Intelligent signal optimization method and system of 5GCPE and medium

Abstract

The invention discloses an intelligent signal optimization method and system of 5GCPE and a medium, and relates to the related technical field of signal optimization, and the method comprises the steps: activating a double-layer collection processing layer, and collecting primary and secondary signals through a 5GCPE terminal to form a collection signal set; performing time sequence prediction on the signal state in combination with historical performance data; analyzing abnormal characteristics by using the secondary acquisition signal, identifying the abnormity of the primary acquisition signal, and correcting a time sequence characteristic prediction result; configuring a signal performance calculation index, and constructing a signal optimization evaluation channel; and executing signal self-adaptive optimization searching to complete intelligent signal optimization management. The technical problems that in the prior art, the signal quality reduction root cause diagnosis capability is insufficient, oscillation risks are caused based on simple threshold rule adjustment, and signal optimization response is lagged are solved, signal performance accurate evaluation, differential regulation and control and multi-parameter collaborative prospective global stable optimization are achieved, and the signal quality is improved. The technical effect of improving the 5G network reliability and the user experience is achieved.

Description

Technical Field

[0001] This invention relates to the field of signal optimization technology, specifically to intelligent signal optimization methods, systems, and media for 5GCPE. Background Technology

[0002] 5G CPEs are critical hubs connecting 5G networks to user-local devices. While 5G networks offer high bandwidth, low latency, and massive connectivity, signal quality in real-world deployments is significantly impacted by factors such as building obstruction, multipath effects, interference source variations, and user mobility, posing a serious challenge to signal stability. Traditional signal optimization relies primarily on fixed parameter configurations or simple threshold adjustments, lacking adaptability to dynamic changes in complex wireless environments and failing to meet users' continuous demands for high-quality, stable network connections. Current 5G CPE signal management lacks in-depth acquisition of signal time-varying characteristics, spectral features, and interference patterns, and only triggers adjustment mechanisms after signal quality degradation is detected, failing to achieve proactive optimization and impacting user experience continuity. Furthermore, key parameters such as antenna beamforming, transmit power, and frequency band selection lack fine-grained adaptive adjustment capabilities based on real-time environmental characteristics, making it impossible to predict potential signal attenuation or increased interference. Additionally, it struggles to differentiate between signal quality degradation caused by external environmental interference, changes in the device's own hardware status, or network-side configuration adjustments, hindering the implementation of precise and differentiated optimization strategies.

[0003] Therefore, current technologies suffer from several technical problems, including insufficient ability to diagnose the root causes of signal quality degradation, the risk of oscillation due to simple threshold rule adjustments, and delayed signal optimization response. Summary of the Invention

[0004] This application provides a smart signal optimization method, system, and medium for 5G CPE, which solves the technical problems of insufficient ability to diagnose the root causes of signal quality degradation, the risk of oscillation caused by simple threshold rule adjustment, and the lag in signal optimization response in the prior art. It achieves accurate evaluation of signal performance, differentiated control, and forward-looking global stable optimization with multi-parameter coordination, thereby improving the technical effect of 5G network reliability and user experience.

[0005] This application provides an intelligent signal optimization method for 5GCPE, the method comprising: activating a dual-layer acquisition and processing layer, performing enhanced data acquisition of the 5GCPE terminal, and establishing an acquisition signal set, the acquisition signal set including primary acquisition signals and secondary acquisition signals; after calling historical performance data, performing time-series prediction of signal state based on the acquisition signal set and the historical performance data, and establishing time-series prediction results at multiple time steps; performing abnormal feature analysis using the secondary acquisition signals in the acquisition signal set, and confirming signal abnormalities of the primary acquisition signals based on the abnormal feature analysis results; performing time-series feature prediction result correction at the corresponding time step based on the confirmation results, configuring signal performance calculation indicators based on the correction results and the acquisition signal set, and constructing a signal optimization evaluation channel using the signal performance calculation indicators; performing adaptive signal optimization using the signal optimization evaluation channel, the adjustment parameters of the adaptive optimization including antenna beam, transmit power, and frequency band switching, and performing intelligent signal optimization management based on the adaptive optimization results.

[0006] In one possible implementation, anomaly feature analysis is performed on the secondary acquisition signals in the acquired signal set, and the signal anomaly of the primary acquisition signal is confirmed based on the anomaly feature analysis results. This includes: constructing a high-resolution signal evolution sequence within a corresponding time window based on the secondary acquisition signals; dividing the high-resolution signal evolution sequence into multiple continuous micro-time units; analyzing the transient evolution relationships of signal amplitude, phase, and subcarrier energy distribution within each continuous micro-time unit, and extracting structural anomaly features characterizing the non-stationarity and abrupt change consistency of the signal; mapping the structural anomaly features to the low-resolution time units corresponding to the primary acquisition signals according to the signal evolution stage correspondence rules, forming a cross-resolution aligned feature group; performing structural distinguishability determination based on the cross-resolution aligned feature group, and using the structural distinguishability determination to confirm the signal anomaly of the primary acquisition signal.

[0007] In one possible implementation, enhanced data acquisition by the 5GCPE terminal includes: after the formation of the primary acquisition signal, performing signal evolution behavior structure analysis within a time window, calculating the discontinuity index of the signal intensity change gradient, the short-term clustering degree of phase jitter, and the local offset trend of subcarrier energy distribution, respectively, and generating a joint analysis result, which characterizes the abnormal evolution tendency of the primary acquisition signal in the time and frequency domains; performing a joint evaluation using the joint analysis result to generate an acquisition trigger indication quantity characterizing the degree of signal anomaly suspicion; and adaptively configuring the acquisition start time, acquisition duration, and acquisition resolution enhancement parameters of the secondary acquisition signal according to the acquisition trigger indication quantity, performing directional enhanced acquisition, and constructing the secondary acquisition signal.

[0008] In one possible implementation, calculating the discontinuity index of the signal intensity change gradient includes: performing segmented gradient calculation on the signal intensity sequence within the time window to obtain the intensity change gradient vector within adjacent time periods; performing joint analysis on the sign consistency and amplitude continuity of the intensity change gradient vector to identify the location of direction reversal or amplitude abrupt change during the gradient change process; and generating a discontinuity index characterizing the discontinuous characteristics of signal intensity evolution based on the degree of gradient consistency disruption.

