Interference identification method and device and electronic equipment

Abstract

The invention discloses an interference identification method and device and electronic equipment. The method comprises the steps that an interference cell set is acquired, the interference cell set comprises historical interference cells, simulation interference cells and interference cells to be analyzed, and each interference cell comprises corresponding time domain index data, frequency domain index data and channel index data; a graph structure corresponding to the interference cell set is determined, node features in the graph structure are determined, each interference cell is a node in the graph structure, and the node features are index features obtained after time domain index data, frequency domain index data and channel index data are fused; and training an interference prediction model according to the graph structure, the node features and the known interference label, and predicting the interference cell to be analyzed through the trained interference prediction model to obtain an interference identification result. According to the method and the device, the technical problems of data scarcity and insufficient practical engineering application availability in related technologies are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wireless communication, in particular to an interference identification method and device and electronic equipment. BACKGROUND

[0002] With the successive deployment and evolution of 3G, 4G and even 5G networks, modern wireless communication systems are facing unprecedented architectural changes and technical challenges. In particular, under the background of the current mixed deployment of 4G and 5G base stations, combined with the implementation of the low-frequency re-cultivation and co-construction and sharing strategy among operators, multiple network standards coexist has become the norm, which significantly improves the complexity of wireless networks. In this complex network environment, the difficulty of external interference troubleshooting increases dramatically, and due to the growth of network deployment time and the cumulative effect of equipment aging, internal interference problems are increasingly prominent, which has a non-negligible impact on network performance.

[0003] In the traditional practice of interference identification and processing, network optimization engineers often rely on manually extracted network performance parameters to visualize the interference status of the interference cell and its co-frequency cell with the help of geographic information systems such as MapInfo, and then supplement the intuitive interpretation of the interference spectrum with artificial experience. This process can reveal some interference clues, but its efficiency is limited by manpower and highly dependent on individual professional judgment and skills. Especially in the face of multiple complex interference scenarios, manual troubleshooting methods are not up to the task and are difficult to quickly and accurately determine the type of interference and develop effective solutions.

[0004] In recent years, with the rapid development of machine learning and big data technologies, significant progress has been made in the field of interference analysis automation. Interference identification methods based on clustering algorithms, such as FINCH, can to some extent identify external interference sources and cluster similar interference spectrum cells through statistical analysis of network performance data, and assist in frequency domain interference type judgment with the help of expert knowledge. However, this kind of method often performs better in simulation environments than in actual engineering applications, mainly due to the data scarcity problem encountered in real-world scenarios and the decline in recognition ability when faced with multiple interference factors coexisting.

[0005] Meanwhile, deep learning techniques are gradually applied to communication interference analysis, especially in the field of radar jamming recognition, showing great potential. Such methods often combine artificial feature extraction and automatic feature learning, and predict the interference type through means such as Naive Bayes classifier. However, even in the field of deep learning, interference analysis technology faces two major challenges: one is the high data demand, most of the literature generates interference signals through simulation as training samples, but the difference between this and the real world may lead to poor model generalization ability; the second is the difficulty of type judgment in complex interference scenarios, when multiple interference factors are intertwined, existing methods are difficult to accurately identify the interference type, especially in practical applications, the problems of data scarcity and recognition accuracy need to be overcome.

[0006] At present, there is no effective solution to the above problems. SUMMARY

[0007] Embodiments of the present application provide an interference recognition method, device and electronic equipment to at least solve the technical problems of data scarcity and insufficient practical engineering application availability of related technologies.

[0008] According to an aspect of an embodiment of the present application, an interference recognition method is provided, comprising: obtaining a set of interference cells, wherein the set of interference cells includes historical interference cells, simulated interference cells and to-be-analyzed interference cells, each interference cell includes corresponding time domain index data, frequency domain index data and channel index data; determining a graph structure corresponding to the set of interference cells, and determining node features in the graph structure, wherein each interference cell is a node in the graph structure, and the node feature is an index feature after fusing the time domain index data, the frequency domain index data and the channel index data; training an interference prediction model according to the graph structure, the node feature and a known interference label, and predicting the to-be-analyzed interference cell through the trained interference prediction model to obtain an interference recognition result.

[0009] Optionally, obtaining the set of interference cells comprises: obtaining a plurality of internal interference samples and first external interference samples from an interference investigation case library, and determining the historical interference cells according to the internal interference samples and the first external interference samples, wherein the number of internal interference samples is less than that of the first external interference samples; generating a plurality of second external interference samples of different interference types through a simulation software according to preset interference conditions, and determining the simulated interference cells according to the second external interference samples; and aggregating the historical interference cells, the simulated interference cells and the to-be-analyzed interference cells to obtain the set of interference cells.

[0010] Optionally, the determining the graph structure corresponding to the set of interfering cells comprises: determining physical positions of each interfering cell in the set of interfering cells, wherein the physical position is longitude and latitude information of the interfering cell; determining interference relationships between adjacent interfering cells in the set of interfering cells, wherein the interference relationship is used to reflect an interference correlation degree between the adjacent interfering cells; and determining the graph structure according to the physical positions and the interference relationships.

[0011] Optionally, the method further comprises: determining distances between the adjacent interfering cells, and determining that there is a connection relationship between the adjacent interfering cells if the distance is less than a preset distance threshold; and determining interference similarities between the adjacent interfering cells, and determining that there is a connection relationship between the adjacent interfering cells if the interference similarity is greater than a preset similarity threshold.

