multi-bandwidth IQ imbalance interference parameters. The quadrature modulation output signal can be the RF output signal processed through the complete transmit link, and can be used as the object of multi-bandwidth interference analysis to reflect the overall IQ imbalance effect. Multi-bandwidth interference analysis can be an analytical mechanism that evaluates the image interference intensity caused by IQ imbalance across multiple frequency bands, and can be used to reveal frequency-dependent imbalance characteristics and guide frequency-selective compensation. The multi-bandwidth IQ imbalance interference parameters can be quantitative indicators of IQ imbalance impact divided by frequency band, obtained through multi-bandwidth interference analysis, and can be used as the basis for adaptive compensation channel configuration.

[0026] Multi-bandwidth interference analysis of the quadrature modulation output signal based on IQ imbalance parameter data from M regions can be achieved by mapping regional parameters to a frequency domain model and simulating or measuring image interference power at different frequency bands. Furthermore, multi-bandwidth interference analysis of the quadrature modulation output signal based on IQ imbalance parameter data from M regions can be achieved by predicting the interference level of each sub-band using a frequency domain convolution model, or by injecting known signals at multiple test frequencies and measuring the image leakage intensity, thereby revealing the differences in the spectral manifestation of imbalance. Determining the multi-bandwidth IQ imbalance interference parameters can be achieved by summarizing the multi-bandwidth interference analysis results and generating an interference parameter table indexed by frequency band, thus providing input for frequency-selective compensation.

[0027] The coefficient setting module 40 is used to establish a multi-bandwidth IQ imbalance adaptive compensation channel, and at the same time set the dynamic compensation coefficient of IQ imbalance according to the transmission requirements of the target communication scenario, the response sensitivity of user equipment and channel environment information. The multi-bandwidth IQ imbalance adaptive compensation channel can be a dynamically configurable signal processing path used to implement targeted compensation based on IQ imbalance interference parameters in different frequency bands. It can achieve frequency-selective compensation and improve image rejection capabilities in broadband or multi-carrier scenarios. Furthermore, the multi-bandwidth IQ imbalance adaptive compensation channel consists of a reconfigurable filter bank, a gain adjustment unit, and a phase correction module in the digital baseband processor, dynamically adjusting the compensation parameters of each sub-band according to the multi-bandwidth IQ imbalance interference parameters. In a specific embodiment, the multi-bandwidth IQ imbalance adaptive compensation channel may include, but is not limited to, low-frequency band compensation sub-channels, mid-frequency band compensation sub-channels, and high-frequency band compensation sub-channels. The IQ imbalance dynamic compensation coefficient can be a set of numerical parameters that adjust in real time according to changes in the communication scenario, used to control the compensation intensity and method. It can be used to adapt the compensation strategy to the current transmission requirements, equipment sensitivity, and channel state, avoiding overcompensation or undercompensation. Furthermore, the IQ imbalance dynamic compensation coefficient is generated by the coefficient setting module based on the transmission requirements of the target communication scenario, the user equipment response sensitivity, and channel environment information. In an exemplary embodiment, the IQ imbalance dynamic compensation coefficient, as an input parameter of the dynamic compensation module, directly affects the determination process of the multi-bandwidth IQ adaptive imbalance compensation parameters. The transmission requirements of the target communication scenario can be the performance requirements of the current service for data rate, latency, reliability, etc., which can be used to participate in setting the dynamic compensation coefficient to ensure that compensation resources match service priorities. User equipment response sensitivity can be the receiver's tolerance to signal distortion (such as EVM), which can be used to influence the setting of the compensation strength threshold, avoiding over-correction and wasted power. Channel environment information can be real-time wireless propagation status data including multipath fading, Doppler shift, and interference levels, which can be used to determine whether enhanced compensation is needed to combat the bit error risk caused by channel degradation.

[0028] Establishing a multi-bandwidth IQ imbalance adaptive compensation channel can be achieved by configuring programmable digital filter banks and correction units to support independent adjustment by frequency band. Furthermore, this multi-bandwidth IQ imbalance adaptive compensation channel can be established by using multi-tap FIR filters to achieve frequency domain equalization, or by using a sub-band decomposition-correction-synthesis architecture to achieve frequency division processing, thereby achieving the technical effect of constructing a compensation path with adjustable frequency response. Setting the dynamic IQ imbalance compensation coefficient based on the transmission requirements of the target communication scenario, the response sensitivity of user equipment, and channel environment information can be achieved by inputting three types of scenario information into the decision logic and outputting compensation strength weights or thresholds. Furthermore, setting the dynamic IQ imbalance compensation coefficient based on the transmission requirements of the target communication scenario, the response sensitivity of user equipment, and channel environment information can be achieved by setting coefficients based on a rule engine (such as if-else logic), or by optimizing the compensation coefficients online through a reinforcement learning agent, thereby achieving the technical effect of making the compensation strategy aware of the environment and services.

[0029] The dynamic compensation module 50 is used to perform compensation analysis on the multi-bandwidth IQ imbalance interference parameters based on the IQ imbalance dynamic compensation coefficient of the multi-bandwidth IQ imbalance adaptive compensation channel, determine the multi-bandwidth IQ adaptive imbalance compensation parameters, and perform dynamic compensation control of the quadrature modulation signal IQ imbalance through the multi-bandwidth IQ adaptive imbalance compensation parameters.

[0030] The multi-bandwidth IQ adaptive imbalance compensation parameters can be the final compensation instruction set generated by combining multi-bandwidth IQ imbalance interference parameters and IQ imbalance dynamic compensation coefficients. This set can be used to directly drive the dynamic compensation module to perform frequency-selective correction operations. The quadrature modulation signal IQ imbalance dynamic compensation control can be a real-time correction behavior implemented on the transmitted signal based on the multi-bandwidth IQ adaptive imbalance compensation parameters. This can be used to restore the signal constellation diagram regularity and reduce the bit error rate and EVM.

[0031] The multi-bandwidth IQ imbalance adaptive compensation channel employs dynamic compensation coefficients to analyze and compensate for multi-bandwidth IQ imbalance interference parameters. This can be achieved by fusing the dynamic compensation coefficients with the interference parameters to calculate the required correction values ​​for each frequency band. Furthermore, the analysis can be achieved by fusing static interference parameters and dynamic coefficients using a weighted average method, or by constraint optimization to solve for compensation parameters that satisfy both power consumption and performance objectives. This allows for the generation of compensation schemes that consider both scenario requirements and imbalance characteristics. Determining the multi-bandwidth IQ adaptive imbalance compensation parameters can be achieved by outputting a final frequency band-compensation value mapping table or filter coefficient set, thus forming executable compensation instructions. Dynamic compensation control of orthogonal modulation signal IQ imbalance using multi-bandwidth IQ adaptive imbalance compensation parameters can be achieved by loading the compensation parameters into a digital predistortion or baseband correction module to adjust the transmitted signal in real time. Furthermore, dynamic compensation control of IQ imbalance in quadrature modulation signals through multi-bandwidth IQ adaptive imbalance compensation parameters can be achieved by inserting a complex gain and phase rotation correction unit before the DAC, or by updating the OFDM subcarrier precoding matrix to achieve frequency-selective compensation, thereby achieving closed-loop correction and restoring signal quality.

[0032] Taking 5G NR multi-carrier aggregation downlink transmission as an example, the IQ imbalance compensation system based on orthogonal modulation in this embodiment can be implemented in the base station transmit link by analyzing three key IQ imbalance impact areas: the mixer, the LO allocation network, and the final-stage PA, and deploying orthogonal signal acquisition units in each area. When the user equipment moves at high speed, causing rapid channel fading and the service switches to eMBB high-throughput mode, the coefficient setting module combines high transmission requirements, medium terminal sensitivity, and strong Doppler channel information to improve the dynamic compensation coefficient; the interference analysis module finds that the high-frequency subcarriers suffer from increased image interference due to PA nonlinearity; the multi-bandwidth IQ imbalance adaptive compensation channel accordingly applies stronger phase correction to the high-frequency subband, while maintaining mild compensation in the low-frequency subband. Finally, the multi-bandwidth IQ adaptive imbalance compensation parameters output by the dynamic compensation module effectively suppress cross-carrier image leakage, maintain EVM below 3%, and ensure the reliability of high-speed data transmission.

[0033] In one embodiment, M key IQ imbalance impact regions are identified, including: Correlation data mining is performed based on the circuit topology and design parameters of the target quadrature modulation transmission system to obtain historical IQ imbalance data of the transmitter and the characteristics of the system power amplifier and mixer module. In this context, correlation data mining can be a data analysis process that extracts implicit correlations from the circuit topology, design parameters, and historical operating data of the target orthogonal modulation transmitter system. It can be used to fuse structural information with measured imbalance performance, enhancing the realism of the simulation model. In this embodiment, correlation data mining can perform feature alignment and correlation modeling between structured circuit information and unstructured historical imbalance data. For example, correlation data mining can employ graph neural networks to jointly embed and learn the circuit topology and imbalance logs, or identify the dependencies between module characteristics and imbalance indicators based on causal inference methods. Transmitter historical IQ imbalance data can be time series or statistical samples of transmitter IQ imbalance measurement results recorded under different operating conditions. This can be used to reflect the impact of actual hardware non-ideals on IQ balance and to calibrate the simulation model. System power amplifier and mixer module characteristics can be a set of non-ideal behavioral parameters of the power amplifier and mixer in terms of gain, phase, linearity, and isolation. These can be used as inputs to the core device model of the IQ imbalance propagation simulation model.

[0034] Correlation data mining based on the circuit topology and design parameters of the target orthogonal modulation transmitter system can be performed by aligning and modeling the structured circuit information with unstructured historical imbalance data. Furthermore, this operation can be achieved by constructing a module-imbalance causal network using a knowledge graph, or by jointly predicting the imbalance sensitivity of each module through multi-task learning, thereby establishing a mapping relationship between module characteristics and imbalance performance. Obtaining historical IQ imbalance data of the transmitter and the characteristics of the system power amplifier and mixer modules can be achieved by extracting relevant data from equipment log databases or device test reports, providing real-world constraints for simulation modeling.

