A communication interference troubleshooting method, device, equipment, storage medium and product

Abstract

The application discloses a communication interference troubleshooting method, device, equipment, storage medium and product. The method comprises the following steps: acquiring interference data corresponding to target interference; the interference data at least comprises an interference value measured by a base station on a frequency domain resource; determining corresponding target interference judgment information according to the scene characteristics and interference characteristics of the interference data; the target interference judgment information at least comprises a first matching coefficient based on characteristic distance measurement, a second matching coefficient based on waveform feature recognition and an expert experience coefficient based on prior experience; determining a target interference type to which the target interference belongs according to the target interference judgment information; and outputting a troubleshooting suggestion corresponding to the target interference type. The scheme realizes the fusion and utilization of multi-dimensional information, effectively overcomes the problems of low accuracy and poor adaptability of single-dimensional recognition in a complex network environment, significantly improves the accuracy and reliability of interference root cause positioning, and provides scientific and efficient decision support for network operation and maintenance.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and in particular to a method, apparatus, device, storage medium, and product for troubleshooting communication interference. Background Technology

[0002] Traditional methods for troubleshooting communication interference typically rely on single-dimensional features for interference identification. For example, they might classify interference waveforms using mathematical modeling or perform feature matching solely through deep learning models. These methods have significant limitations in practical applications: firstly, single-dimensional identification struggles to handle issues such as the mixing and truncation of interference waveforms in complex network environments, resulting in low accuracy; secondly, existing solutions lack the integration and utilization of prior knowledge such as regional features and frequency band characteristics, and cannot be dynamically optimized based on actual troubleshooting results. This makes it difficult to adapt to the differences in interference characteristics across different scenarios, ultimately impacting the efficiency and accuracy of interference troubleshooting. Summary of the Invention

[0003] This invention provides a communication interference investigation method, apparatus, device, storage medium, and product to solve the problems of low accuracy and difficulty in adapting to complex network scenarios caused by existing communication interference investigation methods that rely solely on single-dimensional identification.

[0004] According to one aspect of the present invention, a method for troubleshooting communication interference is provided, the method comprising: Obtain the interference data corresponding to the target interference; the interference data shall include at least the interference values ​​measured by the base station in the frequency domain resources; Based on the scene characteristics and interference characteristics of the interference data, the corresponding target interference determination information is determined; the target interference determination information includes at least: a first matching coefficient based on feature distance metric, a second matching coefficient based on waveform feature recognition, and an expert experience coefficient based on prior experience; Based on the target interference determination information, determine the type of target interference to which the target interference belongs; Output troubleshooting suggestions corresponding to the type of interference from the target.

[0005] According to another aspect of the present invention, a communication interference detection device is provided, the device comprising: The interference data acquisition module is used to acquire the interference data corresponding to the target interference; the interference data includes at least the interference values ​​measured by the base station in the frequency domain resources; The determination information module is used to determine the corresponding target interference determination information based on the scene characteristics and interference characteristics of the interference data; the target interference determination information includes at least: a first matching coefficient based on feature distance metric, a second matching coefficient based on waveform feature recognition, and an expert experience coefficient based on prior experience; The interference type determination module is used to determine the type of target interference based on the target interference determination information. The troubleshooting suggestion output module is used to output troubleshooting suggestions corresponding to the type of target interference.

[0006] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the communication interference investigation method according to any embodiment of the present invention.

[0007] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions, the computer instructions being configured to cause a processor to execute and implement the communication interference detection method according to any embodiment of the present invention.

[0008] According to another aspect of the present invention, a computer program product is provided, the computer program product comprising a computer program that, when executed by a processor, implements the communication interference investigation method described in any embodiment of the present invention.

[0009] The technical solution of this invention involves acquiring interference data corresponding to a target interference. The interference data includes at least the interference values ​​measured by the base station in the frequency domain. Based on the scene characteristics and interference characteristics of the interference data, corresponding target interference determination information is determined. The target interference determination information includes at least: a first matching coefficient based on feature distance measurement, a second matching coefficient based on waveform feature recognition, and an expert experience coefficient based on prior experience. Based on the target interference determination information, the target interference type is determined. A troubleshooting suggestion corresponding to the target interference type is output. This technical solution determines the first matching coefficient, the second matching coefficient, and the expert experience coefficient from three dimensions—feature distance measurement, waveform feature recognition, and prior experience—based on the scene characteristics and interference characteristics of the target interference. It then comprehensively judges the interference type based on these three types of coefficients and outputs targeted troubleshooting suggestions. This achieves the fusion and utilization of multi-dimensional information, effectively overcoming the problems of low accuracy and poor adaptability of single-dimensional identification in complex network environments. It significantly improves the accuracy and reliability of interference root cause localization, providing scientific and efficient decision support for network operation and maintenance.

[0010] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a flowchart of a communication interference troubleshooting method provided in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the frequency band training and intent recognition architecture of the interference recognition model provided in Embodiment 1 of the present invention; Figure 3 This is a flowchart illustrating a communication interference troubleshooting method provided in Embodiment 1 of the present invention; Figure 4 This is a schematic diagram of the structure of a communication interference detection device according to Embodiment 2 of the present invention; Figure 5 This is a schematic diagram of the structure of an electronic device that implements the communication interference investigation method of this invention. Detailed Implementation

[0013] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0014] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0015] Example 1 With the continuous development of wireless communication networks, multiple network standards such as 2G, 3G, 4G, NB-IoT, and 5G coexist. The increasing density of sites and users has exacerbated internal interference and external problems. Meanwhile, interference from external sources such as illegally installed amplifiers, exam jammers, and video surveillance equipment occurs frequently, negatively impacting user experience. Currently, interference is diverse, difficult to identify, and existing analysis and solutions are fragmented. There is an urgent need to develop new methodologies to address mobile network interference and ensure a smooth network experience for users.

[0016] Figure 1 This is a flowchart of a communication interference troubleshooting method provided in Embodiment 1 of the present invention. This embodiment is applicable to situations in wireless communication systems where interference types are automatically identified and troubleshooting suggestions are output through multi-dimensional feature matching. This method can be executed by a communication interference troubleshooting device, which can be implemented in hardware and / or software and can be configured in an electronic device. Figure 1 As shown in the figure, the communication interference troubleshooting method provided in this embodiment includes the following steps: S110. Obtain the interference data corresponding to the target interference; the interference data shall include at least the interference values ​​measured by the base station in the frequency domain resources.

[0017] Target interference refers to abnormal uplink interference signals appearing in the wireless communication system under investigation, and is the object of this interference investigation. Interference data refers to the signal characteristics carrying the target interference, which can be used for subsequent interference analysis and type identification. It includes at least the interference values ​​measured by the base station in the frequency domain resources, such as the interference power measurements of each resource block (RB) in the frequency domain. Interference data usually exists in the form of numerical sequences to characterize the frequency domain distribution characteristics of the interference. RBs are 12 consecutive subcarriers in a frequency domain. The number of RBs varies in different frequency bands and bandwidths; currently, 4G / 5G networks have a maximum of 273 RBs.

[0018] In this embodiment of the invention, interference data corresponding to the target interference to be investigated can be obtained. This data includes at least the interference power value reported by the base station and measured on the RB, which serves as the raw input for subsequent analysis.

[0019] S120. Based on the scene characteristics and interference characteristics of the interference data, determine the corresponding target interference judgment information; the target interference judgment information includes at least: a first matching coefficient based on feature distance measurement, a second matching coefficient based on waveform feature recognition, and an expert experience coefficient based on prior experience.