[0012] Optionally, the determining the node feature in the graph structure comprises: processing time domain index data of a target interfering cell by a time series analysis model to obtain a time domain feature, wherein the target interfering cell is any one of the set of interfering cells; respectively determining a frequency domain feature and a channel feature corresponding to frequency domain index data and channel index data of the target interfering cell, and performing standardization processing on the frequency domain feature and the channel feature; and splicing the time domain feature, the frequency domain feature and the channel feature to obtain the node feature of the target interfering cell.

[0013] Optionally, the method further comprises: generating a node embedding representation corresponding to the target interfering cell by the trained interference prediction model, wherein the node embedding representation is a fusion result of the node feature of the target interfering cell and node features of neighbor interfering cells of the target interfering cell; determining a connection relationship in the graph structure, and propagating data information in the known interference label to the to-be-analyzed interfering cell according to the connection relationship and the node embedding representation to obtain an interference identification result of the to-be-analyzed interfering cell.

[0014] Optionally, the method further comprises: determining an interference type and a prediction probability corresponding to the to-be-analyzed interfering cell during the propagation of the data information in the known interference label; and determining the interference type corresponding to the prediction probability as the interference identification result of the to-be-analyzed interfering cell if the prediction probability is greater than a preset threshold.

[0015] Optionally, the method further comprises: generating a visual report of the to-be-analyzed interfering cell according to the interference identification result, wherein the visual report comprises at least one of the following: an interference time domain analysis graph, an interference frequency domain analysis graph and an interference range analysis graph of the to-be-analyzed interfering cell; and generating a corresponding interference troubleshooting suggestion for different interference types in the interference identification result.

[0016] According to another aspect of the embodiments of the present application, the method further includes: obtaining a set of interference cells, wherein the set of interference cells includes historical interference cells, simulated interference cells and interference cells to be analyzed, each of the interference cells including corresponding time domain index data, frequency domain index data and channel index data; determining a graph structure corresponding to the set of interference cells and determining node features in the graph structure, wherein each of the interference cells is a node in the graph structure, and the node features are index features after fusing the time domain index data, the frequency domain index data and the channel index data; training an interference prediction model according to the graph structure, the node features and known interference labels, and predicting the interference cells to be analyzed by using the trained interference prediction model to obtain an interference identification result.

[0017] According to still another aspect of the embodiments of the present application, an electronic device is provided, including a memory and a processor, wherein the memory is configured to store program instructions, and the processor is connected to the memory and configured to execute the interference identification method.

[0018] According to still another aspect of the embodiments of the present application, a non-volatile storage medium is provided, including a stored computer program, wherein a device in which the non-volatile storage medium is located executes the interference identification method by running the computer program.

[0019] According to still another aspect of the embodiments of the present application, a computer program product is provided, including computer instructions, which, when executed by a processor, implement the interference identification method.

[0020] In the embodiments of the present application, the set of interference cells is obtained, wherein the set of interference cells includes historical interference cells, simulated interference cells and interference cells to be analyzed, each of the interference cells including corresponding time domain index data, frequency domain index data and channel index data; the graph structure corresponding to the set of interference cells is determined, and the node features in the graph structure are determined, wherein each of the interference cells is a node in the graph structure, and the node features are index features after fusing the time domain index data, the frequency domain index data and the channel index data; the interference prediction model is trained according to the graph structure, the node features and known interference labels, and the interference cells to be analyzed are predicted by using the trained interference prediction model to obtain an interference identification result, thereby achieving the purpose of accurately identifying multiple interference types under the condition of data sample scarcity, and realizing the technical effect of improving the efficiency and accuracy of wireless network interference troubleshooting, and further solving the technical problem of insufficient data scarcity and actual engineering application availability of related technologies. BRIEF DESCRIPTION OF DRAWINGS

[0021] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application. In the drawings:

[0022] Figure 1 Fig. 1 is a hardware structure diagram of a computer terminal for implementing an interference identification method according to an embodiment of the present application;

[0023] Figure 2 Fig. 2 is a flowchart of an interference identification method according to an embodiment of the present application;

[0024] Figure 3 Fig. 3 is a flowchart of another interference identification method according to an embodiment of the present application;

[0025] Figure 4 Fig. 4 is a structural schematic diagram of an Encoder module of an Autoformer model according to an embodiment of the present application;

[0026] Figure 5 Fig. 5 is a training flow schematic diagram of a GraphSAGE model according to an embodiment of the present application;

[0027] Figure 6 Fig. 6 is a structural diagram of an interference identification device according to an embodiment of the present application. DETAILED DESCRIPTION

[0028] In order to enable persons skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by persons skilled in the art without creative labor should fall within the scope of protection of the present application.

[0029] It should be noted that the terms "first", "second", and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but can include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0030] First, some of the terms or terms that appear in the process of explaining the embodiments of the present application are applicable to the following explanations:

[0031] Semi-supervised learning: a machine learning method that uses a small amount of labeled data and a large amount of unlabeled data to train the model together, improving the generalization ability of the model and the depth of understanding of the data.