[0009] In one possible implementation, the calculation of the short-time clustering degree of phase jitter includes: extracting the phase change sequence of a single acquisition signal within the time window, dividing the phase change sequence into multiple continuous short-time analysis units according to a preset short-time analysis scale; calculating the distribution density of phase change within each short-time analysis unit to obtain a local clustering degree parameter characterizing the concentrated occurrence of phase disturbances within a local time range; comparing and analyzing the local clustering degree parameters corresponding to adjacent short-time analysis units to identify concentrated enhancement segments of phase disturbances within a local time range, and constructing the short-time clustering degree.

[0010] In one possible implementation, the calculation of the local offset trend of subcarrier energy distribution includes: within the time window, performing subcarrier-level energy analysis on a single acquisition signal to obtain the subcarrier energy distribution state at each time point; using the subcarrier energy distribution state to determine the energy centroid position of the subcarrier energy distribution at each time point in the frequency domain; performing a continuity analysis of the trajectory of the energy centroid position changing over time to identify the directional offset characteristics of the energy centroid position in the local frequency band, and establishing a local offset trend based on the continuous offset direction and offset amplitude.

[0011] In one possible implementation, signal performance calculation indicators are configured based on the correction results and the acquired signal set. A signal optimization evaluation channel is then constructed using these indicators. This includes: obtaining the corrected time-series feature prediction results based on the correction results; extracting predictive performance features characterizing the future evolution trend of the signal; combining the signal observation results of the first and second acquired signals at different time resolutions in the acquired signal set, applying consistency weighting to the predictive performance features to generate a set of signal performance features with credibility identifiers; using the set of signal performance features as performance calculation indicators, and mapping and combining the performance calculation indicators based on the response sensitivity of the optimized control parameters to construct the signal optimization evaluation channel.

[0012] In one possible implementation, the signal optimization evaluation channel is used to perform adaptive signal optimization, including: based on the multi-dimensional evaluation results output by the signal optimization evaluation channel, constructing parameter response correlations corresponding to antenna beam, transmit power, and frequency band switching respectively, forming a multi-parameter linkage optimization space; within the multi-parameter linkage optimization space, according to the weight distribution of each performance dimension in the signal optimization evaluation channel, performing adaptive search and filtering of parameter combinations to construct a candidate optimization strategy set; verifying the performance trend consistency of the candidate optimization strategy set, and simultaneously optimizing the parameters of antenna beam, transmit power, and frequency band switching according to the verification results to complete the adaptive optimization.

[0013] This application also provides an intelligent signal optimization system for 5GCPE, the system comprising: a signal acquisition module for activating a dual-layer acquisition and processing layer, performing enhanced data acquisition of the 5GCPE terminal, and establishing an acquisition signal set, the acquisition signal set including primary acquisition signals and secondary acquisition signals; a timing prediction module for performing timing prediction of signal states based on the acquisition signal set and the historical performance data after calling historical performance data, and establishing timing prediction results at multiple time steps; an anomaly feature analysis module for performing anomaly feature analysis using the secondary acquisition signals in the acquisition signal set, and confirming signal anomalies in the primary acquisition signals based on the anomaly feature analysis results; an evaluation channel establishment module for correcting the timing feature prediction results at the corresponding time step based on the confirmation results, configuring signal performance calculation indicators based on the correction results and the acquisition signal set, and constructing a signal optimization evaluation channel using the signal performance calculation indicators; and a signal optimization management module for performing adaptive signal optimization using the signal optimization evaluation channel, the adjustment parameters of the adaptive optimization including antenna beam, transmit power, and frequency band switching, and performing intelligent signal optimization management based on the adaptive optimization results.

[0014] This application also provides a computer-readable storage medium, including: a computer program stored thereon that, when executed by a processor, implements a smart signal optimization method for 5GCPE.

[0015] This application proposes a smart signal optimization method, system, and medium for 5G CPE. This involves activating a dual-layer acquisition and processing layer, acquiring primary and secondary signals through a 5G CPE terminal to form a signal set, and using historical performance data to predict signal status in a time-series manner. Abnormal features are analyzed using the secondary acquisition signal to identify anomalies in the primary acquisition signal and correct the time-series feature prediction results. Signal performance calculation indicators are configured to construct a signal optimization evaluation channel. Adaptive signal optimization is then performed to complete intelligent signal optimization management. This addresses the technical problems in existing technologies, such as insufficient ability to diagnose the root causes of signal quality degradation, the risk of oscillation due to simple threshold rule adjustments, and delayed signal optimization response. It achieves accurate signal performance evaluation, differentiated control, and forward-looking, globally stable optimization with multi-parameter collaboration, thereby improving the reliability of 5G networks and user experience. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments of this disclosure will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0017] Figure 1 This is a schematic flowchart of the intelligent signal optimization method for 5GCPE provided in an embodiment of this application.

[0018] Figure 2 This is a schematic diagram of the intelligent signal optimization system structure of 5GCPE provided in an embodiment of this application.

[0019] Figure labeling: Signal acquisition module 10, time series prediction module 20, anomaly feature analysis module 30, evaluation channel establishment module 40, signal optimization management module 50. Detailed Implementation

[0020] To further illustrate the technical means and effects adopted by the present invention in order to achieve the intended purpose, the following detailed description is provided in conjunction with the accompanying drawings and preferred embodiments, based on the specific implementation methods, structures, features and effects of the present invention.

[0021] This application provides an intelligent signal optimization method for 5GCPE, such as... Figure 1 As shown, the method includes: Step S100: Activate the dual-layer acquisition and processing layer, perform enhanced data acquisition of the 5GCPE terminal, and establish an acquisition signal set, which includes primary acquisition signals and secondary acquisition signals.

[0022] Preferably, the dual-layer acquisition and processing layer is a dual-layer acquisition and processing unit constructed by calling hardware such as the baseband processor and RF front-end, as well as signal processing firmware, for multi-modal signal acquisition. It includes a basic acquisition layer and an enhanced acquisition layer. The basic acquisition layer is responsible for routine periodic signal parameter measurements, while the enhanced acquisition layer is triggered under specific conditions to perform high-resolution, multi-dimensional signal acquisition. Activating the dual-layer acquisition and processing layer enables enhanced data acquisition by the 5GCPE terminal. Specifically, the basic acquisition layer acquires a signal at a fixed period, such as once per second, or at a low sampling rate, to monitor the macroscopic state of the signal, including at least the Received Signal Strength Indicator (RSSI), Reference Signal Received Power (RSRP), and Reference Signal Gain. The system receives physical layer measurements such as RSRQ and SNR, basic time-frequency domain parameters such as signal bandwidth, subcarrier spacing, and cyclic prefix length, as well as connection status information such as radio resource control status and transmission mode. When the primary acquisition signal exhibits abnormal trends such as sudden changes in signal strength or sharp drops in quality, the enhanced acquisition layer is dynamically activated. After triggering, it continuously acquires secondary acquisition signals at a high sampling rate for a short period of time to obtain the secondary acquisition signal. This is used for in-depth analysis of abnormal signal structures, including but not limited to high-resolution time-domain waveforms for analyzing transient changes in signal amplitude / phase, fine frequency-domain characteristics such as power distribution, phase noise spectrum, and frequency offset of each subcarrier, and multi-antenna spatial characteristics such as channel impulse response and direction-of-arrival estimation.