[0020] In this embodiment, scene features can refer to background information related to the environment in which the target interference is located, which may include at least one of the following: target communication standard, target communication frequency band, and target geographical information, used to limit the context of interference identification and the basis for screening and matching. The target communication standard can refer to the wireless communication technology standard or protocol type involved in the target interference, which may include, but is not limited to, 2G, 3G, 4G, 5G, NB-IoT, etc., used to limit the network standard range to which the interference belongs, so as to screen the corresponding interference knowledge base and waveform recognition model in subsequent matching. The target communication frequency band can refer to the frequency range in which the target interference is located, such as, but not limited to, specific frequency bands such as 700MHz, 2.6GHz, 3.5GHz, etc., used to limit the frequency attributes of the interference, so as to screen the corresponding frequency band's interference knowledge base and waveform recognition model in subsequent matching, avoiding confusion caused by differences in interference characteristics of different frequency bands. The target geographical information can refer to the geographical location information of the target interference, including but not limited to cities, regions (such as coastal / inland), base station locations, etc., used to limit the spatial attributes of the interference, so as to statistically analyze the historical interference distribution patterns of the region in subsequent matching and obtain expert experience coefficients with regional characteristics.

[0021] Interference features can refer to characteristic information extracted from the original interference data of the target interference that can characterize the characteristics of the interference waveform, including at least a standardized target interference feature pattern. Target interference determination information can refer to multi-dimensional quantitative information used to comprehensively determine the type of target interference, including at least a first matching coefficient based on feature distance measurement, a second matching coefficient based on waveform feature recognition, and an expert experience coefficient based on prior experience. The first matching coefficient can be a quantified coefficient used to characterize the similarity of features between the interference features to be investigated and various standard interference features in a preset interference library, obtained by calculating the feature distance between the interference features to be investigated and the feature distance, and then normalizing the result. The second matching coefficient can be a quantified coefficient used to characterize the degree of matching between the interference to be investigated and various preset interference types, output after extracting and recognizing the deep features such as the frequency domain waveform morphology and variation patterns of the interference to be investigated through waveform feature recognition algorithms or models. The expert experience coefficient can be a quantified coefficient used to characterize the probability of occurrence of various interference types in a given scenario, obtained based on historical interference investigation data of the target interference scenario and statistical analysis of prior knowledge from experts in the communications field.

[0022] In this embodiment of the invention, corresponding scene features and interference features can be extracted from the original interference data corresponding to the target interference to be investigated. The scene features include at least one of the following: target communication standard, target communication frequency band, target geographical information, etc. These features provide a screening range and prior basis for subsequent matching. The interference features are used to standardize and characterize the waveform features of the interference to be investigated. Then, based on the extracted scene features and interference features, target interference determination information for determining the type of interference to be investigated is analyzed from different dimensions. The target interference determination information may include at least: a first matching coefficient based on feature distance measurement, a second matching coefficient based on waveform feature recognition, and an expert experience coefficient based on prior experience. S130. Based on the target interference determination information, determine the target interference type to which the target interference belongs.

[0023] Among them, the target interference type can refer to the specific category label of the interference determined by multi-dimensional analysis of the interference data of the target interference. The interference type can be used to distinguish interference with different causes or characteristics, such as intra-system interference or extra-system interference, and can be further subdivided into specific categories such as repeater interference, atmospheric waveguide interference, and privately installed amplifier interference.

[0024] In this embodiment of the invention, based on the obtained target interference determination information, the first matching coefficient, the second matching coefficient, and the expert experience coefficient corresponding to each candidate interference type are extracted; then, by fusing all the coefficients, the comprehensive discrimination rate corresponding to each candidate interference type is obtained; finally, the candidate interference type with the largest comprehensive discrimination rate is determined as the target interference type to which the target interference to be investigated belongs.

[0025] S140, Output troubleshooting suggestions corresponding to the target interference type.

[0026] Among them, the investigation suggestions can refer to the practical investigation operation guidelines and solutions provided for eliminating the identified type of interference.

[0027] In this embodiment of the invention, based on the finally determined target interference type, the solution suggestions can be extracted from a preset interference knowledge base or the latest solutions can be obtained through large-scale model network search, etc., and finally, the investigation suggestions that can be directly implemented and correspond to the target interference type can be output, thereby completing the closed loop of the entire interference investigation process.

[0028] The technical solution of this invention determines the corresponding first matching coefficient, second matching coefficient, and expert experience coefficient from three dimensions—feature distance measurement, waveform feature recognition, and prior experience—based on the scene features and interference features of the target interference. It then comprehensively judges the interference type based on these three types of coefficients and outputs targeted troubleshooting suggestions. This achieves the fusion and utilization of multi-dimensional information, effectively overcoming the problems of low accuracy and poor adaptability of single-dimensional identification in complex network environments. It significantly improves the accuracy and reliability of interference root cause localization and provides scientific and efficient decision support for network operation and maintenance.

[0029] Furthermore, based on the above embodiments of the invention, the interference features include a target interference feature pattern, which is obtained through the following methods: The interference values ​​of each resource block are extracted from the interference data to obtain the resource block interference value sequence; The mean normalization process is performed on the resource block interference value sequence to obtain the target interference feature map.

[0030] The resource block interference value sequence can refer to a set of interference values ​​arranged in the order of resource block indices. The target interference feature pattern can refer to a resource block value sequence obtained after mean normalization, which is a numerical and structured expression of the target interference waveform features and will serve as direct input data for subsequent multi-dimensional matching.

[0031] In this embodiment of the invention, the interference features of the target interference can be characterized by a target interference feature pattern, and the extraction process of the target interference feature pattern includes: (1) Extract the interference values ​​of each resource block (RB) measured by the base station in the frequency domain from the acquired target interference data. Assuming the total number of RBs corresponding to the system bandwidth is N, the original sequence of resource block interference values ​​is obtained as follows: This sequence reflects the power distribution of the target interference in the frequency domain and is the basis for subsequent feature extraction.

[0032] (2) The extracted resource block interference value sequence is normalized to eliminate the influence of the difference in absolute power level between different interferences on waveform feature recognition and to highlight the shape features of the interference waveform. The normalization process is calculated as follows: In the formula, This represents the original interference value of the i-th RB; The normalized value of the i-th RB represents the relative proportion of the interference intensity of that RB in the total interference of all RBs.

[0033] (3) Combine all the normalized resource block values ​​according to the resource block index order to obtain the target interference feature map. This feature pattern is a numerical and structured representation of the target interference waveform features, and will serve as the direct input data for the subsequent calculation of the first matching coefficient (based on feature distance metric) and the second matching coefficient (based on waveform feature recognition).

[0034] Furthermore, based on the above embodiments of the invention, the first matching coefficient is obtained in the following manner: According to the target communication standard and target communication frequency band, retrieve the pre-classified and constructed preset interference knowledge base, and determine the corresponding candidate interference types and the candidate interference feature patterns contained in each candidate interference type in the preset interference knowledge base. Determine the feature distance between the target interference feature map and each candidate interference feature map under each candidate interference type; The minimum value among the feature distances determined under the same candidate interference type is determined as the baseline distance for the corresponding candidate interference type; Determine the mean of the baseline distance for all candidate interference types, and use the mean as the temperature coefficient of the preset variable-temperature Softmax function; The temperature coefficient and the baseline distance of each candidate interference type are input into the preset variable temperature Softmax function for normalization processing to obtain the first matching coefficient corresponding to each candidate interference type.

[0035] The pre-built interference knowledge base can refer to a pre-constructed database that stores interference types and their corresponding waveform feature patterns categorized by standard and frequency band. Candidate interference types can refer to the types of interference that may exist in the knowledge base under the current standard and frequency band, such as repeater interference, atmospheric duct interference, and unauthorized amplifier interference. Candidate interference feature patterns can refer to the normalized resource block sequence corresponding to each candidate interference type, used to characterize the typical waveform shape of that type of interference. Feature distance can refer to a quantitative indicator used to measure the similarity between the target interference feature pattern and the candidate interference feature pattern; the smaller the distance, the higher the feature similarity between the two. For example, this feature distance can include at least Euclidean distance. The baseline distance can refer to the minimum feature distance between the target interference feature pattern and all candidate interference feature patterns under the same candidate interference type, representing the best matching degree between that type of interference and the target interference. The variable-temperature Softmax function can refer to an improved Softmax normalization function whose temperature parameter can be dynamically adjusted according to the statistical characteristics of the input data to enhance the distinguishability between categories. The temperature coefficient can refer to the temperature parameter in the preset variable-temperature Softmax function, which is calculated from the mean of the baseline distances of all candidate interference types and is used to adjust the smoothness of the distribution of the first matching coefficient.