[0035] By combining historical IQ imbalance data of the transmitter with the characteristics of the system power amplifier and mixer module, a simulation model of IQ imbalance transmission is established. The signal offset propagation simulation can inject I / Q amplitude or phase deviations into a simulation environment and track their propagation and deformation process in the transmit link. This can be used to generate training or validation data for the IQ imbalance propagation simulation model. In an exemplary embodiment, the signal offset propagation simulation can reproduce typical imbalance scenarios and record the signal evolution process in a simulation platform. For example, the signal offset propagation simulation can inject I / Q errors using SPICE-level circuit simulation or embed non-ideal module behavior into a system-level Simulink model. The IQ imbalance propagation simulation model can be a computational model built based on historical IQ imbalance data of the transmitter and non-ideal characteristics of key modules, used to simulate the propagation and accumulation process of IQ imbalance in the transmit link. It can be used to reveal the propagation mechanism of IQ imbalance from the source to the output, supporting imbalance path identification and key node location. Furthermore, the IQ imbalance propagation simulation model can be established by combining circuit topology, design parameters, power amplifier / mixer characteristics, and historical imbalance data through signal flow graph modeling and error propagation equation simulation.

[0036] By combining historical I / Q imbalance data from the transmitter with the characteristics of the system's power amplifier and mixer modules, signal offset propagation simulation can be performed. This allows for the reproduction of typical imbalance scenarios and the recording of signal evolution within a simulation platform. Furthermore, this operation can be achieved by injecting I / Q errors using SPICE-level circuit simulation or by embedding non-ideal module behaviors into the system-level Simulink model, thereby generating imbalance propagation samples covering multiple operating conditions. Establishing an I / Q imbalance propagation simulation model can involve fitting the simulation results into an analytical or data-driven model that allows for rapid inference. Further, this operation can be achieved by constructing a linear transfer function based on state-space equations or by training a graph convolutional network to directly predict the node imbalance response, thus enabling efficient prediction of imbalance propagation.

[0037] Based on the transmitter's different operating bandwidth and power conditions, the set of IQ imbalance sources is determined. The imbalance propagation simulation model is used to simulate the imbalance propagation of the IQ imbalance source set and determine the set of imbalance impact propagation paths. The transmitter's different operating bandwidths can be the set of signal bandwidths supported by the transmitter in multi-band or multi-carrier configurations. These can be used as conditional variables for identifying IQ imbalance sources, as bandwidth variations affect frequency response flatness and image rejection capability. Power condition information can be a description of the transmitter's operating state at different output power levels, including static bias point and thermal state. This can be used to influence the IQ imbalance performance of nonlinear modules such as power amplifiers, and to construct a set of imbalance sources related to operating conditions. The IQ imbalance source set can be a list of initial disturbance sources that may cause IQ imbalance under specific bandwidth and power conditions, and can be used as the input excitation set for imbalance propagation simulation. In a specific embodiment, the IQ imbalance source set may include LO phase noise sources, I / QDAC gain mismatch sources, and baseband filter group delay inconsistencies. The imbalance impact propagation path set can be a set of signal paths from the imbalance source to the system output, output by the IQ imbalance propagation simulation model. Each path represents the propagation trajectory of a specific imbalance component, and can be used to provide a structured view of imbalance propagation for subsequent contribution analysis. Furthermore, the set of imbalance impact transmission paths allows for the tracking of the coupling and diffusion paths of each source item in the IQ imbalance source set across different modules within the IQ imbalance transmission simulation model.

[0038] Determining the set of IQ imbalance sources based on transmitter operating bandwidth and power conditions can activate potential imbalance sources strongly correlated with bandwidth / power under the current operating conditions. Furthermore, this operation can be achieved by matching pre-calibrated operating conditions-imbalance source mappings using a lookup table method, or by online estimation of the current dominant imbalance mechanism type, thus enabling condition-adaptive imbalance source initialization. Simulating the imbalance propagation of the IQ imbalance source set based on an IQ imbalance propagation simulation model can involve using each imbalance source as an input stimulus and running the model to output the complete propagation path. Further, this operation can generate a set of imbalance impact propagation paths. Determining the set of imbalance impact propagation paths can involve summarizing the propagation trajectories corresponding to all source items to form a path database, providing a comprehensive view of imbalance propagation.

[0039] The transmission contribution of the imbalance impact transmission path set is statistically analyzed and key nodes are extracted to obtain the key impact node set of IQ imbalance. Based on the key impact node set of IQ imbalance, the region is divided and marked to determine M key IQ imbalance impact regions.

[0040] The transmission contribution statistics can be a process of quantitatively evaluating the imbalance amplification / attenuation effect of each node or edge in the imbalance impact transmission path, which can be used to identify the path segments with the greatest impact on the final IQ imbalance. Key node extraction can be based on the transmission contribution statistics results, filtering out circuit nodes or functional units whose contribution weight exceeds a threshold, which can be used to narrow down the area to be monitored and improve resource utilization efficiency. The set of key impact nodes for IQ imbalance can be a set of circuit nodes or functional modules that have a significant impact on the overall IQ imbalance, filtered after performing contribution statistics and key node extraction on the set of imbalance impact transmission paths. This can be used as a basis for area division, focusing on high-influence locations to improve compensation efficiency. Furthermore, the set of key impact nodes for IQ imbalance can be sorted and thresholded based on indicators such as amplitude / phase perturbation gain, signal correlation attenuation rate, or information entropy change of each node in the path. In an exemplary embodiment, the set of key impact nodes for IQ imbalance may include high-gain amplification stage nodes, orthogonal mixer cross-coupling nodes, local oscillator phase jitter sensitive nodes, etc. Furthermore, the set of key impact nodes of IQ imbalance can provide input for regional delineation markers, directly affecting the spatial boundary definition of the M key IQ imbalance impact regions.

[0041] Region partitioning can be an operation that delineates physical or logical region boundaries centered on key nodes affecting IQ imbalance, combined with electrical distance or signal correlation. This can be used to transform discrete key nodes into continuously deployable sensing areas. In one specific embodiment, region partitioning can use a Voronoi diagram to divide the node space, or set the region radius based on PCB trace delay. Performing contribution statistics and key node extraction on the set of imbalance impact propagation paths can involve calculating the average disturbance gain or information retention rate of each node across multiple paths and filtering for high-contribution nodes. Further, this operation can employ PageRank-like algorithms to evaluate node centrality, or allocate each node's marginal contribution to the total imbalance based on Shapley values, thereby focusing on the most influential imbalance propagation hubs. Obtaining the set of key nodes affecting IQ imbalance can be achieved by outputting a filtered list of high-contribution nodes, forming the core basis for region partitioning. Region partitioning based on the set of key nodes affecting IQ imbalance can be performed by delineating region boundaries centered on each key node, according to electrical length or signal attenuation threshold. Furthermore, this operation can be achieved by dividing the node space using a Voronoi diagram or by setting the region radius based on PCB trace delay, thereby generating spatial partitions for deployable acquisition units. Identifying M key IQ imbalance-affected regions can output a final set of region definitions, with each region corresponding to a key influencing node and its neighborhood, thus completing the transformation from nodes to regions and supporting subsequent distributed acquisition.

[0042] Taking the calibration of a broadband transmit link for a millimeter-wave base station as an example, the IQ imbalance compensation system based on quadrature modulation in this embodiment can be implemented in a 28GHz millimeter-wave AAU. The system performs correlation mining based on historical IQ imbalance logs and PA / mixer S-parameters, discovering that there is strong coupling between AM-PM distortion and LO leakage in the final-stage PA under high-power conditions. By simulating the imbalance propagation under different bandwidths (400MHz vs 800MHz) through an IQ imbalance propagation simulation model, the third-order intermodulation products between the DAC output and PA input are identified as the main imbalance source. After statistical analysis of the propagation contribution, the output buffer stage of the quadrature modulator is identified as the key influencing node. Based on this, the key IQ imbalance influence area, including the buffer stage and the electrical delay range of 10ps before and after it, is divided, and an on-chip sampling unit is deployed in this area to enable subsequent compensation to accurately suppress the image interference of the broadband signal edge subcarriers.

[0043] In one embodiment, IQ imbalance parameter data for M regions are obtained, including: Based on the distribution characteristics of the orthogonal baseband signals in M ​​regions and the requirements for IQ imbalance analysis, the sliding acquisition window length and sliding step size of the orthogonal signals are determined. The orthogonal baseband signals of M regions are divided into blocks according to the acquisition window length and sliding step size to obtain a set of orthogonal signal blocks of M regions. For each signal block in the set of M orthogonal signal blocks, the amplitude and phase characteristics are calculated to obtain the amplitude difference and phase deviation sequence of the I and Q paths; Imbalance features are extracted and quantified from the amplitude difference and phase deviation sequences of the I and Q paths to obtain IQ imbalance parameter data information for M regions.

[0044] The distribution characteristics of the M orthogonal baseband signals in each region can be the variation patterns of the I / Q baseband signals in the time domain, frequency domain, or statistical significance, and can be used as one of the bases for configuring the window length and step size. In this embodiment, the distribution characteristics of the M orthogonal baseband signals in each region can include, but are not limited to, one or more of the following: signal stability level, modulation order complexity, and probability of sudden interference. The IQ imbalance analysis requirements can be the constraints of the current task on the accuracy of imbalance parameter estimation, latency tolerance, and computational resources, used to guide the optimal selection of sliding window parameters. For example, IQ imbalance analysis requirements can include high-precision offline calibration requirements, low-latency online compensation requirements, and energy-efficiency-first lightweight analysis requirements. The orthogonal signal sliding acquisition window length can be a time or sample length parameter used for segmenting the regional orthogonal baseband signals, determining the duration of the signal covered by a single signal block, and can be used to control the balance between the temporal resolution and statistical stability of imbalance feature extraction. Furthermore, the sliding acquisition window length for orthogonal signals can be dynamically set based on the distribution characteristics (such as stationarity and transient frequency) of the M regions of orthogonal baseband signals and the IQ imbalance analysis requirements (such as real-time performance and accuracy). The sliding step size can be the offset of two adjacent orthogonal signal acquisition windows on the time axis, controlling the degree of overlap or update rate between signal blocks, and can be used to adjust the trade-off between the update frequency of imbalance parameters and computational overhead. In a specific embodiment, the sliding step size can be configured in conjunction with the sliding acquisition window length for orthogonal signals, determined based on the dynamic change rate of the signal and the system processing capability. Furthermore, the sliding step size, together with the sliding acquisition window length for orthogonal signals, can determine the temporal structure of signal blocks; a smaller step size can improve temporal continuity but increase computational load.