[0023] Furthermore, step S100 also includes, after the formation of the first acquisition signal, performing signal evolution behavior structure analysis within a time window, calculating the discontinuity index of the signal intensity change gradient, the short-term clustering degree of phase jitter, and the local offset trend of subcarrier energy distribution, respectively, and generating a joint analysis result, the joint analysis result characterizing the abnormal evolution tendency of the first acquisition signal in the time and frequency domains; using the joint analysis result to perform joint evaluation, generating an acquisition trigger indication quantity characterizing the degree of signal anomaly suspicion; based on the acquisition trigger indication quantity, adaptively configuring the acquisition start time, acquisition duration, and acquisition resolution enhancement parameters of the second acquisition signal, performing directional enhancement acquisition, and constructing the second acquisition signal.

[0024] Preferably, after acquiring the first acquisition signal, it is analyzed in real time, and a high-resolution second acquisition signal is adaptively triggered and configured. Specifically, the first acquisition signal is analyzed within a fixed time window, and the discontinuity index of the signal strength change gradient, the short-term clustering degree of phase jitter, and the local offset trend of subcarrier energy distribution are calculated to capture abnormal tendencies in the signal's microstructure. The fixed time window may be the most recent 10 sampling points or 100 milliseconds of data. Calculating the discontinuity index of the signal strength change gradient analyzes the smoothness of the received signal strength change over time. Ideally, the signal change is relatively smooth, while sudden drops or rises may indicate obstruction, interference, or equipment switching. The phase jitter is calculated... The short-time clustering degree is used to analyze the short-term stability of the signal phase. Phase noise is usually randomly distributed. If the phase changes rapidly within a short time, it may indicate the presence of sudden pulse interference or hardware transient faults. Calculating the local offset trend of subcarrier energy distribution is used to analyze the stability of the signal's energy distribution in the frequency domain. In OFDM systems, the energy of each subcarrier should be relatively balanced or distributed according to a known pattern. A continuous offset of the energy centroid in the frequency domain may indicate the presence of frequency-selective fading or narrowband interference. Then, the discontinuity index, short-time clustering degree, and local offset trend are combined into a multi-dimensional feature vector as a joint analytical result to characterize the potential and structurally abnormal evolutionary tendencies of a single-acquired signal in the time and frequency domains.

[0025] Preferably, the joint analysis results are input into the evaluation classifier to perform joint evaluation, which is used to quantify the degree of suspicion of signal anomaly and determine whether to initiate secondary acquisition. The evaluation classifier calculates the acquisition trigger indication based on predefined or self-learned rules for the current signal state, which comprehensively reflects the severity and combination pattern of the anomaly in the three dimensions. The higher the acquisition trigger indication, the greater the possibility of complex and structural anomalies in the signal, and the more necessary it is to initiate secondary acquisition for in-depth diagnosis.

[0026] Preferably, based on the acquisition trigger indication, the acquisition start time, acquisition duration, and acquisition resolution enhancement parameters of the secondary acquisition signal are adaptively configured to achieve accurate capture. Specifically, the acquisition start time of the secondary acquisition signal is configured according to the time and location of the abnormal feature. For example, if the phase convergence is highest at the end of the window, it is started immediately; if it is in the middle of the window, it may be slightly delayed in combination with prediction to capture the key period. The acquisition duration is configured according to the persistence of the abnormal feature and the magnitude of the acquisition trigger indication. For example, for short-lived pulse interference, the acquisition duration may cover the pulse; for continuous fading, the acquisition time is longer to observe the evolution process. The configuration of acquisition resolution enhancement parameters mainly includes increasing the sampling rate from the lower sampling rate of the basic acquisition to a higher multiple to obtain more detailed time-domain waveforms, temporarily enabling higher bit-level analog-to-digital conversion to reduce quantization noise, and allocating more buffer space to store high sampling rate data. After configuration, the RF front-end and baseband processing unit are dynamically reconfigured, and directional enhancement acquisition is performed at a specified time with set high resolution parameters to capture the specific performance of a specific abnormal pattern in the primary acquisition signal at a higher resolution, thereby obtaining the secondary acquisition signal.

[0027] Furthermore, step S100 also includes, within the time window, performing segmented gradient calculation on the signal intensity sequence to obtain intensity change gradient vectors within adjacent time periods; performing joint analysis on the sign consistency and amplitude continuity of the intensity change gradient vectors to identify the locations of direction reversals or amplitude abrupt changes during gradient changes, and generating a discontinuity index characterizing the discontinuous characteristics of signal intensity evolution based on the degree of gradient consistency disruption.

[0028] Preferably, the signal intensity sequence within the entire fixed time window is divided into multiple continuous and non-overlapping or partially overlapping time periods. Linear fitting is performed on the intensity values ​​within each time period, and the slope of the fitted line is used as the average rate of change of signal intensity within that time period to determine the segmented gradient. This yields the intensity change gradient vector between adjacent time periods, where each element represents the average rate of change of the current time period. Next, a joint analysis is performed on the sign consistency and amplitude continuity of the intensity change gradient vector. This includes checking the sign of adjacent gradients; if the current time period is positive (intensity is increasing) and the next time period is negative (intensity is decreasing), it indicates that the direction of signal intensity change reverses at the boundary between the two time periods. Simultaneously, the relative rate of change or absolute difference of adjacent gradient amplitudes is calculated. If the relative rate of change or absolute difference between adjacent time periods exceeds a preset amplitude abrupt change threshold, it indicates that a sudden change in gradient amplitude has occurred. Finally, all locations of sign inconsistency or amplitude abrupt changes are identified as potential gradient consistency violation points, corresponding to moments in the original signal intensity sequence where abnormal abrupt changes may occur. Finally, each identified gradient consistency violation point is scored, and the scores of all gradient consistency violation points within a fixed time window are accumulated to determine the total score. The total score is then normalized by dividing by the maximum possible number of violation points within the window or by an empirical constant to generate a discontinuity index that characterizes the discontinuous characteristics of signal intensity evolution. The higher the discontinuity index value, the more unstable the signal intensity change and the worse the continuity within that time window.