[0036] In this embodiment of the invention, the specific method for obtaining the first matching coefficient includes: (1) Based on the target communication standard and target communication frequency band to which the target interference belongs, retrieve the corresponding candidate interference type from the pre-classified interference knowledge base, as well as all candidate interference feature patterns contained in each candidate interference type. The pattern is a resource block sequence obtained by normalizing historical interference data.

[0037] (2) Calculate the feature distance between the target interference feature map and all candidate interference feature maps under each candidate interference type. For example, the feature distance can be a multidimensional Euclidean distance.

[0038] (3) For each candidate interference type, select the smallest feature distance from all the calculated feature distances under that type and determine it as the reference distance corresponding to that candidate interference type. The reference distance represents the closest relationship between the target interference and that type of interference on the mathematical waveform.

[0039] (4) Calculate the reference distances corresponding to all candidate interference types, calculate the arithmetic mean of these reference distances, and use this average as the temperature coefficient T of the preset variable temperature Softmax function, as follows: In the formula, M represents the number of candidate interference types; is the baseline distance for the i-th candidate interference type.

[0040] (5) Input the temperature coefficient and the reference distance corresponding to each candidate interference type into the preset variable temperature Softmax function for normalization. This function converts the reference distances of different ranges into coefficients with a uniform value range, and finally obtains the first matching coefficient corresponding to each candidate interference type. as follows: In the formula, the first matching coefficient The value of is between [0,1], and all candidate interference types The sum of these coefficients is 1. This coefficient reflects the probability that the target interference belongs to the i-th type of interference under the mathematical distance metric. The smaller the distance (i.e. the more similar the waveforms), the higher the matching coefficient obtained for the category.

[0041] This embodiment uses variable-temperature Softmax adaptive normalization to convert feature distances (such as Euclidean distances) into probabilistic matching coefficients. This preserves the objectivity of mathematical distances while enhancing the distinguishability between different interference types through temperature coefficients. It can also adapt to the differences in interference characteristics across different frequency bands and standards, providing an accurate and reliable objective basis for subsequent multi-dimensional fusion and discrimination.

[0042] In one embodiment, this solution also provides a method for constructing an interference database and interference knowledge base for wireless communication systems, aiming to provide a standardized data foundation for subsequent intelligent interference identification. The method specifically includes: (1) Raw interference data under different frequency bands and standards is collected from base stations operating in the current network. The interference data reported by the base stations is measured in units of resource blocks (RBs), with each RB corresponding to 12 consecutive subcarriers in the frequency domain. The total number of RBs N varies under different system bandwidths; for example, in a 5G system, the maximum number of RBs can reach 273. For each measurement moment, the base station records the interference power value on each RB, forming a raw interference data sequence. These unprocessed raw data are categorized and stored according to information such as time, frequency band, and standard to form an interference database.

[0043] (2) In order to extract waveform features that can be used for matching from the raw data, it is necessary to clean and normalize the raw data in the interference database to obtain the normalized RB sequence. As a feature pattern of interference.

[0044] (3) After obtaining the cleaned interference feature patterns, each interference sample is annotated in multiple dimensions based on expert knowledge and historical investigation records to form a structured interference knowledge base. Each knowledge base entry contains at least the following six-dimensional vector information: ① Interference Category: Identify the type of interference source, such as internal interference (inter-device interference, adjacent channel interference, etc.) or external interference (privately installed amplifiers, atmospheric waveguides, external signal sources, etc.). ② Interference type: Specific interference name, such as repeater interference, atmospheric waveguide interference, video surveillance equipment interference, etc.; ③ Interference waveform: The normalized RB sequence, i.e., the interference feature pattern; ④ Interference characteristics: Qualitative description of the interference waveform, such as sawtooth, plateau, pulse, broadband flat, etc., to facilitate human understanding and classification; ⑤ Interference frequency band: Identifies the frequency band to which the interference sample belongs, such as 700MHz, 2.6GHz, etc.; ⑥ Expert advice: Suggestions for troubleshooting and handling this type of interference, such as checking for unauthorized amplifiers nearby, adjusting the tilt angle of the base station antenna, and checking the working status of the weather radar.

[0045] The aforementioned six-dimensional information is categorized and stored according to frequency band and standard, thus forming a pre-defined interference knowledge base. This knowledge base not only provides standard feature pattern references for subsequent Euclidean distance matching, but also provides labeled data for training the waveform recognition model, and provides expert knowledge support for the final investigation and suggestion output.

[0046] Furthermore, based on the above embodiments of the invention, the second matching coefficient is obtained in the following manner: According to the target communication standard and target communication frequency band, retrieve the pre-trained preset waveform recognition model; The target interference feature pattern is input into the preset waveform recognition model, and the matching degree between the target interference and each candidate interference type is obtained as the second matching coefficient.

[0047] The preset waveform recognition model can refer to a deep learning model trained in advance using a large number of labeled interference samples. It is used to infer the input target interference feature pattern and output the matching degree between the target interference and each candidate interference type. The model is constructed according to frequency band and standard classification to improve recognition accuracy.

[0048] In this embodiment of the invention, the specific method for obtaining the second matching coefficient includes: (1) According to the target communication standard and target communication frequency band to which the target interference belongs, retrieve the preset waveform recognition model that matches the standard and frequency band from multiple pre-stored waveform recognition models. The model is trained in advance using a large number of labeled interference samples under the corresponding frequency band and standard, and can accurately capture the subtle features of various interference waveforms under the specific frequency band / standard.

[0049] (2) The preprocessed target interference feature pattern is input into the preset waveform recognition model. The model automatically extracts key features (such as sawtooth, plateau, and pulse patterns) from the waveform through a multi-layer neural network structure, and compares them with the waveform patterns of various interferences learned by the model during the training phase. Finally, after model inference, an M-dimensional probability distribution vector is output. , where M is the number of candidate interference types; This represents the confidence or matching degree of the model's determination that the target interference belongs to the i-th candidate interference type. This vector is the second matching coefficient, which quantifies the similarity between the target interference and each candidate interference type at the waveform feature level. The higher the value, the more similar the waveform of the target interference is to the typical waveform of the i-th type of interference.

[0050] This embodiment uses a deep learning model trained by frequency band and system type to identify interference feature patterns. It can accurately capture complex waveform features that are difficult to distinguish by mathematical distance (such as sawtooth, truncated, etc.), significantly improving the ability to identify mixed and truncated interference, and providing accurate and reliable waveform feature matching basis for subsequent multi-dimensional fusion discrimination.

[0051] In one embodiment, because Euclidean distance may lead to misjudgment in some interference situations, such as sawtooth or plateau waveforms, while these features can be identified relatively well manually, Euclidean distance judgment may suffer from significant deviations in the final decision due to factors such as peak shift and phase shift. Therefore, this embodiment employs a large language model for waveform feature matching and recognition. Figure 2 As shown, the preset waveform recognition model can use the DeepSeek V3-0324 large language model, which is trained in the following way: For complex waveforms such as sawtooth and plateau shapes that are difficult to accurately identify due to factors such as peak offset and phase offset, Euclidean distance is used to label various features that may appear in the interference, and input them into the model in both image and numerical ways for parameter tuning; at the same time, in order to avoid the knowledge base being too complex and thus causing a high illusion rate and to improve the matching accuracy, a frequency band and system-specific training method is adopted. Different large models are trained according to the feature differences of different frequency bands and systems. For example, interference types that do not appear in certain frequency bands do not need to be included in the corresponding knowledge base; by optimizing parameters such as TOP P and Temperature, iterative training is carried out with the goal of maximizing the probability of actual interference waveforms, so that the model can output the matching degree of the waveform corresponding to the interference type as the second matching coefficient.