[0045] Based on the distribution characteristics of the orthogonal baseband signals in M ​​regions and the requirements for IQ imbalance analysis, the sliding acquisition window length and sliding step size of the orthogonal signals are determined. This can be done by evaluating the time-varying characteristics of the signals and the task objectives, and selecting an appropriate combination of window length and step size. Furthermore, this operation can be achieved by automatically setting the window length based on the decay rate of the signal's autocorrelation function, dynamically adjusting the step size in conjunction with CPU load, or by pre-setting multiple sets of window-step templates and switching them according to the service type. This allows the signal segmentation strategy to match the actual imbalance dynamics, avoiding feature distortion or redundancy caused by a fixed window. Segmenting the orthogonal baseband signals in the M regions according to the acquisition window length and sliding step size can be done by truncating the continuous signal stream with a specified length and step, forming overlapping or non-overlapping segments. Furthermore, this operation can be achieved by using a circular buffer to implement real-time sliding segmentation, or by using index mapping to virtually segment in memory to reduce data copying, thereby constructing a signal block sequence suitable for time-varying imbalance modeling.

[0046] Obtaining a set of M regional orthogonal signal blocks can be achieved by organizing all block results into a set data structure by region, thus providing an input format for subsequent parallel processing. The set of regional orthogonal signal blocks can be an ordered set of signal segments formed by dividing a single regional orthogonal baseband signal according to a specified window length and step size, which can be used to provide a time-varying modeling basis for non-stationary IQ imbalance. In this embodiment, the set of regional orthogonal signal blocks can be generated by framing a continuous baseband signal stream using a sliding window mechanism. For example, the set of regional orthogonal signal blocks may include a subset of steady-state signal blocks, a subset of transition-state signal blocks, a subset of sudden interference signal blocks, etc. Calculating the amplitude and phase characteristics of each signal block in the M regional orthogonal signal block set can be achieved by calculating the instantaneous or statistical amplitude-phase relationship between the I(t) and Q(t) samples within each signal block. Further, this operation can be implemented by calculating the instantaneous amplitude ratio and phase difference at each point, or by performing an FFT on the entire signal block and estimating the IQ imbalance parameters in the frequency domain, thereby generating raw deviation data reflecting the local imbalance state. Each signal block can be a single signal segment from a set of orthogonal signal blocks in the region, and can be used as the basic unit for calculating amplitude and phase characteristics. The amplitude difference and phase deviation sequence of the I and Q channels are obtained, which can be a sequence of deviation values ​​arranged in time order, thus preserving the time-varying details of the imbalance.

[0047] The amplitude difference and phase deviation sequence can be time-series data of the instantaneous amplitude difference and phase deviation between the I and Q signals calculated within each signal block, which can be used to characterize the dynamic behavior of IQ imbalance within a local time period. In a specific embodiment, the amplitude difference and phase deviation sequence can be obtained by performing point-by-point or statistical amplitude-phase operations on the I(t) and Q(t) samples within the signal block. For example, the amplitude difference and phase deviation sequence can include envelope imbalance sequence, carrier phase error sequence, cross-modulation leakage phase sequence, etc. Extracting imbalance features and quantifying them from the amplitude difference and phase deviation sequences of the I and Q channels can be achieved by applying feature engineering methods to extract key indicators from the sequence and converting them into a standard format. Furthermore, this operation can be achieved by calculating the mean, variance, kurtosis, and other statistical quantities of the sequence as feature vectors, or by using an autoencoder to compress and encode the deviation sequence to generate a low-dimensional quantized representation, thereby improving the interpretability of parameters and the applicability of compensation.

[0048] Imbalance features can reflect the essential attributes of IQ signals deviating from an ideal orthogonal state, and can be used to form the basis of quantization description. Quantization description can be the process result of converting imbalance features into a storable, transmittable, and processable numerical form, and can be used to achieve the engineering usability of the features. Imbalance feature quantization description can be a structured numerical expression formed after feature extraction from the amplitude difference and phase deviation sequences, and can be used to transform the raw deviation data into standardized parameters that can be used for compensation decisions. In this embodiment, imbalance feature quantization description can use statistical estimation, frequency domain analysis, or machine learning methods to extract key indicators from the deviation sequence and normalize or encode them. For example, imbalance feature quantization description can include mean-variance eigenvectors, principal component projection coefficients, time-frequency energy distribution moments, etc. Obtaining IQ imbalance parameter data information from M regions can summarize the quantization description results of all regions to form a unified parameter set, thereby completing the transformation from the raw signal to structured imbalance parameters. The IQ imbalance parameter data information from the M regions can be the final output set of structured imbalance parameters, which can be used by subsequent modules.

[0049] Taking the burst IQ imbalance correction in millimeter-wave broadband communication as an example, the IQ imbalance compensation system based on quadrature modulation in this embodiment can be described as follows: In a 28GHz millimeter-wave transmission link, a sudden temperature change in a certain area causes mixer gain drift, resulting in non-stationary IQ imbalance. The system detects that the quadrature baseband signal in this area exhibits high transient characteristics. Based on this, the sliding acquisition window length is set to the short-time steady-state window length (corresponding to approximately 10 OFDM symbols), and the sliding step size is set to a high refresh rate small step size (updating every 2 symbols). In the set of quadrature signal blocks obtained after segmentation, some signal blocks contain burst amplitude jumps. The amplitude and phase characteristics are calculated for each block to obtain a fine amplitude difference and phase deviation sequence; the mean-variance feature vector is extracted from it as a quantitative description of the imbalance features. The finally generated regional IQ imbalance parameter data accurately captures the time-varying profile of the imbalance, enabling the subsequent dynamic compensation module to complete the correction within 3 symbol periods, effectively suppressing constellation diagram rotation and stretching, and maintaining EVM stability.

[0050] In one embodiment, determining the multi-bandwidth IQ imbalance interference parameters includes: Based on the transmitter's operating mode and the operating bandwidth range of the quadrature modulation system, the bandwidth division interval for IQ imbalance analysis is set.

[0051] The transmitter operating mode can be the current service type or configuration state of the transmitter, such as single-carrier, carrier aggregation, high-power backoff, etc., which can influence the bandwidth allocation strategy because the spectrum usage and nonlinear effect distribution differ under different modes. The operating bandwidth range of the quadrature modulation system can be the total signal bandwidth currently supported by the system, determined by the modulation scheme and RF front-end capabilities. It can be used as the upper limit boundary of the bandwidth allocation to ensure that the analysis interval covers the effective spectrum. The IQ imbalance analysis bandwidth allocation interval can be a set of non-overlapping frequency sub-bands defined according to the transmitter operating mode and the system bandwidth range, used for frequency-domain granular analysis of IQ imbalance. In this embodiment, the IQ imbalance analysis bandwidth allocation interval can be divided according to modulation bandwidth, carrier aggregation configuration, or channel allocation strategy, such as equal bandwidth, equal interference sensitivity, or channel edge alignment.

[0052] The IQ imbalance analysis bandwidth division interval is set according to the transmitter operating mode and the operating bandwidth range of the quadrature modulation system. This can be achieved by parsing the current operating mode (such as CA configuration) and the total bandwidth to dynamically generate a set of analysis subbands. Furthermore, this operation can be implemented by dividing the interval according to the OFDM subcarrier group boundary or by adaptively dividing the frequency band based on the PA or filter group delay inflection point, thereby achieving an imbalance analysis granularity aligned with actual spectrum usage.

[0053] According to the bandwidth division interval, the IQ imbalance parameter data information of the M regions is divided into M regional sub-bandwidth IQ imbalance sets.

[0054] The M regional bandwidth-divided IQ imbalance sets can be structured datasets formed by frequency-domain segmentation of IQ imbalance parameter data from M regions according to the IQ imbalance analysis bandwidth. Each region corresponds to imbalance parameters in multiple frequency bands, which can be used to preserve the differences in imbalance characteristics of key affected regions in different frequency bands, supporting frequency-varying imbalance modeling. In an exemplary embodiment, the M regional bandwidth-divided IQ imbalance sets can perform frequency-domain interpolation or filter bank decomposition on the original regional IQ imbalance parameters (such as amplitude imbalance factor and phase error) and map them to each analysis interval. Furthermore, the M regional bandwidth-divided IQ imbalance sets can serve as input excitations for simulating multi-bandwidth IQ imbalance propagation paths, and can be linked with the IQ imbalance propagation simulation model.

[0055] The IQ imbalance parameter data of M regions are divided into M regional sub-bandwidth IQ imbalance sets according to the bandwidth division interval. This can be achieved by frequency domain mapping or interpolation of the IQ imbalance parameters of each region and assigning them to the corresponding analysis interval. In a specific embodiment, this operation can be achieved by re-estimating the imbalance parameters after frequency division of the baseband signal using a digital filter bank, or by extrapolating the frequency response model based on the original parameters to the center of each sub-band, thereby establishing a two-dimensional imbalance parameter matrix of region-frequency band.

[0056] Based on the IQ imbalance propagation simulation model, the M regional bandwidth IQ imbalance sets and the final modulation output port are integrated and simulated to obtain the multi-bandwidth IQ imbalance propagation path.

[0057] The final modulation output port can be the physical or logical interface for the RF signal output of the quadrature modulation transmit link. It can be used as the endpoint of the integrated transmission simulation to evaluate the combined effect of imbalances in each frequency band at the output. The multi-bandwidth IQ imbalance transmission path can be a set of frequency-selective signal transmission trajectories generated for each analysis bandwidth interval in the IQ imbalance transmission simulation model, from the imbalance source in each region to the final modulation output port. This can be used to reveal the propagation gain and phase accumulation characteristics of IQ imbalance in the transmit link under different frequency bands. In this embodiment, the multi-bandwidth IQ imbalance transmission path can be simulated in the IQ imbalance transmission simulation model using the set of IQ imbalances in each region's sub-bandwidth as the frequency domain excitation, simulating its coupling and synthesis process between modules at each level, up to the output port. For example, the multi-bandwidth IQ imbalance transmission path can include, but is not limited to, low-frequency band imbalance transmission paths, mid-frequency band imbalance transmission paths, and high-frequency band imbalance transmission paths. Furthermore, the multi-bandwidth IQ imbalance transmission path can, together with output interference analysis, constitute the basis for generating multi-bandwidth IQ imbalance interference parameters.