[0032] GraphSAGE (Graph Sample and Aggregate): a variant of graph neural network for node representation learning on large-scale graphs. GraphSAGE learns the embedding representation of nodes by sampling the neighbor information of nodes and aggregating, which is suitable for handling semi-supervised learning problems.

[0033] Autoformer: a Transformer architecture model designed for time series prediction. After introducing Auto-Correction into the Tranformer architecture, its features can more efficiently capture long-term dependencies and periodic changes in time series data.

[0034] PRB (Physical Resource Block): a basic resource unit in LTE (Long Term Evolution) and NR (New Radio) systems, used to carry wireless signals. Each PRB occupies a certain area in the time-frequency resource, specifically one slot and 12 consecutive subcarriers in a subframe, used to schedule user data and control information.

[0035] RSSI (Received Signal Strength Indicator): a commonly used parameter in wireless communication systems, used to measure the total power of the signal received by the receiving antenna. In a wireless network, RSSI can indicate the strength of the signal, including the sum of the desired signal and all interference signals.

[0036] To solve the problem of poor interference identification efficiency in the related art, the embodiments of the present application provide an interference identification method, which can be run in Figure 1 The computer terminal is illustrated in the following.

[0037] The interference identification method provided by the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 A hardware structure block diagram of a computer terminal for implementing the interference identification method is shown. As Figure 1As shown, the computer terminal 10 may include one or more processors (shown as 102a, 102b, ..., 102n in the figure) (the processor may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 104 for storing data, and a transmission module 106 for communication functions connected via wired and / or wireless networks. In addition, it may also include: a display, a keyboard, a cursor control device, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, and a BUS bus. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0038] It should be noted that the aforementioned one or more processors and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be implemented wholly or partially as software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be wholly or partially integrated into any other element in the computer terminal 10. As involved in the embodiments of this application, the data processing circuits serve as processor control (e.g., selection of a variable resistor termination path connected to an interface).

[0039] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the interference identification method in the embodiments of this application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the above-mentioned interference identification method. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor, and these remote memories can be connected to the computer terminal 10 via a network. Examples of the above-mentioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0040] The transmission module 106 is configured to receive or send data via a network. The network can include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission module 106 includes a network interface controller (NIC) that can connect to other network devices through a base station to communicate with the Internet. In one example, the transmission module 106 can be a radio frequency (RF) module that is configured to communicate with the Internet wirelessly.

[0041] The display can be a liquid crystal display (LCD) that is touch screen, for example, which can enable a user to interact with a user interface of the computer terminal 10.

[0042] It is noted that in some alternative embodiments, the above Figure 1 The computer terminal can include hardware elements (including circuitry), software elements (including computer code stored on a computer readable medium), or a combination of both hardware and software elements. It should be noted that in some embodiments, one or more of the functions described above can be implemented using software instructions that can be read by a computer system and executed using processing circuitry. In some embodiments, the software instructions can be stored in a non-transitory computer-readable medium, such as a memory device. Figure 1 is merely one example of a particular implementation and is intended to demonstrate a type of component that can exist in the computer terminal described above.

[0043] In the above operating environment, the embodiments of the present disclosure provide a method for interference identification. It is noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described herein can be executed in an order different from that shown herein.

[0044] Figure 2 is a flowchart of a method for interference identification according to an embodiment of the present disclosure, as Figure 2 shown, the method includes the following steps:

[0045] In step S202, a set of interference cells is obtained, wherein the set of interference cells includes historical interference cells, simulated interference cells, and interference cells to be analyzed, each interference cell includes corresponding time domain index data, frequency domain index data, and channel index data.

[0046] In step S204, a graph structure corresponding to the set of interference cells is determined, and node features in the graph structure are determined, wherein each interference cell is a node in the graph structure, and the node features are index features after fusing the time domain index data, the frequency domain index data, and the channel index data.

[0047] Step S206, the interference prediction model is trained according to the graph structure, node features and known interference labels, and the trained interference prediction model is used to predict the to-be-analyzed interference cell to obtain an interference identification result.

[0048] Through the above steps S202 to S206, the purpose of accurately identifying multiple interference types under the condition of data sample scarcity is achieved, thereby realizing the technical effect of improving the efficiency and accuracy of wireless network interference troubleshooting, and further solving the technical problems of related technologies in data scarcity and insufficient usability in actual engineering applications.

[0049] Figure 3 is a flowchart of another interference identification method according to an embodiment of the present application, as shown in Figure 3 The above interference identification method is more detailed in the implementation process, and is intended to solve the interference identification problem caused by data sample scarcity and coexistence of multiple interference factors in a wireless network, so as to improve the accuracy and efficiency of interference identification. The following Figure 3 The above steps S202 to S206 are described in detail.

[0050] For the data set construction stage, as shown in the above step S202, a set of interference cells is obtained, including: obtaining a plurality of internal interference samples and first external interference samples from an interference troubleshooting case library, and determining historical interference cells according to the internal interference samples and the first external interference samples, wherein the number of internal interference samples is less than that of the first external interference samples; generating a plurality of second external interference samples of different interference types by simulation software according to preset interference conditions, and determining simulation interference cells according to the second external interference samples; and aggregating the historical interference cells, the simulation interference cells and the to-be-analyzed interference cells to obtain the set of interference cells.

