CN-122002353-A - Data acquisition system of wireless communication network

CN122002353ACN 122002353 ACN122002353 ACN 122002353ACN-122002353-A

Abstract

The invention provides a data acquisition system of a wireless communication network, which relates to the technical field of data acquisition and comprises a first edge process and a wireless protocol stack process which are deployed on the same base station server, wherein the first edge process is used for analyzing KPI data streams formed by one or more KPI data generated by the wireless protocol stack process in real time, generating a first data acquisition strategy associated with network abnormality under the condition that the network abnormality occurs in the wireless communication network based on analysis judgment, and sending the first data acquisition strategy to the wireless protocol stack process, and the wireless protocol stack process is used for acquiring original wireless communication data for deep diagnosis of the network abnormality according to the first data acquisition strategy. The invention realizes effective balance between acquisition depth and acquisition cost.

Inventors

HUANG YONGMING
ZHANG BINGZHI
HE SHIWEN
YI YUNSHAN

Assignees

紫金山实验室

Dates

Publication Date: 20260508
Application Date: 20260122

Claims (10)

1. The data acquisition system of the wireless communication network is characterized by comprising a first edge process and a wireless protocol stack process which are deployed on the same base station server; The first edge process is used for analyzing KPI data flow formed by one or more KPIs generated by the wireless protocol stack process in real time, generating a first data acquisition strategy associated with network abnormality when the network abnormality occurs in the wireless communication network based on the analysis, and transmitting the first data acquisition strategy to the wireless protocol stack process; The wireless protocol stack process is used for collecting original wireless communication data for deeply diagnosing the network abnormality according to the first data collection strategy.
2. The data acquisition system of a wireless communication network according to claim 1, wherein the first edge process is configured to, in real time, analyze KPI data streams formed by one or more KPI data generated by the wireless protocol stack process, in particular: The KPI data streams are ordered according to time to form an input KPI vector, and the input KPI vector is reconstructed through a built-in anomaly detection model to obtain a reconstructed KPI vector; Calculating a reconstruction error based on a difference between the input KPI vector and the reconstructed KPI vector; And when the reconstruction error exceeds an error threshold, judging that the wireless communication network is abnormal.
3. The data acquisition system of a wireless communication network according to claim 2, wherein the first edge process, when generating a first data acquisition policy associated with the network anomaly, is specifically configured to: decomposing the reconstruction error, determining one or more KPIs with the largest contribution to the network anomaly, and combining the one or more KPIs with the largest contribution as key KPIs; Searching in a rule base based on the key KPI combination to obtain the first data acquisition strategy, wherein the rule base stores a mapping relation between the key KPI combination and the first data acquisition strategy.
4. The data acquisition system of the wireless communication network of claim 1, wherein the first edge process is further configured to: Receiving a service acquisition demand, translating the service acquisition demand into a second data acquisition strategy based on a built-in knowledge graph, and sending the second data acquisition strategy to the wireless protocol stack process; The knowledge graph is obtained by carrying out structural representation on a plurality of entities in the wireless communication network and interrelation among the entities, wherein the entities comprise a business target entity, a network index entity and a data field entity.
5. The data acquisition system of claim 4, wherein the first edge process is configured to, when translating the traffic acquisition requirement into a second data acquisition policy based on a built-in knowledge-graph: Mapping the service acquisition requirement to the service target entity in the knowledge graph; Starting from the business target entity, executing a weighted graph traversing algorithm on the knowledge graph to identify the network index entity associated with the business target entity; Starting from the network index entity, determining a data field entity associated with the network index entity along a path determined by the weighted graph traversal algorithm to form a target data field set; The second data acquisition strategy is generated based on the set of target data fields.
6. The data acquisition system of the wireless communication network of claim 4, wherein the wireless protocol stack process comprises a policy management module and one or more data acquisition anchor functions deployed at key data generation locations in a code execution path of the wireless protocol stack process; The strategy management module is used for receiving and caching the data acquisition strategy sent by the first edge process, wherein the data acquisition strategy comprises the first data acquisition strategy and the second data acquisition strategy; the data collection anchor function, when invoked, is to: writing predefined one or more KPI data into IPC message queues, and Inquiring the cached data acquisition strategy, acquiring wireless communication data according to the inquired data acquisition strategy, and copying the wireless communication data to a memory buffer area shared with the first edge process.
7. The data acquisition system of the wireless communication network of claim 6, wherein the first edge process is further configured to: acquiring the wireless communication data from the memory buffer area; in the case that the wireless communication data is a KPI data stream, analyzing the wireless communication data in real time; In the case where the wireless communication data is original wireless communication data, the wireless communication data is written into a local database or transmitted to a remote storage engine according to the data type and the data size of the wireless communication data.
8. The data acquisition system of claim 2, wherein the base station server is further configured with a second edge process for: and updating the abnormality detection model to generate a new version of abnormality detection model.
9. The data acquisition system of the wireless communication network of claim 8, wherein the first edge process is further configured to: carrying out statistical test on the KPI data flow to obtain statistical distribution of the KPI data flow; And under the condition that the concept drift is determined to occur based on the statistical distribution of the KPI data flow, sending a drift trigger signal to a second edge process, wherein the drift trigger signal is used for triggering the updating of the anomaly detection model, and the drift trigger signal comprises at least one part of the KPI data flow.
10. The data collection system of the wireless communication network according to claim 9, wherein the second edge process is specifically configured to, when updating the anomaly detection model to generate a new version of the anomaly detection model: Detecting the current load of the base station server after receiving the drift trigger signal; Loading a copy of the anomaly detection model currently used by the first edge process under the condition that the current load of the base station server is lower than a safety threshold; and based on at least one part of the KPI data flow contained in the drift trigger signal, adopting an elastic weight consolidation algorithm to carry out incremental update on the copy to obtain the new version of abnormality detection model.