[0058] Based on the IQ imbalance propagation simulation model, an integrated propagation simulation of the IQ imbalance sets across M regions and the final modulation output port is performed. This can be achieved by injecting imbalance excitations from each frequency band and region into the simulation model in parallel and tracking their composite effect at the output. Furthermore, this operation can be implemented using a frequency-domain state-space model for frequency-by-frequency simulation, or by leveraging a multi-frequency parallel GPU-accelerated simulation framework to improve efficiency. This allows for the generation of frequency-selective imbalance propagation paths that reflect frequency-dependent link responses. The resulting multi-bandwidth IQ imbalance propagation paths can be obtained by summarizing the propagation simulation results across different analysis intervals to form structured path data, thus providing a complete view of frequency-varying imbalance propagation.

[0059] Output interference analysis is performed based on the M regional bandwidth IQ imbalance sets and the multi-bandwidth IQ imbalance propagation paths to determine the multi-bandwidth IQ imbalance interference parameters.

[0060] Output interference analysis can be a process of calculating the image interference power and constellation distortion generated at the output end of each frequency band based on regional bandwidth imbalance data and corresponding transmission paths. It can be used to quantify the actual impact of frequency selection imbalance on system performance. In a specific embodiment, output interference analysis can calculate the leakage energy at the image frequency point through frequency domain convolution, or evaluate the constellation offset of each sub-band using the error vector magnitude (EVM) simulation model.

[0061] Output interference analysis based on M regional bandwidth-divided IQ imbalance sets and multi-bandwidth IQ imbalance propagation paths can be performed by combining imbalance excitation and propagation gain to calculate image interference and EVM for each frequency band at the output. Furthermore, this operation can be achieved by superimposing the contributions of each path to calculate the total image power spectral density, or by constructing a frequency domain error transfer function to directly output the sub-band EVM prediction value, thereby obtaining an interference intensity index with frequency band resolution. Determining multi-bandwidth IQ imbalance interference parameters can be achieved by formatting the output interference analysis results into an interference parameter vector indexed by frequency band, thus providing accurate input for multi-bandwidth adaptive compensation channels.

[0062] Taking Sub-6GHz 5G NR carrier aggregation downlink transmission as an example, the IQ imbalance compensation system based on orthogonal modulation in this embodiment can be a dual-carrier aggregation mode with the base station operating in a 200MHz bandwidth. The system divides the analysis bandwidth into two 100MHz intervals according to this operating mode. After the regional acquisition unit obtains the IQ imbalance parameters of the mixer and PA region, it splits them into two sets of regional bandwidth-specific IQ imbalance sets according to the intervals. The IQ imbalance propagation simulation model shows that the high-frequency carrier has a higher phase error propagation gain due to the PA memory effect. The output interference analysis calculates that the high-frequency carrier image interference is 4dB higher than that of the low frequency. The finally generated multi-bandwidth IQ imbalance interference parameters indicate that the compensation channel needs to apply stronger phase correction in the high-frequency subband, thereby effectively suppressing the degradation of high-frequency EVM while maintaining low-frequency power consumption.

[0063] In one embodiment, determining the multi-bandwidth IQ imbalance interference parameters includes: The historical IQ imbalance data of the transmitter is divided into bandwidth intervals and the transfer deviation is fitted to generate a multi-bandwidth imbalance transfer function. The multi-bandwidth imbalance transfer function (MWF) is a mathematical mapping relationship describing the frequency-selective accumulation and nonlinear propagation characteristics of IQ imbalance in the transmit link, obtained by fitting historical IQ imbalance data within each bandwidth division interval. It can be used to accurately calculate the actual output deviation of regional imbalance after transmission through the link in each frequency band. In an exemplary embodiment, the MWF can perform regression modeling or system identification on historical IQ imbalance data of the transmitter divided by bandwidth intervals, fitting the frequency domain transmission relationship between input imbalance and output interference. Furthermore, the MWF can serve as a core tool for cumulative deviation calculation, operating on M regional bandwidth-divided IQ imbalance sets and multi-bandwidth IQ imbalance transmission paths.

[0064] Transfer bias fitting can be a process of model fitting historical IQ imbalance data in the frequency domain to extract transfer patterns, and can be used to generate transfer functions that can be used to predict imbalance propagation behavior under new operating conditions. In a specific embodiment, dividing the transmitter's historical IQ imbalance data into bandwidth intervals and fitting the transfer bias can involve dividing the historical imbalance data into preset frequency bands and fitting the input-output imbalance mapping relationship within each sub-band. Further, this operation can be achieved by fitting linear transfer coefficients using the least squares method or modeling nonlinear frequency-varying responses using Gaussian process regression, thus forming the basis for a frequency-selective imbalance transfer model. Generating multi-bandwidth imbalance transfer functions can be achieved by outputting the transfer function expressions or parameter sets corresponding to each frequency band. Further, this operation can be achieved by local linearization fitting near the center frequency of each sub-band or by continuous modeling using sliding window frequency domain kernel regression, thus enabling frequency-varying modeling of imbalance propagation.

[0065] Based on the multi-bandwidth imbalance transfer function, the cumulative deviation of the M regional sub-bandwidth IQ imbalance sets and the multi-bandwidth IQ imbalance transfer paths is calculated to obtain the M regional sub-bandwidth cumulative IQ imbalance sets. The M regional bandwidth-divided cumulative IQ imbalance sets can be data sets reflecting the actual equivalent imbalance strength at the output end, obtained by accumulating the original regional bandwidth-divided IQ imbalance sets through a multi-bandwidth imbalance transfer function (MBF). This can provide a more realistic imbalance characterization considering device nonlinearity and frequency response. In a specific embodiment, the M regional bandwidth-divided cumulative IQ imbalance sets can input the imbalance parameters of each region in each frequency band into the MBF, outputting the accumulated imbalance value after modulation by link gain, phase offset, and cross-coupling. The accumulated deviation calculation can be a computational process that maps regional imbalance parameters to equivalent output imbalance values ​​considering the full-link effect using the MBF, which can improve the realism of imbalance assessment and the targeting of compensation. Furthermore, calculating the accumulated deviation of the M regional bandwidth-divided IQ imbalance sets and the multi-bandwidth IQ imbalance transfer path based on the MBF can be done by taking the regional imbalance parameters as input and calculating their actual contribution at the output end through the transfer function. This operation can be achieved by directly substituting functions into the calculation or by obtaining the transfer gain through table lookup and interpolation, and then scaling the original imbalance, thus obtaining a cumulative imbalance value that more closely resembles physical reality. Obtaining the M regional bandwidth-divided cumulative IQ imbalance sets can summarize the cumulative imbalance results of all regions in each frequency band. Furthermore, this operation can be implemented by structured storage of the output values ​​of each frequency band, thereby forming a complete imbalance database for filtering.

[0066] Based on the bit error rate performance requirements of the quadrature modulation system, a multi-bandwidth IQ imbalance threshold is set. According to the multi-bandwidth IQ imbalance threshold, the deviation comparison and screening of the M regional multi-bandwidth cumulative IQ imbalance sets and the cumulative deviation sum calculation are performed to obtain the total intensity of multi-bandwidth IQ imbalance. The multi-bandwidth IQ imbalance threshold can be a set of tolerable upper limits for IQ imbalance divided by frequency bands, set according to the bit error rate performance requirements of the quadrature modulation system. This can be used to screen imbalance components that have a substantial impact on system performance, avoiding overcompensation for minor imbalances. In an exemplary embodiment, the multi-bandwidth IQ imbalance threshold can be established through link-level simulation or actual measurement to establish the mapping relationship between IQ imbalance and bit error rate for each frequency band, and then inversely deduce the maximum allowable imbalance value to meet the target BER. Setting the multi-bandwidth IQ imbalance threshold according to the bit error rate performance requirements of the quadrature modulation system can be based on the target BER (e.g., 10⁻⁻⁴). 5 The maximum allowable IQ imbalance for each frequency band can be calculated by reverse calculation. Furthermore, this operation can be achieved by deriving theoretical thresholds based on the Shannon limit and modulation order, or by calibrating measured thresholds through Monte Carlo simulation, thereby establishing performance-oriented imbalance tolerance boundaries.

[0067] Deviation comparison and filtering can be performed by comparing each item in the M regional bandwidth-divided cumulative IQ imbalance sets with the corresponding frequency band multi-bandwidth IQ imbalance threshold and removing items below the threshold. This can be used to focus on key interference sources and reduce subsequent calculation and compensation overhead. Cumulative deviation summation calculation can be performed by aggregating the filtered cumulative imbalance items to generate the total multi-bandwidth IQ imbalance intensity, which can be used to form a system-level imbalance severity metric. Deviation comparison and filtering and cumulative deviation summation calculation of the M regional bandwidth-divided cumulative IQ imbalance sets according to the multi-bandwidth IQ imbalance threshold can be performed item by item, retaining items exceeding the threshold, and then aggregating them into the total intensity. Furthermore, this operation can be implemented through hard threshold truncation (setting items below the threshold to zero) or soft weighting (attenuating contribution based on relative distance from the threshold), thereby enabling performance-related filtering and quantification of interference sources. Obtaining the total multi-bandwidth IQ imbalance intensity can be output as a structured total intensity vector or scalar. Furthermore, this operation can be implemented by weighting and summing the retained cumulative imbalance items according to frequency band energy, service weight, or channel quality, thus providing a unified input for interference analysis. Among them, the total intensity of multi-bandwidth IQ imbalance can be a comprehensive index that characterizes the severity of the overall frequency selection imbalance of the system by weighting or summing the cumulative IQ imbalance sets of M regional sub-bandwidths after threshold screening. It can be used as a unified input for interference analysis to support the horizontal comparison and priority ranking of the impact of imbalance between frequency bands.

[0068] Obtain the IQ imbalance assessment index set, perform interference analysis on the total intensity of multi-bandwidth IQ imbalance according to the IQ imbalance assessment index set, and determine the multi-bandwidth IQ imbalance interference parameters.