[0052] Furthermore, based on the above embodiments of the invention, the expert experience coefficient is obtained in the following way: Based on the target communication standard, target communication frequency band, and target geographical information, obtain historical interference investigation data for the corresponding scenario; The frequency of occurrence of each candidate interference type in historical interference investigation data is statistically analyzed, and the frequency is used as the expert experience coefficient for the corresponding candidate interference type.

[0053] Historical interference investigation data can refer to records of past interference events that have been manually confirmed and whose interference type has been clearly identified. For example, it may include information such as the system, frequency band, region, and finally confirmed type of interference when the interference occurred.

[0054] In this embodiment of the invention, the specific method for obtaining the expert experience coefficient includes: (1) Based on the target communication standard, target communication frequency band, and target geographical area information of the target interference, retrieve the confirmed historical interference investigation records for the corresponding scenario from databases such as the operation and maintenance database and the historical investigation record database to form historical interference investigation data. Among them, the data includes all the interference events that have been located in the scenario within a certain period of time (such as one year) and their finally confirmed interference types.

[0055] (2) Classify and statistically analyze the acquired historical interference investigation data, and calculate the number of occurrences of each candidate interference type (such as repeater interference, atmospheric waveguide interference, privately installed amplifier interference, etc.) in the scenario. And divide by the total number of interference events in that scenario. The occurrence frequency of the i-th candidate interference type is obtained. Next, the calculated frequency of occurrence will be... The expert experience coefficient directly serves as the corresponding candidate interference type. This coefficient reflects the prior probability of various types of interference occurring in a specific scenario (system + frequency band + region), and simulates the predictive tendency formed by interference experts based on long-term experience.

[0056] This embodiment transforms the implicit experience of domain experts into explicit, quantifiable prior coefficients through statistical analysis of real historical interference data. This enables the interference identification system to form reasonable predictive tendencies based on scene characteristics such as region and frequency band, much like an experienced expert, before formally analyzing waveforms. This data-driven prior knowledge construction method not only ensures the objectivity and scene adaptability of the coefficients but also provides an important weighting foundation for subsequent fusion and discrimination, complementing the objective matching results and significantly improving the accuracy and robustness of interference identification.

[0057] Furthermore, based on the above embodiments of the invention, the target interference type is determined according to the target interference determination information, including: For each candidate interference type, the product of the corresponding first matching coefficient, second matching coefficient, and expert experience coefficient is used as the comprehensive discrimination rate of the corresponding candidate interference type. The candidate interference type with the highest overall discrimination rate is determined as the target interference type.

[0058] In this embodiment of the invention, for each candidate interference type i, the corresponding first matching coefficient can be calculated. Second matching coefficient Coefficient of Expert Experience The product between them is used as the overall discrimination rate for the candidate type. This coefficient fusion process integrates information from three different dimensions: mathematical shape similarity, waveform semantic matching degree, and prior experience probability, achieving a multi-faceted comprehensive judgment of interference types. Next, the comprehensive discrimination rates of all candidate interference types are compared. The candidate interference type with the largest value is determined as the final target interference type.

[0059] Furthermore, based on the above embodiments of the invention, troubleshooting suggestions corresponding to the target interference type are output, including: Extract the first investigation suggestion corresponding to the target interference type from the preset interference knowledge base, and / or obtain the second investigation suggestion corresponding to the target interference type through online search; Output the first investigation suggestion and / or the second investigation suggestion.

[0060] The first investigation suggestion can refer to a standardized investigation plan corresponding to the target interference type extracted from a pre-set interference knowledge base. This may include, but is not limited to, interference cause analysis, investigation steps, and handling measures. Online search can refer to the process of retrieving the latest solutions, case studies, or technical information related to the target interference type in real time through external search engines, professional technical databases, or online knowledge platforms. The second investigation suggestion can refer to real-time solutions or case studies related to the target interference type obtained through online search, used to supplement or update the standardized suggestions in the pre-set knowledge base.

[0061] In this embodiment of the invention, after determining the type of target interference, corresponding troubleshooting suggestions can be output to complete the closed loop of the entire interference troubleshooting process. The specific process is as follows: (1) Based on the determined target interference type, extract the corresponding first investigation suggestion from the preset interference knowledge base. The first investigation suggestion can cover standardized content that has been verified in actual scenarios, such as the basic investigation process, standard handling steps, historical successful investigation cases, and compliance operation specifications for that interference type.

[0062] (2) Obtain second investigation suggestions corresponding to the target interference type through online search. These second investigation suggestions can cover supplementary content not included in the static knowledge base, such as cutting-edge interference handling solutions in the industry, the latest successful cases of similar interference in the same scenario, real-time updated operation and maintenance specifications of operators, and targeted solutions for new types of interference.

[0063] (3) Output the obtained first and / or second investigation suggestions, specifically including: ① If only the first investigation suggestion is obtained, then the first investigation suggestion is directly determined as the final output content; ② If only the second investigation suggestion is obtained, then the second investigation suggestion is directly determined as the final output content; ③ If both the first and second investigation suggestions are obtained, then the two types of suggestions can be integrated and optimized to form a comprehensive investigation suggestion that takes into account both standardization and cutting-edge technology, which is the final output content. Then, the final determined investigation suggestions are output to the entity performing the interference investigation (such as maintenance personnel, automated network maintenance systems, base station centralized management platforms, etc.) through methods not limited to, but including, visualization of the operation and maintenance interface, export of standardized reports, push of operation and maintenance system instructions, SMS / platform message notifications, etc., thereby completing the entire investigation suggestion output process.

[0064] This embodiment combines standardized suggestions from a pre-set knowledge base with real-time information from online searches, providing maintenance personnel with a reliable basic troubleshooting solution and enabling them to access the latest technological advancements and case experiences. This effectively shortens interference troubleshooting time, improves problem-solving efficiency, and upgrades the interference location system from a simple "identification tool" to an intelligent auxiliary system with "problem-solving capabilities."

[0065] Furthermore, based on the above embodiments of the invention, before determining the overall discrimination rate, a step of correcting the first matching coefficient and the second matching coefficient is included, specifically including: Obtain the correction coefficients corresponding to each candidate interference type; For each candidate interference type, the corresponding corrected matching coefficient is determined based on the corresponding first matching coefficient, second matching coefficient, and correction coefficient; Based on the corrected matching coefficients and expert experience coefficients, the comprehensive discrimination rate corresponding to each candidate interference type is re-determined.

[0066] The correction coefficient refers to the quantified compensation value used to correct the inherent matching degree deviation of different candidate interference types. Specifically, it compensates for the inherent deviation caused by waveform characteristic differences in the objective matching process (first matching coefficient and second matching coefficient), and each candidate interference type corresponds to an independent correction coefficient. The corrected matching coefficient refers to the quantified matching degree without baseline deviation obtained by combining the first matching coefficient, the second matching coefficient, and the corresponding correction coefficient to correct the inherent deviation. It is the core intermediate parameter for calculating the overall discrimination rate.

[0067] In actual interference assessment, some interference types naturally have a lower objective matching degree than others, leading to a bias in the final comprehensive judgment. Specifically, the first matching coefficient... Second matching coefficient Essentially, the matching degree is calculated based on objective algorithms (such as Euclidean distance and waveform feature recognition), and theoretically should be fair. However, due to the differences in the mathematical and graphical significance of waveform features of different interference types, unfairness exists in the matching process.

[0068] Suppose that three types of interference exist in a certain frequency band: A (terrain interference), B (peak interference), and C (sawtooth interference). If an equal number of samples of each type of interference are input into the matching system, from a fairness perspective, what is the objective matching degree of the three types of interference (i.e., The statistical mean of each type should be roughly equivalent. However, actual statistics show that the mean of some types is significantly lower, indicating that this type of interference has an inherent disadvantage in the objective matching process.