[0051] In the embodiments of the present application, a comprehensive and diverse set of interference cells is constructed, which involves collecting interference samples from different sources and finally integrating them into a comprehensive data set containing historical interference cells, simulation interference cells and to-be-analyzed interference cells. The specific analysis is as follows:

[0052] 1. Obtain internal and external interference samples from the interference troubleshooting case library.

[0053] First, real interference samples are collected from different troubleshooting case libraries of companies or enterprises to obtain structured data. The specific interference samples can include: 200 internal interference samples and 400 external interference samples (i.e. the first external interference samples), each of which contains time dimension, frequency domain dimension and channel dimension interference performance indicators of the interfered cell.

[0054] 2. Generate external interference samples using simulation software.

[0055] Through the communication simulation tool, 10 typical external interferences (such as common typical external interference samples of co-channel interference, spurious interference (adjacent channel interference), out-of-band interference, intermodulation interference, blocking interference, frame step-out interference, pseudo base station interference, repeater interference, and the like) are constructed according to interference principles or preset interference conditions, 1900 effective simulation samples, that is, the second external interference samples are generated. Each interference sample also contains time dimension, frequency domain dimension and channel dimension interference performance indicators of the interference cell.

[0056] 3. Aggregation of the interference cell set.

[0057] The information of the historical interference cell (the internal and first external interference samples based on actual cases), the simulation interference cell (the second external interference samples) and the interference cell to be analyzed is merged to construct an interference cell set containing 2500 effective samples. This set not only covers real interference cases, but also contains simulated interference scenarios, providing comprehensive and diverse data support for subsequent data processing, feature extraction and model training.

[0058] It should be noted that before determining the interference cell to be analyzed, a preliminary state evaluation needs to be performed. Specifically, first, a comprehensive network management alarm check needs to be performed on the cell to exclude performance abnormalities caused by device faults, configuration errors or software problems. If there is an alarm, such as an alarm information issued when the standing wave ratio is above 1.5, fault troubleshooting and processing must be performed in priority to ensure that all known problems are solved. Only when the alarm is completely cleared and the noise level of the cell is still higher than the normal range, that is, there is a problem of interference that cannot be solved by conventional maintenance, the cell enters the advanced interference identification process based on heterogeneous data fusion and semi-supervised network representation learning. To ensure the pertinence and efficiency of the identification process and avoid waste of resources on non-interference related problems.

[0059] For the multi-source heterogeneous data fusion stage, the interference performance indicator data of the time dimension, the frequency domain dimension and the channel dimension are fused to generate the node features of each interference cell, including: processing the time domain indicator data of the target interference cell through a time series analysis model to obtain time domain features, wherein the target interference cell is any one of the interference cell set; respectively determining the frequency domain features and channel features corresponding to the frequency domain indicator data and channel indicator data of the target interference cell, and performing standardization processing on the frequency domain features and channel features; splicing the time domain features, the frequency domain features and the channel features to obtain the node features of the target interference cell.

[0060] In the embodiments of the present application, the above-mentioned manner embodies the essence of cross-dimension data analysis, aiming to maximize the mining and integration of the implicit information of various interference signals and improve the accuracy of subsequent identification and prediction. The specific process analysis is as follows:

[0061] 1. Time series feature extraction.

[0062] An advanced time series analysis model, such as Autoformer, is used to perform deep processing on the time-domain index data of the target interference cell (any one cell in the interference cell set), and to extract time series interference features (i.e., time-domain features) reflecting periodicity and trend. The Encoder module of the Autoformer model can effectively capture the fluctuation rules of performance data in different periods such as days and weeks through autocorrelation analysis of time series, which is crucial for understanding the dynamic nature of interference.

[0063] The specific structure of the Encoder module is shown in Figure 4 , which mainly consists of three key parts: Series Decomposition Block, Auto-Correlation Mechanism, and Feed Forward. The Series Decomposition Block is used to decompose the input time dimension feature sequence (i.e., time-domain index data) into trend and periodicity parts. The periodicity part relies on the Auto-Correlation Mechanism to capture periodic dependence. The Feed Forward further processes and extracts the features.

[0064] 2. Static feature extraction and fusion.

[0065] The frequency-domain index data and channel index data of the target interference cell are standardized to obtain frequency-domain (interference) features and channel (interference) features. Standardization ensures that different types of interference performance indicators can be compared and spliced on the same scale, avoiding the bias of the numerical range of a certain dimension indicator on the overall analysis result. Frequency-domain features reflect the average interference level of each physical resource block (PRB), and channel features show the received signal strength indication (RSSI) of each antenna channel.

[0066] These two static interference features are closely combined with dynamic time-domain features through data normalization and cross-filling technology to form a comprehensive, multi-dimensional feature matrix, i.e., the node features of the target interference cell.

[0067] In this stage, through cross-dimension feature extraction, standardization and integration, high-quality input is provided for subsequent semi-supervised learning and node representation of the GraphSAGE model, effectively improving the accuracy and robustness of the interference identification method, and providing strong technical support for solving the interference troubleshooting problem in wireless communication networks.

[0068] For the semi-supervised learning stage, as shown in steps S204 and S206 above, first, the graph structure needs to be constructed using the physical location (latitude and longitude information) of the interference cells and the interference correlation (interference relationship between adjacent cells). Then, the interference prediction model (such as the GraphSAGE model) and its classifier are trained, and the trained model is used to predict and identify the cells to be analyzed to obtain the interference identification result.