[0029] Furthermore, step S100 also includes: within the time window, extracting the phase change sequence of a single acquisition signal, dividing the phase change sequence into multiple continuous short-time analysis units according to a preset short-time analysis scale; within each short-time analysis unit, calculating the distribution density of the phase change quantity, and obtaining a local clustering parameter characterizing the concentrated occurrence of phase disturbances within a local time range; comparing and analyzing the local clustering parameters corresponding to adjacent short-time analysis units, identifying concentrated enhancement segments of phase disturbances within a local time range, and constructing short-time clustering.

[0030] Preferably, the phase change sequence refers to the phase difference between adjacent sampling points, reflecting the instantaneous fluctuation of the phase. The phase change sequence within the entire fixed time window is divided into multiple continuous, non-overlapping or partially overlapping short-time analysis units according to a preset short-time analysis scale, such as one unit every 5 milliseconds. This allows observation of the statistical characteristics of phase perturbations within a local time frame. Within each short-time analysis unit, the distribution density of phase changes is calculated; that is, for all phase change values ​​within each short-time analysis unit, the variance or standard deviation is calculated to determine the statistical dispersion. This generates a local clustering parameter for each short-time analysis unit, characterizing the concentrated occurrence of phase perturbations within a local time range. A higher clustering parameter value indicates a more concentrated distribution of phase change values ​​within the short-time analysis unit, meaning a greater likelihood of clustered phase perturbations. A comparative analysis of the local clustering parameters corresponding to adjacent short-time analysis units is performed to identify whether there is a significant local increase in the local clustering parameter value on the time axis, i.e., whether one or several consecutive short-time analysis units have clustering parameters significantly higher than their surrounding background units. Then, by setting a threshold or using change point detection, concentrated enhancement segments where all phase perturbations exceed the set threshold within a local time range are identified and determined. Finally, the short-time clustering degree is determined, which may include the maximum clustering degree, the total duration of the clustering segment, the integral of the clustering intensity, and the suddenness of the clustering.

[0031] Furthermore, step S100 also includes performing subcarrier-level energy analysis on the acquired signal within the time window to obtain the subcarrier energy distribution state at each time point; using the subcarrier energy distribution state to determine the energy centroid position of the subcarrier energy distribution at each time point in the frequency domain; performing continuity analysis on the trajectory of the energy centroid position changing over time to identify the directional shift characteristics of the energy centroid position in the local frequency band, and establishing a local shift trend based on the continuous shift direction and shift amplitude.

[0032] Preferably, within a fixed time window, subcarrier-level energy analysis is performed on the time-domain data of a single acquisition signal. This involves performing a Fast Fourier Transform on the time-domain signal at each sampling moment to calculate the energy on each OFDM subcarrier at each moment, thus obtaining the subcarrier energy distribution state at each time point. Using the subcarrier energy distribution state, the energy centroid frequency at each time point is calculated and weighted averaged to determine the energy centroid position of the subcarrier energy distribution at each time point in the frequency domain. Here, the energy centroid represents the average position of the signal energy in the frequency domain. If all subcarrier energies are equal, the centroid is at the center of the spectrum; if the energy in the high-frequency part is stronger, the centroid moves towards the high-frequency direction; and vice versa. Then, all energy centroid positions are arranged in chronological order, and the changes in centroid positions at adjacent time points are calculated to determine their direction and amplitude. The directional shift characteristics of the energy centroid positions within a local frequency band are identified, including checking whether the signs of the changes in centroid positions at adjacent time points are consistent over several consecutive time points, and checking whether the magnitude of the shift is relatively stable or consistent. If the signs are consistent and the cumulative shift exceeds the noise threshold in multiple consecutive sampling points, it indicates the existence of directional shift characteristics. Then, the main sign of the local time period is taken, with positive indicating a shift to higher frequencies and negative indicating a shift to lower frequencies, and the total change in centroid positions within the local time period is calculated as the shift amplitude. Finally, the local shift trend is output.

[0033] Step S200: After calling historical performance data, perform time-series prediction of signal state based on the acquired signal set and the historical performance data, and establish time-series prediction results at multiple time steps.

[0034] Preferably, time-series data related to signal performance stored in the 5GCPE local or cloud database over a past period are acquired to determine historical performance data, which may include historical sequences of signal strength, throughput, bit error rate, and latency, historical records of device operating parameters, and environmental label data. Based on the acquired signal set and historical performance data, a time-series prediction of the signal state is performed. Specifically, the historical performance data is aligned with the current acquired signal set on a timeline and spliced ​​into a complete data sequence extending from the past to the present. Periodic and trend features are extracted from the historical data, and dynamic features, especially anomalous features parsed from secondary acquired signals, are extracted from the current acquired signal set to form a high-dimensional feature vector as input to the prediction model. The prediction model is constructed and trained based on a long short-term memory network. The fused feature sequence is input into the trained prediction model, which outputs a signal state prediction sequence for multiple future time steps. This sequence describes the most likely evolution trajectory of signal performance assuming the current environment and device parameters remain unchanged. The final output is a time-series prediction result, which includes at least a sequence of predicted signal state estimates, the fluctuation range of the predicted values, and the time steps in the predicted trajectory where significant changes occur.

[0035] Preferably, multivariate time series are extracted from historical performance data. The sliding window length is determined using a historical time step of 10, and the input sequence is determined. The output sequence is a continuous sequence with a length of 5 prediction time steps following each input sequence. Features of the continuous sequence in each time step sample are extracted. Each feature is Z-score standardized and linear interpolation or forward padding is used. The set is divided into training set (70%), validation set (15%), and test set (15%) according to time order. Mean squared error or mean absolute error is used as the loss function to measure the difference between the predicted sequence and the real sequence. The Adam optimizer is used, and its adaptive learning rate characteristic is suitable for sequence data. Then, the input sequence is input into the LSTM network to obtain the predicted output and compare it with the real output sequence to calculate the loss. The gradient is calculated through time backpropagation. At the same time, the network weights are updated using the optimizer. The performance is monitored on the validation set. When the validation loss no longer decreases for several consecutive cycles, early stopping is triggered, and the model parameters with the best performance on the validation set are saved. The LSTM time-series prediction model employs an encoder-decoder structure. The encoder consists of 64 or 128 stacked LSTM units, responsible for encoding the historical signal state sequence into multiple hidden states and the final context state vector. The decoder uses independent 2-4 LSTM layers, initialized with the encoder's final hidden state, and operates in an autoregressive manner. The prediction output of each step serves as the input for the next step, gradually generating predictions for multiple future time steps. Finally, the high-dimensional hidden state of the decoder's LSTM is mapped to the target prediction dimension to determine the time-series prediction results for multiple time steps.