[0069] The IQ imbalance assessment index set can be a standardized set of technical indicators used to quantify the impact of IQ imbalance on communication performance. It can be used to transform the total intensity of multi-bandwidth IQ imbalance into interpretable and actionable interference parameters. For example, the IQ imbalance assessment index set may include, but is not limited to, Image Rejection Ratio (ISR), Error Vector Magnitude (EVM) degradation, and constellation rotation angle deviation. Furthermore, the IQ imbalance assessment index set can guide the interference analysis process, determining the specific constituent dimensions of the multi-bandwidth IQ imbalance interference parameters. Obtaining the IQ imbalance assessment index set can be done by loading predefined assessment dimensions from system specifications or configuration files. Further, this operation can be achieved by reading device configuration registers or calling standard communication protocol libraries, thereby determining the output format of the interference analysis. Performing interference analysis on the total intensity of multi-bandwidth IQ imbalance according to the IQ imbalance assessment index set can map the total intensity to specific indicators such as ISR and EVM degradation. Further, this operation can be achieved by directly converting using analytical formulas (e.g., EVM ∝ imbalance intensity) or by calling pre-trained performance prediction models for index estimation, thereby generating interpretable and actionable interference parameters. Determining multi-bandwidth IQ imbalance interference parameters can be achieved by organizing the output into a structure of interference parameters organized by frequency band and containing multiple evaluation metrics. Furthermore, this operation can be accomplished by structured encapsulation of ISR, EVM degradation, and rotation angle deviation for each frequency band, thereby providing high-fidelity, multi-dimensional interference input for the adaptive compensation channel.

[0070] For example, in the scenario of high-order QAM transmission under millimeter-wave eMBB services, the IQ imbalance compensation system based on quadrature modulation in this embodiment can operate in 64-QAM mode with a bit error rate requirement of 10⁻⁻⁶. 6 Based on this requirement, the multi-bandwidth IQ imbalance threshold is set to a lower value in the high-frequency edge subband (because higher-order modulation is sensitive to this). The multi-bandwidth imbalance transfer function fitted from historical data shows that the PA introduces significant nonlinear phase accumulation in the high-frequency band. After calculating the accumulated deviation, imbalance values ​​exceeding the threshold in the high-frequency region are retained, while weak imbalances in the low-frequency region are filtered out; the total intensity of the multi-bandwidth IQ imbalance is mainly contributed by the high frequency. The IQ imbalance evaluation index set converts it into EVM degradation (3.8%) and image rejection ratio (28dB). The finally generated multi-bandwidth IQ imbalance interference parameters indicate that the compensation channel needs to apply strong phase correction in the high-frequency subband to ensure that the EVM is below the modulation tolerance of 4%, while avoiding ineffective compensation for the low frequency and saving power consumption.

[0071] In one embodiment, establishing a multi-bandwidth IQ imbalance adaptive compensation channel includes: Historical IQ compensation adjustment datasets are collected and then divided into bandwidth segments according to bandwidth division intervals to obtain multi-bandwidth IQ compensation datasets. The multi-bandwidth IQ compensation dataset is divided and labeled to obtain multi-bandwidth IQ imbalance data, multi-bandwidth IQ adaptive compensation parameters, and corresponding compensation effect data. Based on the compensation effect data, the multi-bandwidth IQ imbalance data and multi-bandwidth IQ adaptive compensation parameters are optimized to obtain an available multi-bandwidth IQ compensation sample set. A deep neural network structure is used to train the available multi-bandwidth IQ compensation sample set in parallel to establish an adaptive compensation channel for multi-bandwidth IQ imbalance.

[0072] The historical IQ compensation adjustment dataset can be a time-series or statistical log of the IQ imbalance state, applied compensation parameters, and their actual effects recorded by the system in past operations. It can be used as the basic raw material for building a data-driven compensation model. In this embodiment, the historical IQ compensation adjustment dataset can be obtained by extracting historical compensation event records from system logs, calibration databases, or online monitoring modules. For example, the historical IQ compensation adjustment dataset can be used to infer the compensation effect from EVM data fed back by user devices via OTA, or by using on-chip diagnostic circuits to record internal compensation parameters and output quality indicators, thereby accumulating real-world imbalance-compensation-effect closed-loop data. The multi-bandwidth IQ compensation dataset can be a structured data set containing compensation records for each frequency band, formed by dividing the historical IQ compensation adjustment dataset into bandwidth intervals. It can be used to preserve the differences in compensation behavior across frequency bands, supporting subsequent sample selection and model training. Furthermore, the multi-bandwidth IQ compensation dataset can be obtained by frequency domain alignment and reorganization based on the previously defined IQ imbalance analysis bandwidth intervals. In one exemplary embodiment, the multi-bandwidth IQ compensation dataset can be distributed to each sub-band after frequency domain decomposition of the broadband compensation records, or only historical segments matching the current bandwidth configuration can be retained, thereby enabling frequency domain structured reorganization of historical data.

[0073] Historical IQ compensation adjustment datasets can be acquired by extracting historical compensation event records from system logs, calibration databases, or online monitoring modules. Furthermore, the acquisition of historical IQ compensation adjustment datasets can be achieved by inferring the compensation effect from EVM data fed back from user equipment via OTA, or by using on-chip diagnostic circuits to record internal compensation parameters and output quality indicators, thereby accumulating real-world imbalance-compensation-effect closed-loop data. The historical IQ compensation adjustment datasets are then divided into bandwidth intervals to obtain multi-bandwidth IQ compensation datasets. This can be done by mapping each historical record to a corresponding frequency band subset based on the previously defined IQ imbalance analysis bandwidth intervals. In one specific embodiment, this operation can be achieved by frequency domain decomposition of the broadband compensation records and allocation to each sub-band, or by retaining only historical segments matching the current bandwidth configuration, thereby enabling the frequency domain structured reorganization of historical data.

[0074] Multi-bandwidth IQ imbalance data can be a set of IQ amplitude imbalance and phase error parameters organized by frequency band, which can be used to characterize the current imbalance state of each frequency band. Multi-bandwidth IQ adaptive compensation parameters can be correction parameters historically used for imbalances in specific frequency bands, such as complex gain, phase rotation angle, or filter coefficients, which can be used as the output target of the compensation model. Compensation effect data can be actual performance improvement indicators after applying a set of compensation parameters, such as EVM reduction, bit error rate change, constellation diagram recovery, etc., which can be used as the evaluation criteria for sample selection. Dividing and labeling the multi-bandwidth IQ compensation dataset to obtain multi-bandwidth IQ imbalance data, multi-bandwidth IQ adaptive compensation parameters, and corresponding compensation effect data can be achieved by parsing each record to extract input imbalance features, used compensation parameters, and validated performance indicators. Furthermore, this operation can be achieved by automatically parsing log fields to generate triples, or by re-evaluating the compensation effect through simulation playback of historical conditions, thereby constructing a standard supervised learning sample format.

[0075] The available multi-bandwidth IQ compensation sample set can be a high-quality triplet dataset retained after compensation effect screening. It includes multi-bandwidth IQ imbalance data, corresponding compensation parameters, and validated compensation effect indicators. This dataset can be used as a supervisory signal source for deep neural network training, ensuring the model learns effective compensation strategies. In an exemplary embodiment, the available multi-bandwidth IQ compensation sample set can be obtained by setting a threshold based on compensation effect data (such as EVM improvement and BER reduction rate) to eliminate inefficient or over-compensated samples. For example, the available multi-bandwidth IQ compensation sample set may include one or more of the following: a high EVM improvement sample subset, a low-power compensation sample subset, or a sample subset specifically for high-speed mobile scenarios. The available multi-bandwidth IQ compensation sample set is obtained by optimizing the multi-bandwidth IQ imbalance data and multi-bandwidth IQ adaptive compensation parameters according to the compensation effect data. This can be achieved by setting a compensation effect threshold (such as EVM improvement ≥ 1%) to filter invalid or negative samples. Furthermore, this operation can be implemented by using Pareto fronts to screen samples that balance performance and power consumption, or by using clustering to retain effective samples under diverse operating conditions, thereby improving the quality of training data and avoiding the model learning suboptimal strategies.

[0076] A deep neural network structure can be a trainable computational model used to model the complex nonlinear relationship between multi-bandwidth IQ imbalance and optimal compensation parameters. It can be used to learn frequency-selective compensation strategies from historical high-quality compensation samples, supporting generalization to unseen operating conditions. In one specific embodiment, the deep neural network structure can employ parallel sub-networks or a shared backbone + frequency band branch structure, with each sub-band corresponding to an independent output head or obtained through frequency domain attention weighting. For example, the deep neural network structure can include multi-task learning neural networks, graph neural networks (with frequency bands as nodes), or time-frequency joint Transformer architectures. The multi-bandwidth IQ imbalance adaptive compensation channel can be a nonlinear mapping system built based on a deep neural network, capable of outputting frequency band adaptive compensation parameters in real time according to the input multi-bandwidth IQ imbalance interference parameters. It can be used to achieve high-precision, low-latency, scene-aware dynamic correction of broadband IQ imbalance. In this embodiment, the multi-bandwidth IQ imbalance adaptive compensation channel can be obtained by parallel training of a deep neural network on an available multi-bandwidth IQ compensation sample set. The input is IQ imbalance data for each frequency band, and the output is the compensation parameters for the corresponding frequency band. Furthermore, the multi-bandwidth IQ imbalance adaptive compensation channel can receive multi-bandwidth IQ imbalance interference parameters as input and output multi-bandwidth IQ adaptive imbalance compensation parameters for the dynamic compensation module to execute. For example, the multi-bandwidth IQ imbalance adaptive compensation channel may include one or more of the following: a frequency domain compensation channel based on a convolutional architecture, a cross-frequency band collaborative compensation channel based on an attention mechanism, or a lightweight embedded inference compensation channel.

[0077] Parallel training of available multi-bandwidth IQ compensation sample sets using deep neural network structures establishes a multi-bandwidth IQ imbalance adaptive compensation channel. This can be achieved by using multi-bandwidth IQ imbalance data as input and multi-bandwidth IQ adaptive compensation parameters as labels to train an end-to-end mapping model. Furthermore, this operation can employ a multi-output head network to predict compensation parameters for each frequency band separately, or use a frequency domain attention mechanism to achieve cross-frequency band information fusion before output, thereby obtaining an intelligent compensation engine with frequency band adaptation and scene generalization capabilities.

