Outlier detection method and system in lte network

Abstract

The present invention provides an outlier detection method and system in an LTE network. By dividing the measured data into a training set and a test set, defining clusters and parameters in the training set, a clustering algorithm is used to find the cluster to which each data point belongs, and according to the parameters Value and clustering results, calculate the likelihood value of each data point, divide the likelihood value into abnormal area, intermediate area and normal area according to the set early warning threshold and alarm threshold; apply the calculated model to the test set , the likelihood of each data point is calculated and grouped into regions to find outliers in the test set. In the method and system, adding a time axis to the model can better understand the change of data points in time, and then can discover multiple outliers from a sequence composed of multiple points instead of a single outlier. This method can quickly detect outliers, and can be found in advance after an outlier appears, and the error rate is very low.

Description

technical field [0001] The invention relates to an abnormal value detection method and system in an LTE network. Background technique [0002] Over the past few years, the amount of data generated over telecommunication networks has grown exponentially. Anomaly detection, which aims to find anomalies in unexpected data patterns by artificial patterns becomes difficult. Due to the huge amount of data, even business experts cannot find anomalies by browsing log files. [0003] The rapid development of LTE networks produces more and more network traffic data. Therefore, it is not possible to manually process and analyze the resulting data traffic. In particular, automatic detection of outliers from continuous data streams remains one of the remaining challenges. This area is critical because anomalies can lead to network inefficiencies. Indeed, the origin of these anomalies may be a technical problem in a cell or a fraudulent intrusion in network usage, which need to be id...

AI Technical Summary

Problems solved by technology

[0009] The purpose of the present invention is to provide an outlier detection method and system in an LTE network, which takes the time axis as an important factor, thereby being able to detect outliers in a new mode and quickly and accurately find outliers under unexpected conditions, so as to solve the problem of Problems existing in the existing technology may not be able to detect unexpected outliers, or cannot discover multiple outliers from a sequence composed of multiple data points

Images