[0036] Step S300: Perform abnormal feature analysis on the secondary acquisition signals in the acquisition signal set, and confirm the signal abnormality of the primary acquisition signal based on the abnormal feature analysis results.

[0037] Step S300 further includes: based on the secondary acquisition signal, constructing a high-resolution signal evolution sequence within the corresponding time window, dividing the high-resolution signal evolution sequence into multiple continuous micro-time units; within each continuous micro-time unit, analyzing the transient evolution relationship of signal amplitude, phase, and subcarrier energy distribution, and extracting structural anomaly features characterizing the non-stationarity and abrupt change consistency of the signal; mapping the structural anomaly features to the low-resolution time unit corresponding to the primary acquisition signal according to the signal evolution stage correspondence rules, forming a cross-resolution aligned feature group; performing structural discriminability determination based on the cross-resolution aligned feature group, and using the structural discriminability determination to confirm the signal anomaly of the primary acquisition signal.

[0038] Preferably, the anomalous feature analysis is performed on the secondary acquisition signal in the acquisition signal set, including detecting and quantifying pulse, clipping, and oscillation attenuation on the I / Q waveform within the corresponding time window, extracting transient time-domain features such as pulse width, peak amplitude, rise / fall time, and oscillation frequency, performing high-resolution spectrum analysis on the signal, and extracting the center frequency, bandwidth, and power spectral density of narrowband interference, the floor rise of broadband noise, and the depth and location of the spectral dip; then, the extracted multiple fine features are combined into a structured high-resolution signal evolution sequence to describe the state of signal anomalies in the time and frequency domains; then, the high-resolution signal evolution sequence is divided into multiple continuous micro-time units, the duration of each continuous micro-time unit being much shorter than the sampling interval of one acquisition.

[0039] Preferably, within each micro-time unit, joint time-frequency analysis is performed to capture the dynamic coupling and evolution relationship between amplitude, phase, and subcarrier energy distribution. This includes amplitude-phase joint analysis to identify whether the phase undergoes synchronous and specific distortion when the amplitude changes rapidly, in order to determine amplifier saturation and pulse interference; analyzing the transient evolution of subcarrier energy distribution to observe whether the energy changes of different subcarriers within the micro-time unit are a synchronous rise / fall or a drastic change in some subcarriers, in order to distinguish between broadband and narrowband interference; and further extracting structural anomaly features characterizing the non-stationarity and abrupt change consistency of the signal. The non-stationarity feature quantifies the rate of change of the signal's statistical properties within the micro-time unit, for example, by calculating the local Hearst exponent or time-varying spectral entropy of the amplitude envelope within the unit. High non-stationarity indicates that the signal is in a drastic transient process. The abrupt change consistency feature quantifies the degree of coordination of amplitude, phase, and multiple subcarriers undergoing abrupt changes at the same time, for example, by calculating a cross-dimensional abrupt change correlation matrix. High consistency indicates that strong interference pulses simultaneously affect all dimensions of the signal.

[0040] Preferably, the structural anomaly features are mapped to the low-resolution time unit corresponding to the first acquisition signal according to the signal evolution stage correspondence rule. That is, the micro time unit of the second acquisition is matched with the low-resolution time unit of the first acquisition on the time scale. The 1 second duration corresponding to the sampling point of the first acquisition is defined as a low-resolution time unit. All features of the micro time units falling within 1 second are considered to be related to the sampling point, thus forming a cross-resolution aligned feature group. That is, each macro unit of the first acquisition is associated with a set of micro structural features of the second acquisition arranged in chronological order. Then, based on the cross-resolution aligned feature group, structural distinguishability determination is performed. This includes analyzing all micro-unit features associated with each sampling point from the first acquisition. If all micro-unit features have no obvious anomalies or the anomalies are weak, random, and cannot be associated with the anomaly patterns from the first acquisition, the structure is determined to be indistinguishable. If, within the time period corresponding to the macro-unit, there are one or more micro-units whose structural anomalies significantly and continuously deviate from the normal background and whose patterns can reasonably explain the anomaly types observed in the first acquisition, the structure is determined to be distinguishable. This formally confirms that the anomaly in the first acquisition signal at that time point truly exists, and its root cause is the specific structural anomaly resolved by the second acquisition, such as impulse interference or frequency-selective fading. Finally, the signal anomaly of the first acquisition signal is confirmed.

[0041] Step S400: Based on the confirmation result, perform time series feature prediction result correction at the corresponding time step, configure signal performance calculation index based on the correction result and the acquired signal set, and construct a signal optimization evaluation channel using the signal performance calculation index.

[0042] Step S400 further includes: after obtaining the corrected time-series feature prediction result based on the correction result, extracting prediction performance features that characterize the future evolution trend of the signal; combining the signal observation results of the first and second acquisition signals in the acquisition signal set at different time resolutions, performing consistency weighting on the prediction performance features to generate a set of signal performance features with credibility identifiers; using the set of signal performance features as performance calculation indicators, mapping and combining the performance calculation indicators based on the response sensitivity of the optimized control parameters to construct a signal optimization evaluation channel.