[0069] Taking tiered and sawtooth interference as examples for analysis: Tiered interference, due to its relatively flat waveform, easily achieves a high matching degree with various data in the database, such as tiered, peaked, and sawtooth patterns, during Euclidean distance matching and waveform feature matching. This dispersion in matching degree results in its features not being prominent, making it difficult to obtain a specific high matching degree. Conversely, sawtooth interference, due to its more obvious waveform variation characteristics, although achieving better matching results for similar patterns, is extremely sensitive to waveform alignment. Even for the same sawtooth waveform, a phase shift or misalignment by one bit can cause a precipitous drop in the Euclidean matching coefficient.

[0070] Therefore, interference types with indistinct waveform features (such as truncated shapes) are naturally at a disadvantage in objective matching, while types with obvious features (such as jagged shapes), although they are more likely to obtain high scores, are also more likely to lose points due to small deviations. If this difference in matching degree caused by the features themselves is not corrected, the probability of judging interference with high matching degree features will be much higher than that of interference with low matching degree features, thus affecting the overall recognition accuracy.

[0071] Therefore, this embodiment introduces a correction coefficient to compensate for the first matching coefficient and the second matching coefficient, so that various types of interference can compete fairly on the same baseline in the objective matching process, thereby improving the accuracy and robustness of overall interference identification.

[0072] Specifically, the steps for correcting the first and second matching coefficients include: (1) Before formally calculating the overall discrimination rate of each candidate interference type, obtain the correction coefficient pre-configured for each candidate interference type; wherein, the correction coefficient can be pre-calculated based on the statistical characteristics of a large number of confirmed interference samples, and is used to measure the systematic deviation of this type of interference in the objective matching process.

[0073] (2) For each candidate interference type i, set its corresponding first matching coefficient Second matching coefficient By combining the results, a preliminary objective matching degree can be obtained. Then, the correction coefficient corresponding to the candidate interference type is used. The initial matching degree is compensated to obtain the corrected matching coefficient. This compensation operation provides a fair boost to interference types that might be underestimated due to their insignificant waveform characteristics, while preventing types with overly prominent characteristics from gaining an undue advantage.

[0074] (3) The corrected matching coefficients are compared with the expert experience coefficients. Multiply by each product to obtain a new overall discrimination rate. The new comprehensive discrimination rate is then used to determine the type of interference.

[0075] The correction step in this embodiment effectively eliminates the identification bias caused by different waveform characteristics by compensating for the inherent differences in objective matching of different interference types. This allows interference types with inconspicuous characteristics to be judged fairly, significantly improving the accuracy and robustness of interference identification and laying a reliable foundation for subsequent output of accurate troubleshooting suggestions.

[0076] Furthermore, based on the above embodiments of the invention, the correction coefficient is obtained in the following manner: Obtain the first training set, which contains multiple first interference samples under each candidate interference type. Each first interference sample is associated with a corresponding interference type label. Determine the first matching coefficient and the second matching coefficient corresponding to each first interference sample, and take the product of the first matching coefficient and the second matching coefficient as the joint matching coefficient of the corresponding first interference sample; The average joint matching coefficient of all first interference samples under the same candidate interference type is determined as the average joint matching coefficient of the corresponding candidate interference type. The maximum value among the average joint matching coefficients of all candidate interference types is determined as the baseline matching coefficient; For each candidate interference type, the difference between the corresponding average joint matching coefficient and the baseline matching coefficient is used as the correction coefficient for the corresponding candidate interference type.

[0077] Here, the first training set can refer to the set of labeled interference samples used to calculate the correction coefficients. Each sample (the first interference sample) contains a true interference type label and is stored according to the candidate interference type. The joint matching coefficient can refer to the value obtained by multiplying the first matching coefficient and the second matching coefficient, used to comprehensively measure the overall performance of the interference samples in the objective matching dimension. The average joint matching coefficient can refer to the arithmetic mean of the joint matching coefficients of all first interference samples under the same candidate interference type, used to represent the average level of this type of interference in objective matching. The baseline matching coefficient can refer to the maximum value among the average joint matching coefficients of all candidate interference types, which can be used as a benchmark level to measure the compensation required for other interference types.

[0078] In this embodiment of the invention, the specific method for obtaining the correction coefficient includes: (1) Obtain the first training set for calculating the correction coefficient. This training set can completely cover all candidate interference types to be identified. Each candidate interference type contains a sufficient number of first interference samples, and each sample is associated with a unique interference type label confirmed by on-site investigation, providing a reliable and real data source for subsequent statistical calculations. Taking the three types of interference (table-shaped, sawtooth, and peak-shaped) in the 700M frequency band as an example, 1000 waveform samples of confirmed interference types can be selected as the first training set for each type.

[0079] (2) For each first interference sample in the first training set, determine the first matching coefficient based on feature distance metric and the second matching coefficient based on waveform feature recognition, and then use the product of the two as the joint matching coefficient corresponding to the sample. This coefficient reflects the comprehensive performance of the sample in the objective matching dimension.

[0080] (3) For each candidate interference type, the joint matching coefficient of all first interference samples under the candidate interference type is averaged to obtain the average joint matching coefficient of the candidate interference type. This coefficient represents the overall level of objective matching of the interference type.

[0081] (4) Traverse the average joint matching coefficients corresponding to all candidate interference types, select the maximum value among them, and determine it as the benchmark matching coefficient shared by all candidate interference types, as a unified reference standard for subsequent correction coefficient calculation.

[0082] (5) For each candidate interference type, calculate the difference between the average joint matching coefficient and the benchmark matching coefficient. This difference is the correction coefficient for that candidate interference type. This coefficient can accurately quantify the inherent gap between the interference type and the highest matching benchmark, and is used to compensate and correct the matching results of real-time interference samples. Continuing with the example of the three types of interference (platform, sawtooth, and peak) in the 700MHz band, assume that the average joint matching coefficients for the three types of interference are 0.8, 0.3, and 0.5, respectively. Since Euclidean distance matching and waveform feature recognition are both objective matching methods, theoretically, for the same number of input samples, the average joint matching coefficients of each type of interference should tend to be consistent. Therefore, the maximum value of 0.8 among all categories is selected as the benchmark matching coefficient to measure the inferiority of other types, that is, the correction coefficients for the three types of interference are 0, 0.5, and 0.3, respectively.

[0083] This embodiment calculates correction coefficients by statistically analyzing the objective matching performance of confirmed interference samples. This quantifies the inherent differences in waveform characteristics of different interference types, providing fair compensation for interference types with inconspicuous characteristics. This effectively eliminates the unfairness of the objective matching process, significantly improves the accuracy and stability of interference identification, and makes the identification results more reliable.

[0084] Furthermore, based on the above embodiments of the invention, the method further includes a step of optimizing the expert experience coefficient, specifically including: Obtain the second training set, which contains multiple second interference samples classified according to the combination of target communication standard, target communication frequency band and target geographical information. Each second interference sample is associated with a corresponding interference type label, and each interference type label is associated with a correction coefficient. For each second interference sample, expert experience coefficients are obtained according to the target communication standard, target communication frequency band and target geographical information for the corresponding scenario. For each second interference sample, the corresponding predicted interference type is determined based on the corresponding first matching coefficient, second matching coefficient, correction coefficient, and expert experience coefficient. The number of target samples in the second training set whose predicted interference type matches the associated interference type label is counted, and the corresponding interference recognition accuracy is determined based on the number of target samples and the total number of samples in the second training set. With the objective function of maximizing the accuracy of interference identification, and with the expert experience coefficient as the optimization variable, an expert experience coefficient optimization model is constructed. The expert experience coefficient optimization model is iteratively adjusted using a preset gradient descent method. After each iteration, the interference recognition accuracy corresponding to the second training set is recalculated until the interference recognition accuracy converges to a preset threshold or the number of iterations reaches a preset upper limit. The expert experience coefficients obtained after iterative convergence are determined as the optimized expert experience coefficients.