[0069] Optionally, the graph structure corresponding to the set of interference cells is determined, including: determining the physical location of each interference cell in the set of interference cells, wherein the physical location is the latitude and longitude information of the interference cell; determining the interference relationship between adjacent interference cells in the set of interference cells, wherein the interference relationship is used to reflect the interference correlation degree between adjacent interference cells; determining the graph structure according to the physical location and the interference relationship.

[0070] In the embodiments of the present application, the graph structure corresponding to the set of interference cells is created, providing a framework and context for subsequent GraphSAGE model learning and interference prediction. This process includes determining the physical location of each cell, evaluating the interference correlation degree between adjacent cells, and constructing the graph structure based on this information.

[0071] Specifically, first, the physical location (latitude and longitude information) of each cell in the set of interference cells can be determined by collecting and organizing data to form the node coordinates in the graph structure. Second, by analyzing the interference data (such as interference level fluctuations, correlation coefficients, etc.) between adjacent interference cells, the interference relationship (interference correlation degree) between adjacent interference cells can be measured. This interference relationship can be based on statistical methods such as cosine similarity to identify adjacent cells that are easily affected by common or mutual interference due to similar geographical location or similar frequency point configuration. Finally, based on these physical locations and interference relationships, the nodes and edges of the graph are constructed to form the graph structure. Each node represents an interference cell, and the edges between nodes are given different weights according to the proximity of geographical location and interference correlation. The smaller the distance between adjacent cells, the higher the interference similarity, and the greater the weight, to reflect the strength and direction of interference between cells.

[0072] Optionally, in the process of constructing the graph structure, the distance between the adjacent interference cells is determined, and the connection relationship between the adjacent interference cells is determined if the distance is less than a preset distance threshold.

[0073] In the embodiments of the present application, considering the influence of frequency point configuration and coverage range on interference, the concept of distance threshold function is introduced to optimize the construction rule of the edges in the graph, ensuring that only the cells that are physically possible to interfere with each other can form a connection. Specifically, by setting a distance threshold, only when the physical distance between two real cells is less than the preset distance threshold, an edge between the two cells in the graph is added to form a connection relationship, making the design of the graph structure closer to the actual scenario of wireless communication, which helps the model focus on the most likely interference relationship and improves the accuracy of interference type prediction.

[0074] Meanwhile, for the evaluation of interference relevance, an intuitive and effective cosine similarity method is adopted and applied to the interference performance index data in the time domain and the frequency domain. Specifically, the cosine similarity of two cells in the time sequence interference feature and the spectrum interference feature is calculated, and then the average value is taken to quantify the interference correlation degree between them. Only when the weighted average similarity exceeds a preset similarity threshold (such as 0.4), it is considered that there is significant interference correlation between the two cells, so that an edge is established in the graph to form a connection relationship, so as to strengthen the understanding of the potential interference relationship by the graph structure.

[0075] In the above step S206, the trained interference prediction model is used to predict the interference cells to be analyzed to obtain the interference identification result, including: generating a node embedding representation corresponding to the target interference cell through the interference prediction model, wherein the node embedding representation is the fusion result of the node features of the target interference cell and the node features of the neighbor interference cells of the target interference cell; determining the connection relationship in the graph structure, and propagating the data information in the known interference label to the interference cells to be analyzed according to the connection relationship and the node embedding representation, to obtain the interference identification result of the interference cells to be analyzed.

[0076] In the embodiments of the present application, the GraphSAGE model is used as the interference prediction model. The GraphSAGE has advantages in processing small sample size and heterogeneous data sets. Through the semi-supervised learning mechanism, the model can use the topological information and node features in the graph structure to improve the accuracy and robustness of prediction under the guidance of a small amount of labeled data. The training process of the GraphSAGE model is shown in Figure 5 as follows:

[0077] 1. Initialize the graph structure and node features.

[0078] Before training the GraphSAGE model, the graph structure needs to be initialized, where nodes represent individual cells, and edges are constructed based on physical location relationships (Euclidean distance of latitude and longitude) and interference relevance (cosine similarity in time and frequency domains).

[0079] Node features include time series features extracted by Autoformer, average interference level per PRB, and RSSI of each antenna channel. These features are processed through normalization and cross-padding techniques to ensure data quality and consistency of model input.

[0080] 2. Label propagation of semi-supervised learning.

[0081] Use historical interference cells with known interference labels and simulated interference cells as labeled data, and the remaining interference cells to be analyzed as unlabeled data.

[0082] The GraphSAGE model generates node embedding representations through domain sampling and feature aggregation, which are the result of combining node features and neighbor node features. This can reflect the node's position in the graph structure and its potential interference relationship with other nodes.

[0083] Then, combine the connection relationship in the graph structure and the weight of each edge to propagate the known interference labels in the graph, gradually enriching the information of unlabeled cells, and guiding the model to learn how to predict interference types based on structure and features during training.

[0084] 3. Model training and optimization.

[0085] Use a multi-class classifier (such as Softmax layer) to classify nodes, and the training goal is to minimize the difference between the predicted labels and the actual interference labels of labeled cells (usually using cross-entropy loss function).

[0086] Adjust the model parameters through backpropagation algorithm until convergence or reach the predetermined training rounds to optimize the generalization ability and prediction accuracy of the model.

