High-latency anomaly detection method and system based on swarm intelligence network measurement data

Abstract

The invention provides a high-delay abnormity detection method and system based on measurement data of the swarm intelligence network. The method comprises that characteristic engineering is carried out on an original data set, abnormal and missing values are removed, and numeralization and discretization are carried out on original data in a unified way; original data with the same characteristicis clustered into one instance, a corresponding weight is calculated, and the instance is labeled, and serves as a basic unit input to a decision tree model; the preprocessed original data set is sampled randomly to generate sub data sets, and each sub data set is modeled by a CART decision tree to obtain a sub tree; high-delay abnormity rules are excavated based on topology and node informationof each sub tree; each rule is calibrated and scored from the aspect of confidence based on network delay condition of a sample subspace of the tree nodes; and rule excavation results of all the sub trees are merged to generate a final network high-delay abnormity detection result. The generalization and robustness of an algorithm are improved, and the high-delay network abnormity can be detectedeffectively.

Description

technical field [0001] The present invention relates to the field of communication technology, in particular to a high-delay anomaly detection method and system based on swarm intelligence network measurement data. Background technique [0002] Nowadays, due to the wide coverage of mobile Internet, Over The Top (OTT) services are developing rapidly. In OTT, users can provide various services, such as video streaming and text transmission, through traditional network operating systems. Unlike traditional communication services, OTT only utilizes the operator's network, while the service is provided by Internet companies. For example, with the help of broadband network services leased from network operators, Skype can provide global IP (VOIP) services at a relatively low cost. In order for OTT services to be available globally, the coverage of the underlying network must be large enough to reach as many users as possible. Therefore, an integrated network of many Internet Se...

AI Technical Summary

Problems solved by technology

However, for such large-scale crowd intelligence data containing a lot of noise, traditional mathematical statistical methods lack robustness, and it is difficult to find the correlation between data

Images