[0085] The second training set refers to the dataset used to optimize the expert experience coefficients. This dataset is classified according to the combination of target communication standard, target communication frequency band, and target geographical information. It may contain multiple second interference samples with confirmed interference types. Each sample has a true interference type label and pre-calculated correction coefficients. The expert experience coefficient optimization model can be a mathematical model constructed with maximizing interference recognition accuracy as the objective function and expert experience coefficients as the optimization variable, used to achieve automated iterative optimization of the expert experience coefficients. The pre-configured gradient descent method can be a pre-configured iterative optimization algorithm used to solve the gradient of the expert experience coefficient optimization model. For example, it may include, but is not limited to, multidimensional gradient descent and stochastic gradient descent, used to achieve targeted iterative adjustment of the expert experience coefficients.

[0086] In this embodiment of the invention, as can be seen from the foregoing embodiments, the initial expert experience coefficient is based solely on frequency statistics of historical interference investigation data, representing a static prior knowledge. However, the wireless communication network environment is dynamically changing (e.g., the emergence of new interference sources, evolution of geographical characteristics, changes in frequency band usage, etc.), and static statistical values ​​may not accurately reflect the current reality. Therefore, it is necessary to optimize the expert experience coefficient so that it can be dynamically adjusted based on feedback from actual samples, thereby more accurately guiding interference type identification. Specifically, the optimization process of the expert experience coefficient includes: (1) Based on the combination scenario of target communication standard, target communication frequency band and target geographical information, collect multiple confirmed types of second interference samples to form a second training set. Each sample is labeled with a real interference type label and is associated with a pre-calculated correction coefficient according to its interference type.

[0087] (2) For each second interference sample in the second training set, the current expert experience coefficients for the corresponding combined scenario are retrieved based on the target communication standard, target communication frequency band, and target geographical information. It should be noted that when performing expert experience coefficient optimization, the retrieved current expert experience coefficients refer to the coefficient values ​​currently stored for that scenario. If the scenario has not yet been optimized, the value is the expert experience coefficient initially calculated for that scenario; if the scenario has undergone historical optimization, the value is the optimized expert experience coefficient obtained and stored after the most recent optimization convergence. In this way, each optimization is performed based on the previous optimization, allowing the expert experience coefficients to continuously approach the optimal value as data accumulates, achieving continuous adaptive improvement.

[0088] (3) For each second interference sample, first use its first matching coefficient. Second matching coefficient Combined with the corresponding correction coefficient Calculate the corrected matching coefficients Then, the corrected matching coefficient is compared with the expert experience coefficient for the current scenario. Multiply by the product to obtain the overall discrimination rate for each candidate interference type. The candidate interference type with the highest overall discrimination rate is then used as the predicted interference type for that sample.

[0089] (4) Traverse all samples in the second training set and count the number of samples whose predicted interference type matches the actual interference type label, i.e. the number of target samples; then divide this number by the total number of samples in the second training set to obtain the interference recognition accuracy under the current expert experience coefficient.

[0090] (5) Taking the maximization of interference recognition accuracy as the objective function and the expert experience coefficient as the optimization variable, the expert experience coefficient optimization model is constructed as follows: In the formula, This interferes with the accuracy of recognition.

[0091] Next, based on the second training set, a preset gradient descent method (such as multidimensional gradient descent) can be used to iteratively adjust the model by defining the maximum downward slope: in each iteration, the gradient direction is calculated based on the current interference recognition accuracy, and the expert experience coefficients are updated along the gradient ascent direction; after updating, the interference recognition accuracy of the second training set is recalculated; this process is repeated until the accuracy converges (the change is less than the preset threshold) or the number of iterations reaches the preset upper limit.

[0092] (6) The expert experience coefficients obtained after iterative convergence are used as the optimized expert experience coefficients for this scenario. Stored and used for subsequent troubleshooting of actual interference in the same scenario.

[0093] In this embodiment, by performing the above optimization process, the expert experience coefficient is no longer limited to simple historical frequency statistics, but can be dynamically adjusted based on actual sample feedback, making prior knowledge more accurately reflect the interference distribution patterns in the current network environment. The optimized expert experience coefficient and the correction coefficient work synergistically to significantly improve the accuracy of interference identification, while enhancing the method's adaptability to different regions, frequency bands, and standards, achieving more intelligent and efficient interference root cause localization.

[0094] Furthermore, based on the above embodiments of the invention, determining the target interference type to which the target interference belongs includes: Obtain the correction coefficients corresponding to each candidate interference type, as well as the optimized expert experience coefficients corresponding to the scene where the target interference is located; For each candidate interference type, the corresponding corrected matching coefficient is determined based on the corresponding first matching coefficient, second matching coefficient, and correction coefficient; For each candidate interference type, the product of the corresponding corrected matching coefficient and the expert experience coefficient is used as the comprehensive discrimination rate of the corresponding candidate interference type; The candidate interference type with the highest overall discrimination rate is determined as the target interference type.

[0095] In this embodiment of the invention, the type of target interference can be determined by combining the correction coefficient introduced in the above embodiments with the optimized expert experience coefficient. The specific process is as follows: (1) Obtain the pre-calculated correction coefficients for each candidate interference type. And the optimized expert experience coefficient corresponding to the scene where the target interference is located. .

[0096] (2) For each type of candidate interference, based on the first matching coefficient of the target interference... Second matching coefficient Combined with the correction coefficient of this type The corrected matching coefficient was calculated to be Next, the corrected matching coefficients are compared with the optimized expert experience coefficients. Multiply by the product to obtain the overall discrimination rate for the candidate interference type. .

[0097] (3) Compare the overall discrimination rate of all candidate interference types and determine the candidate interference type with the highest value as the target interference type to which the target interference belongs.

[0098] This embodiment compensates for the unfairness of different waveform features in objective matching by using a correction coefficient, making the objective matching results more balanced. At the same time, it adopts an optimized expert experience coefficient to make the prior probability more consistent with the actual scenario distribution, thereby obtaining more accurate discrimination results when fusing subjective and objective information, effectively improving the accuracy of interference root cause localization and providing a reliable basis for interference investigation.

[0099] Figure 3 This is a flowchart illustrating a communication interference troubleshooting method provided in Embodiment 1 of the present invention. Figure 3 As shown, the method includes: acquiring the raw interference waveform data reported by the base station, performing interference data cleaning and mean normalization preprocessing to generate a standardized target interference feature pattern; subsequently, the feature pattern is input to two parallel processing channels: one channel calculates the first matching coefficient by combining Euclidean distance matching with a variable-temperature Softmax function. Another approach obtains the second matching coefficient through frequency segment classification training based on the DeepSeek V3 large model. Simultaneously, the system extracts expert experience coefficients from a pre-set interference knowledge base based on the target interference's frequency band, standard, and geographical location. Based on this, an identification correction coefficient is introduced. The objective matching results are compensated to generate corrected matching coefficients. and compared it with the expert experience coefficient The comprehensive discrimination rate is obtained by fusion. Finally, the candidate interference type with the highest comprehensive discrimination rate is selected as the root cause of interference. The intelligent agent (DeepSeek large model) combines the preset interference knowledge base and external network search plugin to dynamically generate interference investigation suggestions containing specific investigation steps. At the same time, the expert experience coefficient can be continuously updated through the coefficient iteration optimization mechanism to realize a closed-loop self-optimizing intelligent interference localization process.

[0100] The technical solution of this invention comprehensively considers the preprocessing of interference data, the calculation of the first matching coefficient based on feature distance metric, the calculation of the second matching coefficient based on waveform feature recognition, the introduction of expert experience coefficients based on prior experience, the adaptation of different scene features (including target communication standard, target communication frequency band, and target geographical information), objective matching compensation based on correction coefficients, and iterative optimization of expert experience coefficients. As the iterative optimization proceeds, the expert experience coefficients gradually converge to a better value, enabling the algorithm to exhibit stronger adaptability and intelligence in dynamic network environments. It can achieve adaptive optimization based on actual scene characteristics, thereby achieving more intelligent and efficient interference root cause localization and outputting more scientific and accurate troubleshooting suggestions.