[0087] 4. Interference type prediction and verification.

[0088] After training, use the GraphSAGE model to predict the unlabeled cells in the graph, output the interference type and its corresponding probability, and get the interference identification result.

[0089] In addition, cross-validation or reserving a portion of test data can be used to evaluate the performance of the model, ensuring the effectiveness and reliability of the model in actual interference identification tasks.

[0090] Optionally, the method further comprises: determining the interference type and the prediction probability of the to-be-analyzed interference cell in the propagation process of the data information in the known interference label; and determining the interference type corresponding to the to-be-analyzed interference cell as the interference identification result of the to-be-analyzed interference cell when the prediction probability is greater than a preset threshold.

[0091] In the embodiments of the present application, the GraphSAGE model is responsible for capturing and learning the potential interference relationship between nodes in the graph structure, and determining the interference type and the prediction probability of the to-be-analyzed interference cell through the propagation mechanism of the known interference label. This mechanism is particularly important in the environment of scarce data labeling. It uses limited labeled data to predict the interference type of other cells in a probabilistic manner through the connection and node features in the graph, and realizes effective inference from known to unknown.

[0092] Specifically, the node representation learned by the model during the training process can reflect the characteristics of the cell in the time, frequency and channel dimensions, as well as the interference association with other cells. When propagating the known label, the model will gradually adjust the interference type prediction of the unlabeled cell based on these representations and the edge weights in the graph structure. Through iterative learning, the prediction probability of the unlabeled cell gradually tends to be accurate. Once the prediction probability of a certain interference type exceeds a preset threshold (such as 50%), it will be taken as the interference identification result of the cell. For example, for a certain cell, the model outputs that the probability of the cell existing in the repeater interference in the external interference is 70%, the probability of the cell existing in the intermodulation interference is 60%, and the probabilities of the cell existing in the other interference types are all less than 50%. Therefore, the interference types are identified as repeater interference and intermodulation interference. In this way, the model can accurately give the interference prediction under various interference factors and provide a basis for subsequent interference troubleshooting.

[0093] Optionally, the method further comprises: generating a visual report of the to-be-analyzed interference cell according to the interference identification result, wherein the visual report comprises at least one of the following: an interference time domain analysis graph, an interference frequency domain analysis graph, and an interference range analysis graph of the to-be-analyzed interference cell; and generating interference troubleshooting suggestions corresponding to different interference types in the interference identification result.

[0094] In the embodiments of the present application, an automatic generation mechanism of interference troubleshooting suggestions is also provided.

[0095] Specifically, after the interference identification is completed, the system will generate a visual report of the interference analysis, including but not limited to an interference time domain analysis graph, an interference frequency domain analysis graph, and an interference range analysis graph, to help network optimization engineers quickly understand and judge the interference type and its impact range.

[0096] Further, for the identified internal interference, the system will automatically generate troubleshooting suggestions according to the specific interference type, for example: for the internal interference of the macro station cell, the operation suggestion of the antenna feeder pair is provided, and the expected effect after the pair is predicted; for the internal interference of the room division cell, the RRU and antenna feeder troubleshooting suggestion is provided to help the engineers locate the interference source.

[0097] And for the identified external interference, the system will give specific reference troubleshooting manual, for example: for the interference that needs to be tested by frequency sweep, the frequency sweeper configuration setting suggestion is given, and the branch diagram of the expected frequency sweep result is provided for engineers to verify and troubleshoot on site; for the case of intermodulation detection by intermodulation tester, the corresponding suggestion is also given for reference by network optimization engineers.

[0098] The interference identification method provided in the present application innovatively introduces heterogeneous data fusion technology and semi-supervised network representation learning, which is characterized in that: first, a small data set sample of comprehensive real cases and simulation data is constructed, which ingeniously solves the problem of insufficient structured data sample quantity in related technologies; second, a multi-source heterogeneous data fusion architecture is developed, periodic time sequence features are extracted by Autoformer, and are fused with frequency domain and channel static features, realizing comprehensive analysis of interference performance indicators; third, the GraphSAGE model is creatively used for semi-supervised learning, and the interference label is propagated in the graph structure, effectively identifying multi-factor interference in complex environments, and significantly improving the accuracy and robustness of interference identification.

[0099] Overall, the present application not only optimizes the time-consuming and dependence on manual experience of traditional interference identification methods, but also realizes high-precision interference type prediction under the condition of data scarcity, greatly promoting the development of wireless communication network optimization technology, and providing a new solution for automatic interference troubleshooting.

[0100] According to an embodiment of the present application, an interference identification device is provided. It should be noted that the interference identification device of the present application can be used to execute the interference identification method provided by the present application. The interference identification device provided by the present application is introduced as follows.

[0101] Figure 6 is a structural diagram of an interference identification device according to an embodiment of the present application. As shown in Figure 6 , the device includes:

[0102] The acquisition module 60 is configured to acquire a set of interference cells, wherein the set of interference cells includes historical interference cells, simulation interference cells and interference cells to be analyzed, and each interference cell includes corresponding time domain index data, frequency domain index data and channel index data.

[0103] The determining module 62 is configured to determine a graph structure corresponding to the set of interfering cells, and determine node features in the graph structure, wherein each interfering cell is a node in the graph structure, and the node features are index features after fusion of the time-domain index data, the frequency-domain index data, and the channel index data.