[0078] Taking the adaptive calibration of millimeter-wave communication links in high-speed rail scenarios as an example, the IQ imbalance compensation system based on orthogonal modulation in this embodiment can be based on the base station recording a large number of IQ compensation logs under high-speed mobile scenarios, covering different Doppler spread and carrier aggregation configurations. The system divides the 200MHz bandwidth into four 50MHz intervals to construct a multi-bandwidth IQ compensation dataset. After division and labeling, it is found that high-frequency bands require stronger phase correction under high-speed scenarios but are prone to overcompensation. Through data screening of compensation effect, samples with significant EVM improvement and reasonable power consumption are retained. When a new user accesses the deep neural network compensation channel, it outputs the optimal compensation parameters for each frequency band within 0.5ms based on the real-time multi-bandwidth IQ imbalance interference parameters, so that the EVM under high-speed mobile scenarios is stabilized below 2.8%, which is 1.2 percentage points lower than the traditional lookup table method.

[0079] In one embodiment, setting a dynamic compensation coefficient for IQ imbalance includes: The transmission requirements of the target communication scenario are quantitatively evaluated according to the bandwidth division intervals, and the weighting factors of multiple bandwidth requirements are determined. Based on the bandwidth division interval, user equipment response sensitivity and channel environment information are classified and constrained to obtain multi-bandwidth sensitivity thresholds and multi-bandwidth channel interference ratios. The dynamic compensation coefficient for IQ imbalance is determined based on the empirical weighted average of the multi-bandwidth demand weighting factor, the multi-bandwidth sensitivity threshold, and the proportion of multi-bandwidth channel interference.

[0080] The multi-bandwidth demand weighting factor can be a frequency band-level service priority indicator obtained by quantifying the transmission requirements (such as rate, latency, and reliability) of the target communication scenario according to the bandwidth division interval based on IQ imbalance analysis. It is used to reflect the importance of different frequency bands in the current service context and guide the allocation of compensation resources. In this embodiment, the multi-bandwidth demand weighting factor can assign normalized weights to each analysis interval based on service type, QoS level, and subcarrier scheduling information. Furthermore, the multi-bandwidth demand weighting factor can include, but is not limited to, the weight of high-throughput service frequency bands, the weight of low-latency control channel frequency bands, and the weight of high-reliability retransmission frequency bands. Quantifying and evaluating the transmission requirements of the target communication scenario according to the bandwidth division interval can involve parsing the current service type and QoS policy, mapping the transmission requirements to each analysis frequency band, and assigning weights. For example, this operation can be achieved by inferring the importance of frequency bands based on the number of RBs allocated by the scheduler and the modulation order, thereby realizing the frequency band-level expression of service priority. In an exemplary embodiment, this operation can also be automatically converted into weighting factors using KPI indicators defined in the SLA contract. Determining the multi-bandwidth demand weighting factors can be achieved by outputting a normalized frequency band-level service weight vector. Furthermore, this operation provides a service-driven basis for dynamic compensation.

[0081] Multi-bandwidth sensitivity thresholds can be upper limits for IQ imbalance tolerance set for each bandwidth division interval based on the receiving sensitivity characteristics of the user equipment in the corresponding frequency band. These thresholds can be used to prevent bit error rate exceeding limits due to undercompensation in sensitive frequency bands. In one specific embodiment, the multi-bandwidth sensitivity threshold can be extracted from the UE capability report or link adaptive feedback for each frequency band's EVM or SINR tolerance threshold and mapped to the maximum allowable IQ imbalance. Multi-bandwidth channel interference ratio can be a quantitative indicator of the degree of signal quality degradation caused by external interference and channel factors such as multipath fading within each bandwidth division interval. This can be used to characterize the enhancement effect of the channel environment on the IQ imbalance correction requirement. In this embodiment, the multi-bandwidth channel interference ratio can be calculated using channel state information (CSI), interference power spectral density estimation, or SINR measurement to determine the proportion of interference energy in each frequency band to the total received energy.

[0082] Classifying and constraining user equipment response sensitivity and channel environment information based on bandwidth division intervals can be achieved by grouping UE capabilities and real-time channel measurement data by frequency band and applying upper and lower limits or classification labels. In an exemplary embodiment, this operation can be implemented by extracting frequency band sensitivity specifications according to 3GPPUECategory. For example, this operation can also utilize CSI-RS measurement results for frequency domain channel clustering, thereby establishing joint constraints between the terminal and the channel. Obtaining multi-bandwidth sensitivity thresholds and multi-bandwidth channel interference ratios can be achieved by outputting the sensitivity tolerance threshold and interference ratio values ​​for each analysis interval separately, thus forming a two-dimensional input of terminal perception and environment perception. The empirical weighted average can be the result of a mathematical operation that fuses multi-bandwidth demand weighting factors, multi-bandwidth sensitivity thresholds, and multi-bandwidth channel interference ratios according to preset empirical weights, and can be used to generate a frequency band-level compensation strength benchmark that comprehensively considers the three dimensions of service, terminal, and channel. In this embodiment, the empirical weighted average serves as the direct calculation basis for the IQ imbalance dynamic compensation coefficient. Furthermore, the empirical weighted average can be linearly weighted using fixed weight coefficients (such as 0.4:0.3:0.3), or the weights of each factor can be dynamically adjusted based on historical compensation effect feedback.

[0083] The dynamic compensation coefficient for IQ imbalance is determined by an empirical weighted average of multi-bandwidth demand weighting factors, multi-bandwidth sensitivity thresholds, and multi-bandwidth channel interference proportions. This can be achieved by substituting these three types of parameters into a weighting formula to calculate the compensation intensity coefficient for each frequency band. In one specific embodiment, this operation can be implemented by using a lookup table to match pre-stored coefficients corresponding to typical scenario combinations. For example, this operation can also be implemented by solving online for weighting coefficients that satisfy the power consumption-performance Pareto optimality, thereby generating frequency band adaptive and scenario-aware dynamic compensation commands.

[0084] Taking high-speed train 5G private network communication as an example, the IQ imbalance compensation system based on orthogonal modulation in this embodiment can be used when the train is passing through a tunnel at 300km / h, with the system operating at a 100MHz bandwidth, divided into two 50MHz analysis intervals. Video surveillance services are concentrated in the low-frequency band, which is assigned a high-demand weight factor (0.8); the high-frequency band is used for control signaling, with a lower weight (0.3). The UE reports that the low-frequency band has poor receiving sensitivity (relaxed threshold), but the high-frequency band has a channel interference rate as high as 60% due to severe Doppler spread. After empirical weighted averaging, the system generates a higher IQ imbalance dynamic compensation coefficient in the high-frequency band, strengthening phase correction to counteract rapid channel changes; while the compensation intensity in the low-frequency band is appropriately reduced to save baseband processing power consumption. Overall, while ensuring the reliability of the control link, the system avoids introducing additional noise into the video stream through compensation.

[0085] In one embodiment, determining the multi-bandwidth IQ adaptive imbalance compensation parameters includes: Based on the multi-bandwidth IQ imbalance adaptive compensation channel, the multi-bandwidth IQ imbalance interference parameters are compensated and analyzed, and the basic multi-bandwidth IQ compensation parameters are output. The IQ imbalance dynamic compensation coefficient is used to dynamically correct the basic multi-bandwidth IQ compensation parameters, and the multi-bandwidth IQ adaptive imbalance compensation parameters are determined.

[0086] The basic multi-bandwidth IQ compensation parameters can be an initial set of compensation parameters directly generated by the multi-bandwidth IQ imbalance adaptive compensation channel based on the multi-bandwidth IQ imbalance interference parameters, without considering dynamic factors of the communication scenario. This set can be used as a frequency-band-related static correction benchmark, reflecting the amplitude and phase mismatch caused by hardware non-ideals in each sub-band. In this embodiment, the basic multi-bandwidth IQ compensation parameters can be analytically calculated from the multi-bandwidth IQ imbalance interference parameters using a frequency-selective correction model within the multi-bandwidth IQ imbalance adaptive compensation channel. This frequency-selective correction model includes sub-band filter coefficient mapping or lookup tables, etc. Furthermore, the basic multi-bandwidth IQ compensation parameters can be input as intermediate variables to the dynamic correction calculation stage, working together with the IQ imbalance dynamic compensation coefficients to generate the final compensation parameters. For example, the basic multi-bandwidth IQ compensation parameters may include, but are not limited to, the low-frequency band basic gain correction value, the mid-frequency band basic phase rotation angle, and the high-frequency band basic cross-coupling suppression coefficient.

[0087] This paper analyzes the compensation of multi-bandwidth IQ imbalance interference parameters based on a multi-bandwidth IQ imbalance adaptive compensation channel, and outputs basic multi-bandwidth IQ compensation parameters. This can be achieved by inputting the IQ imbalance interference parameters, divided by frequency band, into a configured adaptive compensation channel, and generating initial compensation values ​​for the corresponding frequency bands through a built-in correction model. Furthermore, this operation can be achieved by interpolation using a pre-stored interference-compensation mapping lookup table, or by generating basic parameters through real-time solving of a minimum mean square error optimization problem. This allows for the establishment of a preliminary compensation scheme matching the frequency response characteristics, preserving the refined modeling results of the multi-bandwidth interference analysis. The basic multi-bandwidth IQ compensation parameters are dynamically corrected using IQ imbalance dynamic compensation coefficients to determine the multi-bandwidth IQ adaptive imbalance compensation parameters. This can be achieved by using the IQ imbalance dynamic compensation coefficients as weighting factors or modulation functions to scale, offset, or nonlinearly map the basic multi-bandwidth IQ compensation parameters. In one specific embodiment, this operation can be achieved by linearly weighting the basic parameters, i.e., the final parameter is equal to the dynamic coefficient multiplied by the basic parameter; or by mapping the dynamic coefficient to a correction ratio through a nonlinear activation function and then applying it to the basic parameter, so that the final compensation parameter has both frequency domain accuracy and scene adaptability, avoiding the performance degradation of static compensation in dynamic environments.