[0101] Example 2 Figure 4 This is a schematic diagram of a communication interference detection device provided in Embodiment 2 of the present invention. Figure 4 As shown, the device includes: The interference data acquisition module 21 is used to acquire the interference data corresponding to the target interference; the interference data includes at least the interference value measured by the base station in the frequency domain resources; The determination information module 22 is used to determine the corresponding target interference determination information based on the scene characteristics and interference characteristics of the interference data; the target interference determination information includes at least: a first matching coefficient based on feature distance measurement, a second matching coefficient based on waveform feature recognition, and an expert experience coefficient based on prior experience; The interference type determination module 23 is used to determine the type of target interference based on the target interference determination information. The troubleshooting suggestion output module 24 is used to output troubleshooting suggestions corresponding to the type of target interference.

[0102] Furthermore, based on the above embodiments of the invention, the scene features include at least one of the following: target communication standard, target communication frequency band, and target geographical information; the interference features include a target interference feature pattern, which is obtained through the following methods: The interference values ​​of each resource block are extracted from the interference data to obtain the resource block interference value sequence; The mean normalization process is performed on the resource block interference value sequence to obtain the target interference feature map.

[0103] Furthermore, based on the above embodiments of the invention, the first matching coefficient is obtained in the following manner: According to the target communication standard and target communication frequency band, retrieve the pre-classified and constructed preset interference knowledge base, and determine the corresponding candidate interference types and the candidate interference feature patterns contained in each candidate interference type in the preset interference knowledge base. Determine the feature distance between the target interference feature map and each candidate interference feature map under each candidate interference type; The minimum value among the feature distances determined under the same candidate interference type is determined as the baseline distance for the corresponding candidate interference type; Determine the mean of the baseline distance for all candidate interference types, and use the mean as the temperature coefficient of the preset variable-temperature Softmax function; The temperature coefficient and the baseline distance of each candidate interference type are input into the preset variable temperature Softmax function for normalization processing to obtain the first matching coefficient corresponding to each candidate interference type.

[0104] Furthermore, based on the above embodiments of the invention, the second matching coefficient is obtained in the following manner: According to the target communication standard and target communication frequency band, retrieve the pre-trained preset waveform recognition model; The target interference feature pattern is input into the preset waveform recognition model, and the matching degree between the target interference and each candidate interference type is obtained as the second matching coefficient.

[0105] Furthermore, based on the above embodiments of the invention, the expert experience coefficient is obtained in the following way: Based on the target communication standard, target communication frequency band, and target geographical information, obtain historical interference investigation data for the corresponding scenario; The frequency of occurrence of each candidate interference type in historical interference investigation data is statistically analyzed, and the frequency is used as the expert experience coefficient for the corresponding candidate interference type.

[0106] Furthermore, based on the above embodiments of the invention, the interference type determination module 23 is specifically used for: For each candidate interference type, the product of the corresponding first matching coefficient, second matching coefficient, and expert experience coefficient is used as the comprehensive discrimination rate of the corresponding candidate interference type. The candidate interference type with the highest overall discrimination rate is determined as the target interference type.

[0107] Furthermore, based on the above embodiments of the invention, the troubleshooting suggestion output module 24 is specifically used for: Extract the first investigation suggestion corresponding to the target interference type from the preset interference knowledge base, and / or obtain the second investigation suggestion corresponding to the target interference type through online search; Output the first investigation suggestion and / or the second investigation suggestion.

[0108] Furthermore, based on the above embodiments of the invention, the interference type determination module 23 is also used for: Before determining the overall discrimination rate, obtain the correction coefficients corresponding to each candidate interference type; For each candidate interference type, the corresponding corrected matching coefficient is determined based on the corresponding first matching coefficient, second matching coefficient, and correction coefficient; Based on the corrected matching coefficients and expert experience coefficients, the comprehensive discrimination rate corresponding to each candidate interference type is re-determined.

[0109] Furthermore, based on the above embodiments of the invention, the correction coefficient is obtained in the following manner: Obtain the first training set, which contains multiple first interference samples under each candidate interference type. Each first interference sample is associated with a corresponding interference type label. Determine the first matching coefficient and the second matching coefficient corresponding to each first interference sample, and take the product of the first matching coefficient and the second matching coefficient as the joint matching coefficient of the corresponding first interference sample; The average joint matching coefficient of all first interference samples under the same candidate interference type is determined as the average joint matching coefficient of the corresponding candidate interference type. The maximum value among the average joint matching coefficients of all candidate interference types is determined as the baseline matching coefficient; For each candidate interference type, the difference between the corresponding average joint matching coefficient and the baseline matching coefficient is used as the correction coefficient for the corresponding candidate interference type.

[0110] Furthermore, based on the above-described embodiments of the invention, It also includes a step of optimizing the expert experience coefficients, and the interference type determination module 23 is also used for: Obtain the second training set, which contains multiple second interference samples classified according to the combination of target communication standard, target communication frequency band and target geographical information. Each second interference sample is associated with a corresponding interference type label, and each interference type label is associated with a correction coefficient. For each second interference sample, expert experience coefficients are obtained according to the target communication standard, target communication frequency band and target geographical information for the corresponding scenario. For each second interference sample, the corresponding predicted interference type is determined based on the corresponding first matching coefficient, second matching coefficient, correction coefficient, and expert experience coefficient. The number of target samples in the second training set whose predicted interference type matches the associated interference type label is counted, and the corresponding interference recognition accuracy is determined based on the number of target samples and the total number of samples in the second training set. With the objective function of maximizing the accuracy of interference identification, and with the expert experience coefficient as the optimization variable, an expert experience coefficient optimization model is constructed. The expert experience coefficient optimization model is iteratively adjusted using a preset gradient descent method. After each iteration, the interference recognition accuracy corresponding to the second training set is recalculated until the interference recognition accuracy converges to a preset threshold or the number of iterations reaches a preset upper limit. The expert experience coefficients obtained after iterative convergence are determined as the optimized expert experience coefficients.

[0111] Furthermore, based on the above embodiments of the invention, the interference type determination module 23 is also used for: Obtain the correction coefficients corresponding to each candidate interference type, as well as the optimized expert experience coefficients corresponding to the scene where the target interference is located; For each candidate interference type, the corresponding corrected matching coefficient is determined based on the corresponding first matching coefficient, second matching coefficient, and correction coefficient; For each candidate interference type, the product of the corresponding corrected matching coefficient and the expert experience coefficient is used as the comprehensive discrimination rate of the corresponding candidate interference type; The candidate interference type with the highest overall discrimination rate is determined as the target interference type.

[0112] The communication interference investigation device provided in this embodiment of the invention can execute the communication interference investigation method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0113] Example 3 Figure 5 A schematic diagram of an electronic device 30 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0114] like Figure 5As shown, the electronic device 30 includes at least one processor 31 and a memory, such as a read-only memory (ROM) 32 or a random access memory (RAM) 33, communicatively connected to the at least one processor 31. The memory stores computer programs executable by the at least one processor. The processor 31 can perform various appropriate actions and processes based on the computer program stored in the ROM 32 or loaded from storage unit 38 into the RAM 33. The RAM 33 can also store various programs and data required for the operation of the electronic device 30. The processor 31, ROM 32, and RAM 33 are interconnected via a bus 34. An input / output (I / O) interface 35 is also connected to the bus 34.

[0115] Multiple components in electronic device 30 are connected to I / O interface 35, including: input unit 36, such as keyboard, mouse, etc.; output unit 37, such as various types of monitors, speakers, etc.; storage unit 38, such as disk, optical disk, etc.; and communication unit 39, such as network card, modem, wireless transceiver, etc. Communication unit 39 allows electronic device 30 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0116] Processor 31 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 31 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 31 performs the various methods and processes described above, such as communication interference troubleshooting methods.