[0104] The identifying module 64 is configured to train an interference prediction model according to the graph structure, the node features, and known interference labels, and perform prediction on the to-be-analyzed interfering cell by using the trained interference prediction model to obtain an interference identification result.

[0105] Through the acquisition module, the determining module, and the identifying module in the interference identification device, the purpose of accurately identifying multiple interference types under the condition of scarce data samples is achieved, thereby achieving the technical effect of improving the efficiency and accuracy of wireless network interference troubleshooting, and further solving the technical problem of insufficient usability of related technologies in data scarcity and actual engineering applications.

[0106] In the interference identification device provided in the embodiments of the present application, the acquisition module is further configured to acquire a plurality of internal interference samples and first external interference samples from an interference troubleshooting case library, and determine historical interfering cells according to the internal interference samples and the first external interference samples, wherein the number of the internal interference samples is less than that of the first external interference samples; generate a plurality of second external interference samples of different interference types by using a simulation software according to preset interference conditions, and determine simulation interfering cells according to the second external interference samples; and aggregate the historical interfering cells, the simulation interfering cells, and the to-be-analyzed interfering cell to obtain the set of interfering cells.

[0107] In the interference identification device provided in the embodiments of the present application, the determining module is further configured to determine physical positions of each interfering cell in the set of interfering cells, wherein the physical position is the latitude and longitude information of the interfering cell; determine interference relationships between adjacent interfering cells in the set of interfering cells, wherein the interference relationship is used to reflect the interference correlation degree between the adjacent interfering cells; and determine the graph structure according to the physical positions and the interference relationships.

[0108] In the interference identification device provided in the embodiments of the present application, the determining module is further configured to determine distances between the adjacent interfering cells, and determine that there is a connection relationship between the adjacent interfering cells when the distance is less than a preset distance threshold; determine interference similarities between the adjacent interfering cells, and determine that there is a connection relationship between the adjacent interfering cells when the interference similarity is greater than a preset similarity threshold.

[0109] In the interference identification device provided in this application embodiment, the determining module is further configured to process the time-domain index data of the target interfering cell through a time-series analysis model to obtain time-domain features, wherein the target interfering cell is any interfering cell in the set of interfering cells; determine the frequency-domain features and channel features corresponding to the frequency-domain index data and channel index data of the target interfering cell respectively, and perform standardization processing on the frequency-domain features and channel features; and splice the time-domain features, frequency-domain features and channel features to obtain the node features of the target interfering cell.

[0110] In the interference identification device provided in this application embodiment, the identification module is further configured to generate a node embedding representation corresponding to the target interfering cell through an interference prediction model, wherein the node embedding representation is the fusion result of the node features of the target interfering cell and the node features of the neighboring interfering cells of the target interfering cell; determine the connection relationship in the graph structure, and based on the connection relationship and the node embedding representation, propagate the data information in the known interference label to the interfering cell to be analyzed, thereby obtaining the interference identification result of the interfering cell to be analyzed.

[0111] In the interference identification device provided in this application embodiment, the identification module is further used to determine the interference type and predicted probability corresponding to the interference cell to be analyzed during the propagation of data information in the known interference tag; when the predicted probability is greater than a preset threshold, the corresponding interference type is determined as the interference identification result of the interference cell to be analyzed.

[0112] In the interference identification device provided in this application embodiment, the identification module is further configured to generate a visualization report of the interfering cell to be analyzed based on the interference identification result. The visualization report includes at least one of the following: an interference time-domain analysis diagram, an interference frequency-domain analysis diagram, and an interference range analysis diagram of the interfering cell to be analyzed; and to generate corresponding interference investigation suggestions for different interference types in the interference identification result.

[0113] This application also provides an electronic device, including: a memory and a processor, wherein the memory is used to store program instructions; and the processor is connected to the memory and used to execute the above-described interference identification method.

[0114] It should be noted that the aforementioned electronic equipment is used to perform Figure 2 The interference identification method shown above is also applicable to this electronic device, and will not be repeated here.

[0115] This application also provides a non-volatile storage medium including a stored computer program, wherein the device containing the non-volatile storage medium executes the above-described interference identification method by running the computer program.

[0116] It should be noted that the non-volatile storage medium is used to execute Figure 2 The interference identification method shown in the above formula, so the relevant explanation of the interference identification method is also applicable to the non-volatile storage medium, and will not be repeated here.

[0117] The embodiment of the application also provides a computer program product, comprising computer instructions, which are executed by a processor to realize the interference identification method.

[0118] It should be noted that the computer program product is used to execute Figure 2 The interference identification method shown in the above formula, so the relevant explanation of the interference identification method is also applicable to the computer program product, and will not be repeated here.

[0119] The serial number of the embodiment of the application is only for description, and does not represent the advantages and disadvantages of the embodiment.

[0120] In the above embodiment of the application, the description of each embodiment has its own emphasis, and the part not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0121] In the several embodiments provided by the present application, it should be understood that the disclosed technology can be implemented in other ways. Of course, the unit described as the division is only a description of logical function division, and there can be another division manner in actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, unit or module, and can be electrical or other forms.

[0122] The unit described as a separate component can or can not be physically separated, and the component shown as a unit can or can not be a physical unit, that is, it can be located in one place, or can be distributed on multiple units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0123] In addition, each functional unit in each embodiment of the application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of software functional unit.