[0088] For example, in the scenario of 5G millimeter-wave communication in high-speed mobile situations, the IQ imbalance compensation system based on orthogonal modulation in this embodiment can be as follows: When a train crosses the cell boundary at a speed of 300 km / h, the channel experiences severe Doppler spread and rapid fading. The system first identifies, through multi-bandwidth interference analysis, that the high-frequency subcarriers in 28GHz carrier aggregation suffer from significantly enhanced image interference due to PA memory effect. Based on this, the multi-bandwidth IQ imbalance adaptive compensation channel outputs a strong basic phase correction value. At the same time, the coefficient setting module detects that the current service is URLLC, the terminal sensitivity is low, and the channel CQI drops sharply, thus generating a high IQ imbalance dynamic compensation coefficient. The dynamic correction calculation unit applies this coefficient to the basic parameters to further enhance the high-frequency compensation intensity. The resulting multi-bandwidth IQ adaptive imbalance compensation parameters effectively suppress constellation rotation and diffusion caused by high-speed movement, maintaining EVM compliance while avoiding over-correction of low-frequency stable subbands, thus saving baseband processing power consumption.

[0089] In one embodiment, after performing dynamic compensation control of the quadrature modulation signal IQ imbalance using multi-bandwidth IQ adaptive imbalance compensation parameters, the method further includes: Set IQ abnormal jump conditions, and perform anomaly judgment analysis on the IQ imbalance parameter data of M regions based on the IQ abnormal jump conditions to obtain IQ abnormal judgment results; If the IQ anomaly determination result is yes, then the adaptive adjustment of the IQ imbalance dynamic compensation coefficient is paused during IQ imbalance dynamic compensation.

[0090] The IQ anomaly jump condition can be a preset set of criteria used to determine whether abnormal changes have occurred in the IQ imbalance parameter data of a region. It can be used as a threshold or logical rule for anomaly detection, distinguishing between parameter jumps caused by real-world scene changes and transient interference. In this embodiment, the IQ anomaly jump condition can be pre-configured or generated online by the system based on historical IQ imbalance parameter statistical characteristics, hardware stability indicators, and typical interference models. The IQ anomaly determination result can be a binary judgment signal output after analyzing the IQ imbalance parameter data information of M regions based on the IQ anomaly jump condition. It can be used to indicate whether the current IQ imbalance parameter is in a reliable state and to determine whether the dynamic compensation coefficient should continue to adaptively adjust. In an exemplary embodiment, the IQ anomaly determination result can output a "yes" or "no" judgment flag by comparing the matching degree between real-time parameters and the IQ anomaly jump condition. The adaptive adjustment of the IQ imbalance dynamic compensation coefficient can be a process of updating the IQ imbalance dynamic compensation coefficient in real time according to changes in the communication scenario. It can be used to ensure that the compensation strategy continuously matches the current transmission requirements, device sensitivity, and channel status. For example, the adaptive adjustment of the IQ imbalance dynamic compensation coefficient is paused when the IQ anomaly determination result is "yes" to prevent erroneous adjustments based on abnormal data.

[0091] Setting IQ abnormal jump conditions can involve configuring thresholds, rate limits, or logical rules to identify abnormal jumps in IQ parameters. Furthermore, setting IQ abnormal jump conditions can be achieved by setting dynamic thresholds based on long-term IQ parameter sliding window statistics, or by using machine learning models to identify normal parameter evolution trajectories and define deviation boundaries. This establishes a benchmark for anomaly detection and improves the system's ability to identify transient interference. Anomaly determination analysis of IQ imbalance parameter data from M regions based on IQ abnormal jump conditions can be performed by comparing the current IQ imbalance parameter data from the M regions with the IQ abnormal jump conditions to determine if abnormal jumps exist. In a specific embodiment, this operation can be achieved by calculating the parameter change during adjacent sampling periods and comparing it with the jump rate limit, or by evaluating the physical consistency between parameters in multiple regions and detecting isolated anomalies. This enables the identification of unreliable parameters caused by sudden interference, hardware transient failures, or measurement noise. Obtaining the IQ anomaly determination result can output the final conclusion (yes / no) of the anomaly determination analysis. Furthermore, this operation can provide a basis for deciding whether to pause adaptive adjustment. Pausing the adaptive adjustment of the IQ imbalance dynamic compensation coefficient during IQ imbalance dynamic compensation can be achieved by freezing the update operation of the coefficient setting module on the IQ imbalance dynamic compensation coefficient when the IQ anomaly determination result is "yes". For example, this operation can be achieved by keeping the current compensation coefficient unchanged until the abnormal state is resolved, or by switching to a conservative fixed compensation mode and recording the abnormal event for subsequent diagnosis. This avoids erroneous compensation due to abnormal data and maintains the stability of signal correction.

[0092] For example, in the scenario of sudden multipath interference in high-speed train communication, the IQ imbalance compensation system based on orthogonal modulation in this embodiment can be as follows: In a 5G high-speed rail coverage scenario, the base station transmit link operates normally while dynamically compensating for IQ imbalance. When the train passes through a tunnel exit, the RF front-end experiences instantaneous nonlinearity due to strong reflection, causing a sharp jump in the IQ imbalance parameters of the mixer region. The anomaly detection module detects that this jump exceeds the preset phase error jump rate limit and is inconsistent with parameters in other regions, thus determining it as an IQ anomaly. The system immediately suspends the adaptive adjustment of the IQ imbalance dynamic compensation coefficients, maintaining the previous stable compensation state to avoid excessive enhancement of high-frequency band compensation due to misjudgment of channel degradation, thereby preventing further distortion of the constellation diagram. After the parameters return to stability, the adaptive mechanism is reactivated to ensure continuous and reliable communication quality in dynamic environments.

[0093] In one embodiment, setting an abnormal IQ jump condition includes: Acquire IQ imbalance parameter data for M regions over N consecutive acquisition cycles, and calculate the cycle-by-cycle change rate of the IQ imbalance parameter for each region. If a change rate threshold is set, and the change rate exceeds the change rate threshold for K consecutive periods in any region, or the change amount in a single instance exceeds the preset maximum allowable deviation, then the IQ anomaly determination result is determined to be yes, where N and K are preset positive integers greater than 1.

[0094] The M regional IQ imbalance parameter data information from N consecutive acquisition cycles can be a set of regional IQ imbalance parameter sequences formed by M orthogonal signal acquisition units acquiring and identifying features within N consecutive sampling cycles in time. This set can be used to provide a time-series data basis for rate of change calculation and anomaly detection. In an exemplary embodiment, the M regional IQ imbalance parameter data information from N consecutive acquisition cycles can be stored in a caching mechanism, forming a sliding window-style historical record. Furthermore, the N consecutive acquisition cycles can be the length of the time window for parameter acquisition and analysis by the system, including N equally spaced sampling times, which can be used to limit the range of historical data on which anomaly detection depends. For example, each region can be any independent region among the M key IQ imbalance influence regions, which can be used as the basic spatial unit for rate of change calculation and anomaly judgment. In a specific embodiment, N can be a preset positive integer greater than 1, representing the number of historical data cycles used for anomaly analysis, which can be used to determine the length of the time window, affecting detection sensitivity and computational cost.

[0095] Obtaining IQ imbalance parameter data for M regions over N consecutive acquisition cycles can be achieved by reading IQ imbalance parameter data for all M regions within the most recent N cycles from a cache. Furthermore, obtaining IQ imbalance parameter data for M regions over N consecutive acquisition cycles can be achieved by using a circular buffer to maintain the latest N-cycle data in real time, thereby constructing a parameter history sequence for time-series analysis. The cycle-by-cycle change rate of the IQ imbalance parameter can be the relative or absolute rate of change of the IQ imbalance parameter between adjacent acquisition cycles within a key IQ imbalance influence region. This can be used to characterize the temporal dynamics of the imbalance state in that region and to identify slow drift or gradual anomalies. In a specific embodiment, the cycle-by-cycle change rate of the IQ imbalance parameter can be obtained by performing a difference operation on the IQ imbalance parameter values ​​of the same region over two consecutive cycles, using either first-order difference or normalized change rate forms. Calculating the cycle-by-cycle change rate of the IQ imbalance parameter for each region can be achieved by sequentially calculating the parameter change rate between adjacent cycles for each region. Furthermore, the periodic rate of change of the IQ imbalance parameter for each region can be calculated by using first-order forward differencing or normalized rate of change, thereby obtaining the dynamic evolution index of the imbalance state in each region.

[0096] The rate of change threshold can be a preset upper limit value used to determine whether the periodic changes in the IQ imbalance parameter are abnormal. It can be used as a criterion for detecting continuous anomalies, distinguishing between normal scenario evolution and potential faults or interference. In an exemplary embodiment, the rate of change threshold can be set based on a device aging model, typical channel dynamic range, or historical statistical distribution. Setting the rate of change threshold can be configured by setting an upper limit value or function for determining whether the rate of change is abnormal. Furthermore, setting the rate of change threshold can be achieved by using a fixed threshold written to the configuration register during system initialization or by dynamically adjusting an adaptive threshold based on the recent standard deviation of the rate of change, thereby establishing a benchmark for continuous anomaly detection. The K consecutive periods can be any continuous subsequence of length K within N acquisition periods, which can be used to detect continuous abnormal trends and avoid misjudgments due to single fluctuations. In a specific embodiment, K can be a preset positive integer greater than 1, representing the minimum number of consecutive out-of-limit periods required to determine continuous anomalies, which can be used to control the response delay and noise immunity to slow drift anomalies.

[0097] Determining whether any region has a rate of change exceeding a threshold for K consecutive periods can be achieved by performing a sliding window scan on the rate of change sequence of each region to detect the existence of consecutive out-of-limit segments with a length ≥ K. Further, determining whether any region has a rate of change exceeding the threshold for K consecutive periods can be achieved by using a state machine to record the consecutive out-of-limit count, triggering a check when K is reached, or by matching consecutive out-of-limit patterns using convolutional kernels. This can identify persistent anomalies such as slow drift or gradual hardware degradation. The single change can be the absolute difference between two adjacent acquisition cycles of the IQ imbalance parameter in a certain region, reflecting the degree of instantaneous jumps and serving as a direct measure of sudden anomalies. The preset maximum permissible deviation can be the maximum absolute change allowed for a single jump in the IQ imbalance parameter, used to capture sudden transient anomalies, such as parameter jumps caused by instantaneous nonlinearity of the RF front end or strong interference. In an exemplary embodiment, the preset maximum permissible deviation can be determined comprehensively based on hardware accuracy limits, ADC quantization noise, and typical interference intensity.