[0117] In some embodiments, the communication interference detection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 38. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 30 via ROM 32 and / or communication unit 39. When the computer program is loaded into RAM 33 and executed by processor 31, one or more steps of the communication interference detection method described above may be performed. Alternatively, in other embodiments, processor 31 may be configured to perform the communication interference detection method by any other suitable means (e.g., by means of firmware).

[0118] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0119] In some embodiments, the communication interference troubleshooting method may be implemented as a computer program, which is implicitly included in a computer program product. When executed by a processor, the computer program implements the communication interference troubleshooting method of the present invention. The computer program product can be understood as a software product that primarily implements its solution through a computer program. The computer program used to implement the method of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer program causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer program may be executed entirely on a machine, partially on a machine, partially on a remote machine as a standalone software package, or entirely on a remote machine or server.

[0120] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0121] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0122] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0123] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0124] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0125] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for troubleshooting communication interference, characterized in that, The method includes: Obtain the interference data corresponding to the target interference; the interference data includes at least the interference values ​​measured by the base station in the frequency domain resources; Based on the scene characteristics and interference characteristics of the interference data, the corresponding target interference determination information is determined; the target interference determination information includes at least: a first matching coefficient based on feature distance metric, a second matching coefficient based on waveform feature recognition, and an expert experience coefficient based on prior experience; Based on the target interference determination information, the target interference type to which the target interference belongs is determined; Output troubleshooting suggestions corresponding to the target interference type.

2. The method according to claim 1, characterized in that, The scene features include at least one of the following: target communication standard, target communication frequency band, and target geographical information; the interference features include a target interference feature pattern, which is obtained through the following methods: The interference values ​​of each resource block are extracted from the interference data to obtain a sequence of resource block interference values; The mean normalization process is performed on the resource block interference value sequence to obtain the target interference feature pattern.

3. The method according to claim 2, characterized in that, The first matching coefficient is obtained in the following way: According to the target communication standard and the target communication frequency band, a pre-classified and constructed preset interference knowledge base is retrieved, and the corresponding candidate interference types and the candidate interference feature patterns contained in each candidate interference type are determined in the preset interference knowledge base. Determine the feature distance between the target interference feature pattern and each candidate interference feature pattern under each candidate interference type; The minimum value among the feature distances determined under the same candidate interference type is determined as the reference distance for the corresponding candidate interference type; Determine the mean of the reference distances for all the candidate interference types, and use the mean as the temperature coefficient of the preset temperature-variable Softmax function; The temperature coefficient and the reference distance of each candidate interference type are input into the preset variable temperature Softmax function for normalization processing to obtain the first matching coefficient corresponding to each candidate interference type.

4. The method according to claim 3, characterized in that, The second matching coefficient is obtained in the following way: According to the target communication standard and the target communication frequency band, retrieve the pre-trained preset waveform recognition model; The target interference feature pattern is input into the preset waveform recognition model to obtain the matching degree between the target interference and each of the candidate interference types as the second matching coefficient.

5. The method according to claim 3, characterized in that, The expert experience coefficient is obtained in the following way: Based on the target communication standard, the target communication frequency band, and the target geographical information, obtain historical interference investigation data for the corresponding scenario; The frequency of occurrence of each candidate interference type in the historical interference investigation data is statistically analyzed, and the frequency of occurrence is used as the expert experience coefficient corresponding to the candidate interference type.

6. The method according to claim 3, characterized in that, The step of determining the target interference type based on the target interference determination information includes: For each of the candidate interference types, the product of the corresponding first matching coefficient, second matching coefficient, and expert experience coefficient is used as the comprehensive discrimination rate for the corresponding candidate interference type. The candidate interference type with the highest overall discrimination rate is determined as the target interference type.

7. The method according to claim 1, characterized in that, The output of troubleshooting suggestions corresponding to the target interference type includes: Extract a first investigation suggestion corresponding to the target interference type from a preset interference knowledge base, and / or obtain a second investigation suggestion corresponding to the target interference type through online search; Output the first investigation suggestion and / or the second investigation suggestion.

8. The method according to claim 6, characterized in that, Before determining the overall discrimination rate, the method further includes a step of correcting the first matching coefficient and the second matching coefficient, specifically including: Obtain the correction coefficients corresponding to each of the candidate interference types; For each of the candidate interference types, a corresponding corrected matching coefficient is determined based on the corresponding first matching coefficient, second matching coefficient, and correction coefficient; Based on the corrected matching coefficients and the expert experience coefficients, the comprehensive discrimination rate corresponding to each candidate interference type is re-determined.

9. The method according to claim 8, characterized in that, The correction coefficient is obtained in the following manner: Obtain a first training set, which contains multiple first interference samples under each candidate interference type, and each first interference sample is associated with a corresponding interference type label. Determine the first matching coefficient and the second matching coefficient corresponding to each of the first interference samples, and take the product of the first matching coefficient and the second matching coefficient as the joint matching coefficient of the corresponding first interference sample; The average joint matching coefficient of all the first interference samples under the same candidate interference type is determined as the average joint matching coefficient of the corresponding candidate interference type. The maximum value among the average joint matching coefficients of all the candidate interference types is determined as the baseline matching coefficient; For each of the candidate interference types, the difference between the corresponding average joint matching coefficient and the baseline matching coefficient is used as the correction coefficient for the corresponding candidate interference type.

10. The method according to claim 8, characterized in that, It also includes a step of optimizing the expert experience coefficient, specifically including: Obtain a second training set, which contains multiple second interference samples classified according to the combined scenario of target communication standard, target communication frequency band and target geographical information. Each second interference sample is associated with a corresponding interference type label, and each interference type label is associated with the correction coefficient. For each of the second interference samples, the expert experience coefficients for the corresponding scenario are obtained according to the target communication standard, the target communication frequency band, and the target geographical information. For each of the second interference samples, the corresponding predicted interference type is determined based on the corresponding first matching coefficient, second matching coefficient, correction coefficient, and expert experience coefficient. The number of target samples in the second training set that match the predicted interference type with the associated interference type label is counted, and the corresponding interference recognition accuracy is determined based on the number of target samples and the total number of samples in the second training set. An expert experience coefficient optimization model is constructed with the objective function of maximizing the interference recognition accuracy and the expert experience coefficient as the optimization variable. The expert experience coefficient optimization model is iteratively adjusted using a preset gradient descent method. After each iteration, the interference recognition accuracy corresponding to the second training set is recalculated until the interference recognition accuracy converges to a preset threshold or the number of iterations reaches a preset upper limit. The expert experience coefficients obtained after iterative convergence are determined as the optimized expert experience coefficients.

11. The method according to claim 10, characterized in that, Determining the target interference type to which the target interference belongs includes: Obtain the correction coefficients corresponding to each of the candidate interference types, and the optimized expert experience coefficients corresponding to the scene where the target interference is located; For each of the candidate interference types, a corresponding corrected matching coefficient is determined based on the corresponding first matching coefficient, second matching coefficient, and correction coefficient; For each of the candidate interference types, the product of the corresponding corrected matching coefficient and the expert experience coefficient is used as the comprehensive discrimination rate for the corresponding candidate interference type. The candidate interference type with the highest overall discrimination rate is determined as the target interference type.

12. A communication interference detection device, characterized in that, The device includes: An interference data acquisition module is used to acquire interference data corresponding to a target interference; the interference data includes at least the interference values ​​measured by the base station in the frequency domain resources; The determination information module is used to determine the corresponding target interference determination information based on the scene characteristics and interference characteristics of the interference data; the target interference determination information includes at least: a first matching coefficient based on feature distance metric, a second matching coefficient based on waveform feature recognition, and an expert experience coefficient based on prior experience; An interference type determination module is used to determine the target interference type to which the target interference belongs based on the target interference determination information; The troubleshooting suggestion output module is used to output troubleshooting suggestions corresponding to the target interference type.

13. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the communication interference investigation method according to any one of claims 1-11.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that are used to cause a processor to execute the communication interference detection method according to any one of claims 1-11.

15. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the communication interference detection method according to any one of claims 1-11.

Images