[0043] Preferably, based on the confirmation result, the temporal feature prediction result at the corresponding time step is corrected. This involves explicitly adding a label representing interference to the features of the prediction model, or using a simplified physical model to directly superimpose a fixed attenuation based on the interference intensity onto the predicted SINR value for the future duration of the anomaly. The corrected temporal feature prediction result is then output, more accurately reflecting the future signal evolution trajectory under the continued influence of the currently known anomaly. Next, the corrected temporal feature prediction result is obtained, and predictive performance features characterizing the future evolution trend of the signal are extracted. These may include the minimum and average predicted SINR values ​​within the future window, the start time of performance degradation, and the degradation rate. The slope, the expected bottoming time, and the maximum value of future uncertainty are calculated. Then, the extracted predictive performance features are compared with the signal observation results containing similar or related features directly calculated from the first and second acquisition signals. If they are highly consistent, the predictive performance feature is given a high confidence weight. If there is a contradiction, its confidence weight is dynamically reduced according to the inherent reliability and historical accuracy of the data source. The predictive performance features are then weighted for consistency, and a set of signal performance features with confidence labels is output. Each predictive performance feature is accompanied by a weight value [0, 1], which represents its reliability in the decision-making at the current moment. Using the signal performance feature set as a performance calculation index, and based on the current signal performance feature set, its reliability weights, and the optimization objective, the response sensitivity of the optimization control parameters is used to describe the expected impact of each optimization control parameter on each performance feature. Then, a dynamic weighted combination is used to evaluate the comprehensive evaluation function of the three optimization actions: antenna beam, transmit power, and frequency band switching. The weights are global weights that are dynamically adjusted according to the current optimization strategy. Finally, a signal optimization evaluation channel is constructed, which can comprehensively reflect the current state of the signal, the predicted future trend, and anomaly diagnosis information, while quantifying the expected impact of different optimization actions on the overall state.

[0044] Step S500: Use the signal optimization evaluation channel to perform adaptive signal optimization. The adjustment parameters of the adaptive optimization include antenna beam, transmit power, and frequency band switching. Perform intelligent signal optimization management based on the adaptive optimization results.

[0045] Step S500 further includes: based on the multi-dimensional evaluation results output by the signal optimization evaluation channel, constructing parameter response correlations corresponding to antenna beam, transmit power, and frequency band switching respectively, forming a multi-parameter linkage optimization space; within the multi-parameter linkage optimization space, performing adaptive search and filtering of parameter combinations according to the weight distribution of each performance dimension in the signal optimization evaluation channel, and constructing a candidate optimization strategy set; verifying the performance trend consistency of the candidate optimization strategy set, and simultaneously optimizing the parameters of antenna beam, transmit power, and frequency band switching according to the verification results to complete adaptive optimization.

[0046] Preferably, the adaptive optimization adjustment parameters include antenna beam, transmit power, and frequency band switching. Based on the multi-dimensional evaluation results output from the signal optimization evaluation channel, a parameter response correlation is constructed corresponding to the adaptive optimization adjustment parameters. Specifically, for the antenna beam, adjustable parameters such as beam direction angle, downtilt angle, and beamwidth are discretized into a finite set of beam patterns supported by the device. For transmit power, the continuous power adjustment range is discretized into finite power levels. For frequency band switching, all available candidate frequency bands / carriers are listed, and the Cartesian product of these three discrete sets constitutes a complete multi-parameter linkage. The optimization space is then optimized. Next, within the multi-parameter linkage optimization space, based on the weight distribution of each performance dimension in the signal optimization evaluation channel, an adaptive search and selection of parameter combinations is performed. For example, if the current weight emphasizes connectivity preservation, indicating a very high SINR weight, then the search bias can greatly improve SINR and temporarily tolerate high power consumption. Based on the genetic algorithm / particle swarm optimization algorithm, the initial strategy of antenna beam, transmit power, and frequency band switching combination is encoded as a gene or particle position. Through iterative selection, crossover, mutation, or evolution by following the optimal particle, excellent strategies are obtained, resulting in a set of candidate optimization strategies with excellent predicted performance in the signal optimization evaluation channel.

[0047] Preferably, the performance trend consistency of the candidate optimization strategy set is verified. Specifically, it checks whether the strategy violates basic physical constraints. For example, if a candidate optimization strategy suggests reducing the power to the minimum while pointing the beam away from the base station, but the predicted SINR is high, then the strategy should be eliminated. Using historical data or empirical rules, it verifies whether the adjustment direction suggested by the strategy is reasonable. At the same time, it checks whether the candidate optimization strategy introduces new anomalies. For example, a significant increase in power may violate radiation standards or cause serious interference to neighboring cells, and should be rejected or corrected. Thus, a reliable candidate optimization strategy set is output as the verification result. Based on the verification result, the parameters of antenna beam, transmit power, and frequency band switching are simultaneously optimized. That is, through the collaborative configuration interface of the device driver layer or RF controller, the new beamforming weights, new transmit power levels, and new frequency band carrier configurations are simultaneously sent out within a very short time window or the same control cycle. After the device parameters are successfully switched, this adaptive optimization is completed. The adaptive optimization results are converted into register configuration values ​​or API call instructions for devices such as RF front-end, baseband processor, and antenna array. A unified device management coordinator ensures that configuration instructions for antenna, power, and frequency band take effect synchronously within the blank interval of the same radio frame. This minimizes momentary connection interruptions, signal jitter, or protocol state conflicts caused by step-by-step parameter adjustments. After the instructions are issued, the hardware status register is read in real time or the instantaneous signal is measured to confirm whether the new beam, power, and frequency band have taken effect as expected. If the execution fails, a rollback mechanism or backup strategy will be triggered to ensure the reliability of the 5G network and the user experience.

[0048] In the above text, refer to Figure 1 The intelligent signal optimization method for 5GCPE according to embodiments of the present invention is described in detail. Next, reference will be made to... Figure 2 A smart signal optimization system for a 5GCPE according to an embodiment of the present invention is described.

[0049] The intelligent signal optimization system for 5G CPE according to embodiments of the present invention addresses the technical problems in existing technologies, such as insufficient ability to diagnose the root causes of signal quality degradation, oscillation risks caused by simple threshold rule adjustments, and delayed signal optimization response. It achieves accurate signal performance evaluation, differentiated control, and forward-looking global stable optimization through multi-parameter collaboration, thereby improving the technical effects of 5G network reliability and user experience. Figure 2 As shown, the intelligent signal optimization system of 5GCPE includes: a signal acquisition module 10, a timing prediction module 20, an anomaly feature analysis module 30, an evaluation channel establishment module 40, and a signal optimization management module 50.

[0050] The signal acquisition module 10 is used to activate the dual-layer acquisition and processing layer, perform enhanced data acquisition of the 5GCPE terminal, and establish an acquisition signal set, which includes primary acquisition signals and secondary acquisition signals. The timing prediction module 20 is used to perform timing prediction of signal status based on the acquisition signal set and the historical performance data after calling historical performance data, and establish timing prediction results at multiple time steps. The anomaly feature analysis module 30 is used to perform anomaly feature analysis using the secondary acquisition signals in the acquisition signal set, and confirm the signal anomaly of the primary acquisition signal based on the anomaly feature analysis results. The evaluation channel establishment module 40 is used to correct the timing feature prediction results at the corresponding time step based on the confirmation results, configure signal performance calculation indicators based on the correction results and the acquisition signal set, and construct a signal optimization evaluation channel using the signal performance calculation indicators. The signal optimization management module 50 is used to perform adaptive signal optimization using the signal optimization evaluation channel, where the adjustment parameters of the adaptive optimization include antenna beam, transmit power, and frequency band switching, and perform intelligent signal optimization management based on the adaptive optimization results.