[0124] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or say the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a mobile hard disk, a magnetic disk or an optical disk, and various media that can store program codes.

[0125] The above is only the preferred embodiment of the present application, and it should be pointed out that for those skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, which should be considered as the protection scope of the present application.

Claims

1. A method of interference identification, characterized by, The method comprises: obtaining a set of interference cells, wherein the set of interference cells comprises historical interference cells, simulated interference cells and to-be-analyzed interference cells, each interference cell comprises corresponding time domain index data, frequency domain index data and channel index data; determining a graph structure corresponding to the set of interference cells, and determining node features in the graph structure, wherein each interference cell is a node in the graph structure, and the node features are index features after fusing the time domain index data, the frequency domain index data and the channel index data; training an interference prediction model according to the graph structure, the node features and known interference labels, and predicting the to-be-analyzed interference cells by using the trained interference prediction model to obtain interference identification results.

2. The method of claim 1, wherein, Obtaining a set of interference cells comprises: obtaining a plurality of internal interference samples and first external interference samples from an interference investigation case library, and determining the historical interference cells according to the internal interference samples and the first external interference samples, wherein the number of internal interference samples is less than that of the first external interference samples; generating a plurality of second external interference samples of different interference types by using simulation software according to preset interference conditions, and determining the simulated interference cells according to the second external interference samples; aggregating the historical interference cells, the simulated interference cells and the to-be-analyzed interference cells to obtain the set of interference cells.

3. The method of claim 1, wherein, Determining a graph structure corresponding to the set of interference cells comprises: determining the physical location of each interference cell in the set of interference cells, wherein the physical location is the latitude and longitude information of the interference cell; determining the interference relationship between adjacent interference cells in the set of interference cells, wherein the interference relationship is used to reflect the interference correlation degree between the adjacent interference cells; determining the graph structure according to the physical location and the interference relationship.

4. The method of claim 3, wherein, The method further comprises: determining the distance between the adjacent interference cells, and determining that there is a connection relationship between the adjacent interference cells if the distance is less than a preset distance threshold; determining the interference similarity between the adjacent interference cells, and determining that there is a connection relationship between the adjacent interference cells if the interference similarity is greater than a preset similarity threshold.

5. The method of claim 1, wherein, Determining the node features in the graph structure comprises: processing the time domain index data of a target interference cell by using a time series analysis model to obtain time domain features, wherein the target interference cell is any one of the interference cells in the set of interference cells; respectively determining frequency domain features and channel features corresponding to the frequency domain index data and the channel index data of the target interference cell, and performing standardization processing on the frequency domain features and the channel features; splicing the time domain features, the frequency domain features and the channel features to obtain the node features of the target interference cell.

6. The method of claim 5, wherein, Predicting the to-be-analyzed interference cells by using the trained interference prediction model to obtain interference identification results comprises: generating, by the interference prediction model, a node embedding representation corresponding to the target interfering cell, where the node embedding representation is a fusion result of a node feature of the target interfering cell and node features of neighbor interfering cells of the target interfering cell; determining a connection relationship in the graph structure, and propagating data information in the known interference label to the to-be-analyzed interfering cell according to the connection relationship and the node embedding representation, to obtain an interference identification result of the to-be-analyzed interfering cell.

7. The method of claim 6, wherein, The method further includes: determining an interference type and a prediction probability corresponding to the to-be-analyzed interfering cell during the propagation of the data information in the known interference label; in a case where the prediction probability is greater than a preset threshold, determining the corresponding interference type as the interference identification result of the to-be-analyzed interfering cell.

8. The method of claim 1, wherein, The method further includes: generating a visual report of the to-be-analyzed interfering cell according to the interference identification result, where the visual report includes at least one of a time domain analysis graph, a frequency domain analysis graph, and a range analysis graph of interference of the to-be-analyzed interfering cell; and generating a corresponding interference troubleshooting suggestion for different interference types in the interference identification result.

9. An interference identification apparatus characterized by comprising: includes: an acquisition module, configured to acquire a set of interfering cells, where the set of interfering cells includes historical interfering cells, simulated interfering cells, and to-be-analyzed interfering cells, and each interfering cell includes corresponding time domain index data, frequency domain index data, and channel index data; a determination module, configured to determine a graph structure corresponding to the set of interfering cells, and determine node features in the graph structure, where each interfering cell is a node in the graph structure, and the node features are index features after fusing the time domain index data, the frequency domain index data, and the channel index data; an identification module, configured to train an interference prediction model according to the graph structure, the node features, and a known interference label, and perform prediction on the to-be-analyzed interfering cell by using the trained interference prediction model, to obtain an interference identification result.

10. An electronic device, comprising: includes: a memory and a processor, where the memory is configured to store program instructions; the processor, connected with the memory, is configured to execute an interference identification method according to any one of claims 1 to 8.

11. A non-volatile storage medium, comprising: The non-volatile storage medium includes a stored computer program, where a device where the non-volatile storage medium is located executes the interference identification method according to any one of claims 1 to 8 by running the computer program.

12. A computer program product comprising computer instructions, characterized in that, The computer instructions are executed by the processor to implement the interference identification method according to any one of claims 1 to 8.

Images