Description

Data acquisition system of wireless communication network Technical Field The invention relates to the technical field of data acquisition, in particular to a data acquisition system of a wireless communication network. Background In 5G and future wireless communication networks, efficient data acquisition of base stations is the basis for realizing intelligent operation and maintenance, fault diagnosis and performance optimization of the network. Currently, the mainstream technology generally adopts a mode of reporting before analyzing, namely, through a centralized platform such as a Network management system (Network MANAGEMENT SYSTEM, NMS) or a service management and orchestrator (SERVICE MANAGEMENT AND Orchestration, SMO), an operation and maintenance person performs static configuration on a base station and subscribes to aggregated key performance index (Key Performance Indicator, KPI) data. The base station accordingly continuously and indifferently reports the data streams to support traditional network monitoring and offline analysis. However, this model faces a fundamental contradiction in that, on the one hand, accurate network diagnostics and research must acquire depth data of the physical layer, medium access control layer at the slot level, and on the other hand, full time acquisition of such depth data can result in unacceptable costs in transmission, storage and computation. Therefore, the prior art cannot achieve effective balance between the acquisition depth and the acquisition cost, which constitutes a technical problem to be solved in the current wireless communication network data acquisition field. Disclosure of Invention The invention provides a data acquisition system of a wireless communication network, which is used for solving the defect that effective balance between acquisition depth and acquisition cost cannot be achieved in the prior art and realizing the effective balance between the acquisition depth and the acquisition cost. The invention provides a data acquisition system of a wireless communication network, which comprises a first edge process and a wireless protocol stack process which are deployed on the same base station server; the first edge process is used for analyzing KPI data flow formed by one or more KPIs generated by the wireless protocol stack process in real time, generating a first data acquisition strategy associated with network abnormality under the condition that the network abnormality occurs in the wireless communication network based on the analysis, and transmitting the first data acquisition strategy to the wireless protocol stack process; The wireless protocol stack process is used for collecting original wireless communication data for deeply diagnosing the network abnormality according to the first data collection strategy. In some embodiments, the first edge process is specifically configured to, when analyzing in real time a KPI data stream composed of one or more KPI data generated by the wireless protocol stack process: The KPI data streams are ordered according to time to form an input KPI vector, and the input KPI vector is reconstructed through a built-in anomaly detection model to obtain a reconstructed KPI vector; Calculating a reconstruction error based on a difference between the input KPI vector and the reconstructed KPI vector; And when the reconstruction error exceeds an error threshold, judging that the wireless communication network is abnormal. In some embodiments, the first edge process, when generating a first data collection policy associated with the network anomaly, is specifically configured to: decomposing the reconstruction error, determining one or more KPIs with the largest contribution to the network anomaly, and combining the one or more KPIs with the largest contribution as key KPIs; Searching in a rule base based on the key KPI combination to obtain the first data acquisition strategy, wherein the rule base stores a mapping relation between the key KPI combination and the first data acquisition strategy. In some embodiments, the first edge process is further to: Receiving a service acquisition demand, translating the service acquisition demand into a second data acquisition strategy based on a built-in knowledge graph, and sending the second data acquisition strategy to the wireless protocol stack process; The knowledge graph is obtained by carrying out structural representation on a plurality of entities in the wireless communication network and interrelation among the entities, wherein the entities comprise a business target entity, a network index entity and a data field entity. In some embodiments, the first edge process is specifically configured to, when translating the service acquisition requirement into the second data acquisition policy based on the built-in knowledge-graph: Mapping the service acquisition requirement to the service target entity in the knowledge graph; Starting from the business targe