[0051] The specific configuration of the anomaly feature analysis module 30 will be described in detail below. The anomaly feature analysis module 30 further includes: constructing a high-resolution signal evolution sequence within a corresponding time window based on the secondary acquisition signal; dividing the high-resolution signal evolution sequence into multiple continuous micro-time units; analyzing the transient evolution relationship of signal amplitude, phase, and subcarrier energy distribution within each continuous micro-time unit; extracting structural anomaly features characterizing the non-stationarity and abrupt change consistency of the signal; mapping the structural anomaly features to the low-resolution time unit corresponding to the primary acquisition signal according to the signal evolution stage correspondence rules, forming a cross-resolution aligned feature group; performing structural discriminability determination based on the cross-resolution aligned feature group, and using the structural discriminability determination to confirm the signal anomaly of the primary acquisition signal.

[0052] The specific configuration of the signal acquisition module 10 will be described in detail below. The signal acquisition module 10 further includes: after the formation of the primary acquisition signal, performing signal evolution behavior structure analysis within a time window, calculating the discontinuity index of the signal intensity change gradient, the short-term clustering degree of phase jitter, and the local offset trend of subcarrier energy distribution, respectively, and generating a joint analysis result, which characterizes the abnormal evolution tendency of the primary acquisition signal in the time and frequency domains; performing a joint evaluation using the joint analysis result to generate an acquisition trigger indication quantity characterizing the degree of signal anomaly suspicion; and adaptively configuring the acquisition start time, acquisition duration, and acquisition resolution enhancement parameters of the secondary acquisition signal according to the acquisition trigger indication quantity, performing directional enhancement acquisition, and constructing the secondary acquisition signal.

[0053] The specific configuration of the signal acquisition module 10 will be described in detail below. The signal acquisition module 10 further includes: performing segmented gradient calculation on the signal intensity sequence within the time window to obtain the intensity change gradient vector within adjacent time periods; jointly analyzing the sign consistency and amplitude continuity of the intensity change gradient vector to identify the locations of direction reversals or amplitude abrupt changes during the gradient change process; and generating a discontinuity index characterizing the discontinuous characteristics of signal intensity evolution based on the degree of gradient consistency violation.

[0054] The specific configuration of the signal acquisition module 10 will be described in detail below. The signal acquisition module 10 further includes: extracting the phase change sequence of a single acquired signal within the time window; dividing the phase change sequence into multiple continuous short-time analysis units according to a preset short-time analysis scale; calculating the distribution density of phase change quantities within each short-time analysis unit to obtain a local clustering parameter characterizing the concentrated occurrence of phase disturbances within a local time range; comparing and analyzing the local clustering parameters corresponding to adjacent short-time analysis units to identify concentrated enhancement segments of phase disturbances within a local time range and constructing short-time clustering.

[0055] The specific configuration of the signal acquisition module 10 will be described in detail below. The signal acquisition module 10 further includes: performing subcarrier-level energy analysis on a single acquired signal within the time window to obtain the subcarrier energy distribution state at each time point; using the subcarrier energy distribution state to determine the energy centroid position of the subcarrier energy distribution at each time point in the frequency domain; performing continuity analysis of the trajectory of the energy centroid position changing over time to identify the directional shift characteristics of the energy centroid position in the local frequency band, and establishing a local shift trend based on the continuous shift direction and shift amplitude.

[0056] The specific configuration of the evaluation channel establishment module 40 will be described in detail below. The evaluation channel establishment module 40 further includes: after obtaining the corrected time-series feature prediction results based on the correction results, extracting predictive performance features characterizing the future evolution trend of the signal; combining the signal observation results of the first and second acquisition signals in the acquisition signal set at different time resolutions, performing consistency weighting on the predictive performance features to generate a set of signal performance features with a credibility identifier; using the set of signal performance features as performance calculation indicators, mapping and combining the performance calculation indicators based on the response sensitivity of the optimized control parameters, and constructing a signal optimization evaluation channel.

[0057] The specific configuration of the signal optimization management module 50 will be described in detail below. The signal optimization management module 50 further includes: based on the multi-dimensional evaluation results output by the signal optimization evaluation channel, constructing parameter response correlations corresponding to antenna beam, transmit power, and frequency band switching respectively, forming a multi-parameter linkage optimization space; within the multi-parameter linkage optimization space, performing adaptive search and filtering of parameter combinations according to the weight distribution of each performance dimension within the signal optimization evaluation channel, constructing a candidate optimization strategy set; verifying the performance trend consistency of the candidate optimization strategy set, and synchronously optimizing the parameters of antenna beam, transmit power, and frequency band switching based on the verification results to complete adaptive optimization.

[0058] The intelligent signal optimization system for 5GCPE provided in this embodiment of the invention can execute the intelligent signal optimization method for 5GCPE provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method execution.

[0059] Based on the foregoing embodiments, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the 5GCPE intelligent signal optimization method as described in any of the preceding embodiments.

[0060] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

The intelligent signal optimization method for 1.5GCPE is characterized by, The method includes: Activate the dual-layer acquisition and processing layer, perform enhanced data acquisition on the 5GCPE terminal, and establish an acquisition signal set, which includes primary acquisition signals and secondary acquisition signals; After calling historical performance data, time-series prediction of signal state is performed based on the acquired signal set and the historical performance data to establish time-series prediction results at multiple time steps; Anomaly feature analysis is performed using the secondary acquisition signals from the acquired signal set, and the signal anomaly of the primary acquisition signal is confirmed based on the anomaly feature analysis results. Based on the confirmation result, the timing feature prediction result at the corresponding time step is corrected. Based on the correction result and the acquired signal set, signal performance calculation indicators are configured, and signal optimization evaluation channels are constructed using the signal performance calculation indicators. The signal optimization evaluation channel is used to perform adaptive signal optimization. The adjustment parameters of the adaptive optimization include antenna beam, transmit power, and frequency band switching. Intelligent signal optimization management is performed based on the adaptive optimization results.