[0098] Determining whether a single change exceeds the preset maximum permissible deviation can be achieved by calculating the absolute difference between the parameters of the current period and the previous period and comparing it with the preset maximum permissible deviation. Furthermore, determining whether a single change exceeds the preset maximum permissible deviation can be achieved by setting independent deviation limits for the amplitude and phase components, or by using Euclidean distance to measure the total jump variable of the complex imbalance parameter, thereby detecting sudden transient anomalies. A "yes" result for obtaining an IQ anomaly determination can be achieved by outputting an anomaly flag when any anomaly condition is met (exceeding limits for K consecutive periods or exceeding the deviation for a single change). Furthermore, a "yes" result for obtaining an IQ anomaly determination can trigger a protection mechanism that pauses subsequent adaptive adjustments.

[0099] Taking the drift of PA characteristics caused by a sudden temperature change in a millimeter-wave base station as an example, the IQ imbalance compensation system based on quadrature modulation in this embodiment can be used in outdoor millimeter-wave AAU equipment where the ambient temperature changes drastically before and after thunderstorms, causing a slow drift in the power amplifier gain. The system collects IQ imbalance parameters every 10ms, with N set to 20 (i.e., a 200ms window) and K set to 5. During the 8th to 13th cycles, the amplitude imbalance rate of the PA region continuously exceeds the set threshold. Although a single jump does not exceed the limit, exceeding the limit for six consecutive cycles triggers an anomaly judgment. The system then suspends the adaptive adjustment of the IQ imbalance dynamic compensation coefficient to avoid misjudging device thermal drift as channel changes caused by user movement and incorrectly enhancing compensation, thus maintaining the stability of the transmitted signal EVM. After the temperature stabilizes, the parameter changes return to the normal range, and the adaptive mechanism resumes operation.

[0100] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. An IQ imbalance compensation system based on quadrature modulation, characterized in that, The system includes: The regional deployment module is used to analyze key modules based on the circuit topology and design parameters of the target quadrature modulation transmission system, determine M key IQ imbalance influence regions, and deploy M quadrature signal acquisition units within the M key IQ imbalance influence regions. The feature recognition module is used to acquire M regional orthogonal baseband signals through the M orthogonal signal acquisition units, perform IQ imbalance feature recognition on the M regional orthogonal baseband signals, and obtain IQ imbalance parameter data information for the M regions. The interference analysis module is used to perform multi-bandwidth interference analysis on the quadrature modulation output signal based on the IQ imbalance parameter data information of the M regions, and to determine the multi-bandwidth IQ imbalance interference parameters. The coefficient setting module is used to establish a multi-bandwidth IQ imbalance adaptive compensation channel, and at the same time set the dynamic compensation coefficient of IQ imbalance according to the transmission requirements of the target communication scenario, the response sensitivity of user equipment and channel environment information. The dynamic compensation module is used to perform compensation analysis on the multi-bandwidth IQ imbalance interference parameters based on the IQ imbalance dynamic compensation coefficient of the multi-bandwidth IQ imbalance adaptive compensation channel, determine the multi-bandwidth IQ adaptive imbalance compensation parameters, and perform dynamic compensation control of the quadrature modulation signal IQ imbalance through the multi-bandwidth IQ adaptive imbalance compensation parameters.

2. The IQ imbalance compensation system based on quadrature modulation as described in claim 1, characterized in that, The determination of the M key IQ imbalance impact areas includes: Based on the circuit topology and design parameters of the target quadrature modulation transmission system, correlation data mining is performed to obtain historical IQ imbalance data of the transmitter and the characteristics of the system power amplifier and mixer module. By combining the historical IQ imbalance data of the transmitter with the characteristics of the system power amplifier and mixer module, a signal offset transmission simulation model is established to simulate the IQ imbalance transmission. Based on the transmitter’s different operating bandwidth and power conditions, the set of IQ imbalance sources is determined. The IQ imbalance propagation simulation model is used to simulate the imbalance propagation of the set of IQ imbalance sources to determine the set of imbalance impact propagation paths. The transmission contribution of the imbalance impact transmission path set is statistically analyzed and key nodes are extracted to obtain the key impact node set of IQ imbalance. Based on the key impact node set of IQ imbalance, the regions are divided and marked to determine the M key IQ imbalance impact regions.

3. The IQ imbalance compensation system based on quadrature modulation as described in claim 1, characterized in that, The obtained IQ imbalance parameter data information for M regions includes: Based on the distribution characteristics of the orthogonal baseband signals in the M regions and the IQ imbalance analysis requirements, the sliding acquisition window length and sliding step size of the orthogonal signals are determined. The orthogonal baseband signals of the M regions are divided into blocks according to the acquisition window length and sliding step size to obtain a set of orthogonal signal blocks of the M regions. Amplitude and phase characteristics are calculated for each signal block in the set of M orthogonal signal blocks in the region to obtain the amplitude difference and phase deviation sequence of the I and Q paths; Imbalance features are extracted and quantified from the amplitude difference and phase deviation sequences of the I and Q paths to obtain the IQ imbalance parameter data information of the M regions.

4. The IQ imbalance compensation system based on quadrature modulation as described in claim 2, characterized in that, The determination of multi-bandwidth IQ imbalance interference parameters includes: Based on the transmitter's operating mode and the operating bandwidth range of the quadrature modulation system, the bandwidth division interval for IQ imbalance analysis is set. According to the bandwidth division interval, the IQ imbalance parameter data information of the M regions is divided into M regional sub-bandwidth IQ imbalance sets. Based on the IQ imbalance propagation simulation model, the M regional bandwidth IQ imbalance sets and the final modulation output port are integrated and propagated to obtain the multi-bandwidth IQ imbalance propagation path. Output interference analysis is performed based on the M regional bandwidth IQ imbalance sets and the multi-bandwidth IQ imbalance propagation paths to determine the multi-bandwidth IQ imbalance interference parameters.

5. The IQ imbalance compensation system based on quadrature modulation as described in claim 4, characterized in that, The determination of multi-bandwidth IQ imbalance interference parameters includes: According to the bandwidth division interval, the historical IQ imbalance data of the transmitter is divided into bandwidth and the transfer deviation is fitted to generate a multi-bandwidth imbalance transfer function. Based on the multi-bandwidth imbalance transfer function, the cumulative deviation of the M regional sub-bandwidth IQ imbalance sets and the multi-bandwidth IQ imbalance transfer paths is calculated to obtain the M regional sub-bandwidth cumulative IQ imbalance sets. Based on the bit error rate performance requirements of the quadrature modulation system, a multi-bandwidth IQ imbalance threshold is set. According to the multi-bandwidth IQ imbalance threshold, the deviation comparison and screening of the M regional multi-bandwidth cumulative IQ imbalance sets and the cumulative deviation sum calculation are performed to obtain the total intensity of multi-bandwidth IQ imbalance. Obtain the IQ imbalance assessment index set, perform interference analysis on the total intensity of the multi-bandwidth IQ imbalance according to the IQ imbalance assessment index set, and determine the interference parameters of the multi-bandwidth IQ imbalance.

6. The IQ imbalance compensation system based on quadrature modulation as described in claim 4, characterized in that, The establishment of a multi-bandwidth IQ imbalance adaptive compensation channel includes: Collect historical IQ compensation adjustment dataset, divide the historical IQ compensation adjustment dataset into bandwidth according to the bandwidth division interval, and obtain multi-bandwidth IQ compensation dataset; The multi-bandwidth IQ compensation dataset is divided and labeled to obtain multi-bandwidth IQ imbalance data, multi-bandwidth IQ adaptive compensation parameters, and corresponding compensation effect data. Based on the compensation effect data, the multi-bandwidth IQ imbalance data and multi-bandwidth IQ adaptive compensation parameters are optimized to obtain an available multi-bandwidth IQ compensation sample set. The available multi-bandwidth IQ compensation sample set is trained in parallel using a deep neural network structure to establish the multi-bandwidth IQ imbalance adaptive compensation channel.

7. The IQ imbalance compensation system based on quadrature modulation as described in claim 4, characterized in that, The setting of the dynamic compensation coefficient for IQ imbalance includes: The transmission requirements of the target communication scenario are quantitatively evaluated according to the bandwidth division intervals to determine the multi-bandwidth requirement weighting factors. Based on the bandwidth division interval, the response sensitivity of the user equipment and the channel environment information are classified and constrained to obtain the multi-bandwidth sensitivity threshold and the multi-bandwidth channel interference ratio. The IQ imbalance dynamic compensation coefficient is determined based on the empirical weighted average of the multi-bandwidth demand weighting factor, the multi-bandwidth sensitivity threshold, and the multi-bandwidth channel interference ratio.

8. The IQ imbalance compensation system based on quadrature modulation as described in claim 1, characterized in that, The determination of multi-bandwidth IQ adaptive imbalance compensation parameters includes: Based on the multi-bandwidth IQ imbalance adaptive compensation channel, the multi-bandwidth IQ imbalance interference parameters are compensated and analyzed, and the basic multi-bandwidth IQ compensation parameters are output. The IQ imbalance dynamic compensation coefficient is used to dynamically correct the basic multi-bandwidth IQ compensation parameters to determine the multi-bandwidth IQ adaptive imbalance compensation parameters.

9. The IQ imbalance compensation system based on quadrature modulation as described in claim 1, characterized in that, After performing dynamic compensation control of the quadrature modulation signal IQ imbalance using the multi-bandwidth IQ adaptive imbalance compensation parameters, the method further includes: Set IQ abnormal jump conditions, and perform abnormal judgment analysis on the IQ imbalance parameter data information of the M regions based on the IQ abnormal jump conditions to obtain IQ abnormal judgment results; If the IQ anomaly determination result is yes, then the adaptive adjustment of the IQ imbalance dynamic compensation coefficient is paused during IQ imbalance dynamic compensation.

10. The IQ imbalance compensation system based on quadrature modulation as described in claim 9, characterized in that, The setting of IQ abnormal jump conditions includes: Acquire IQ imbalance parameter data for the M regions over N consecutive acquisition cycles, and calculate the cycle-by-cycle change rate of the IQ imbalance parameter for each region. A change rate threshold is set. If the change rate of any region exceeds the change rate threshold for K consecutive periods, or the change amount of a single change exceeds the preset maximum allowable deviation, then the IQ anomaly determination result is determined to be yes, where N and K are preset positive integers greater than 1.

Images