2. The intelligent signal optimization method for 5GCPE as described in claim 1, characterized in that, Anomaly feature analysis is performed on the secondary acquisition signals from the acquired signal set. Based on the anomaly feature analysis results, signal anomalies in the primary acquisition signals are confirmed, including: Based on the secondary acquisition signal, a high-resolution signal evolution sequence is constructed within the corresponding time window, and the high-resolution signal evolution sequence is divided into multiple continuous micro-time units; Within each continuous micro-time unit, the transient evolution relationship of signal amplitude, phase, and subcarrier energy distribution is analyzed, and structural anomaly features characterizing the non-stationarity and abrupt change consistency of the signal are extracted; The structural anomaly features are mapped to the low-resolution time unit corresponding to a single acquisition signal according to the signal evolution stage correspondence rules, forming a cross-resolution aligned feature group; Based on the cross-resolution aligned feature group, structural distinguishability determination is performed, and the structural distinguishability determination is used to confirm the signal anomaly of a single acquired signal.

3. The intelligent signal optimization method for 5GCPE as described in claim 1, characterized in that, Perform enhanced data acquisition on the 5GCPE terminal, including: After the first acquisition signal is formed, the signal evolution behavior structure analysis under the time window is performed. The discontinuity index of the signal intensity change gradient, the short-term clustering degree of phase jitter and the local offset trend of subcarrier energy distribution are calculated respectively, and a joint analysis result is generated. The joint analysis result characterizes the abnormal evolution tendency of the first acquisition signal in the time and frequency domains. The joint analysis results are used to perform a joint evaluation and generate a collection trigger indication quantity that characterizes the degree of suspicion of signal anomalies; Based on the acquisition trigger indication, the acquisition start time, acquisition duration, and acquisition resolution enhancement parameters of the secondary acquisition signal are adaptively configured, directional enhancement acquisition is performed, and the secondary acquisition signal is constructed.

4. The intelligent signal optimization method for 5GCPE as described in claim 3, characterized in that, The discontinuity index for calculating the gradient of signal intensity change includes: Within the time window, the signal intensity sequence is segmented and gradient calculation is performed to obtain the intensity change gradient vector within adjacent time periods; The sign consistency and amplitude continuity of the intensity change gradient vector are jointly analyzed to identify the locations of direction reversal or amplitude abrupt change during the gradient change process. Based on the degree of gradient consistency violation, a discontinuity index characterizing the discontinuous characteristics of signal intensity evolution is generated.

5. The intelligent signal optimization method for 5GCPE as described in claim 3, characterized in that, The calculation of short-term clustering of phase jitter includes: Within the time window, the phase change sequence of a single acquisition signal is extracted, and the phase change sequence is divided into multiple continuous short-time analysis units according to a preset short-time analysis scale. Within each short-time analysis unit, the distribution density of phase change is calculated to obtain a local clustering parameter that characterizes the concentrated occurrence of phase perturbations within a local time range. By comparing and analyzing the local clustering parameters of adjacent short-time analysis units, concentrated enhancement segments of phase perturbation within the local time range are identified, and short-time clustering is constructed.

6. The intelligent signal optimization method for 5GCPE as described in claim 3, characterized in that, The calculation of the local offset trend of subcarrier energy distribution includes: Within the time window, subcarrier-level energy analysis is performed on a single acquisition signal to obtain the subcarrier energy distribution status at each time point; Using the subcarrier energy distribution state, the energy centroid position of the subcarrier energy distribution in the frequency domain at each time point is determined; Perform a continuity analysis of the trajectory of the energy center of gravity position over time, identify the directional shift characteristics of the energy center of gravity position in the local frequency band, and establish a local shift trend based on the continuous shift direction and shift amplitude.

7. The intelligent signal optimization method for 5GCPE as described in claim 1, characterized in that, Based on the correction results and the acquired signal set, signal performance calculation indicators are configured, and a signal optimization evaluation channel is constructed using these indicators, including: After obtaining the corrected time-series feature prediction results based on the correction results, the prediction performance features characterizing the future evolution trend of the signal are extracted. By combining the signal observation results of the first and second acquisition signals in the acquisition signal set at different time resolutions, the prediction performance features are weighted for consistency to generate a set of signal performance features with confidence labels. The set of signal performance characteristics is used as a performance calculation index. Based on the response sensitivity of the optimized control parameters, the performance calculation index is mapped and combined to construct a signal optimization evaluation channel.

8. The intelligent signal optimization method for 5GCPE as described in claim 1, characterized in that, The signal optimization evaluation channel is used to perform adaptive signal optimization, including: Based on the multi-dimensional evaluation results output by the signal optimization evaluation channel, the parameter response correlation relationship corresponding to antenna beam, transmit power and frequency band switching is constructed to form a multi-parameter linkage optimization space. Within the multi-parameter linkage optimization space, based on the weight distribution of each performance dimension in the signal optimization evaluation channel, an adaptive search and filtering of parameter combinations is performed to construct a candidate optimization strategy set; The performance trend consistency of the candidate optimization strategy set is verified, and the parameters of antenna beam, transmit power, and frequency band switching are simultaneously optimized based on the verification results to complete adaptive optimization. The intelligent signal optimization system of 9.5GCPE is characterized by, The system is used to implement the intelligent signal optimization method for 5GCPE according to any one of claims 1 to 8, and the system comprises: The signal acquisition module is used to activate the dual-layer acquisition and processing layer, perform enhanced data acquisition of the 5GCPE terminal, and establish an acquisition signal set, which includes primary acquisition signals and secondary acquisition signals. The timing prediction module is used to perform timing prediction of signal states based on the acquired signal set and the historical performance data after calling historical performance data, and to establish timing prediction results at multiple time steps. The abnormal feature analysis module is used to analyze the abnormal features of the secondary acquisition signals in the acquisition signal set, and to confirm the signal abnormality of the primary acquisition signal based on the abnormal feature analysis results. The evaluation channel establishment module is used to correct the timing feature prediction results at the corresponding time step based on the confirmation results, configure signal performance calculation indicators based on the correction results and the acquired signal set, and construct a signal optimization evaluation channel using the signal performance calculation indicators. The signal optimization management module is used to perform adaptive signal optimization using the signal optimization evaluation channel. The adjustment parameters of the adaptive optimization include antenna beam, transmit power, and frequency band switching. Based on the adaptive optimization results, intelligent signal optimization management is performed.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the intelligent signal optimization method of 5GCPE as described in any one of claims 1-8.

Images