The complexity of alarm detection and diagnosis tasks often results in a lack of alarm log data.Due to the strong rule associations inherent in alarm log data,existing data augmentation algorithms cannot obtain good r...The complexity of alarm detection and diagnosis tasks often results in a lack of alarm log data.Due to the strong rule associations inherent in alarm log data,existing data augmentation algorithms cannot obtain good results for alarm log data.To address this problem,this paper introduces a new algorithm for augmenting alarm log data,termed APRGAN,which combines a generative adversarial network(GAN)with the Apriori algorithm.APRGAN generates alarm log data under the guidance of rules mined by the rule miner.Moreover,we propose a new dynamic updating mechanism to alleviate the mode collapse problem of the GAN.In addition to updating the real reference dataset used to train the discriminator in the GAN,we dynamically update the parameters and the rule set of the Apriori algorithm according to the data generated in each epoch.Through extensive experimentation on two public datasets,it is demonstrated that APRGAN surpasses other data augmentation algorithms in the domain with respect to alarm log data augmentation,as evidenced by its superior performance on metrics such as BLEU,ROUGE,and METEOR.展开更多
The evolution of telecommunications has allowed the development of broadband services based mainly on fiber optic backbone networks. The operation and maintenance of these optical networks is made possible by using su...The evolution of telecommunications has allowed the development of broadband services based mainly on fiber optic backbone networks. The operation and maintenance of these optical networks is made possible by using supervision platforms that generate alarms that can be archived in the form of log files. But analyzing the alarms in the log files is a laborious and difficult task for the engineers who need a degree of expertise. Identifying failures and their root cause can be time consuming and impact the quality of service, network availability and service level agreements signed between the operator and its customers. Therefore, it is more than important to study the different possibilities of alarms classification and to use machine learning algorithms for alarms correlation in order to quickly determine the root causes of problems faster. We conducted a research case study on one of the operators in Cameroon who held an optical backbone based on SDH and WDM technologies with data collected from 2016-03-28 to “2022-09-01” with 7201 rows and 18. In this paper, we will classify alarms according to different criteria and use 02 unsupervised learning algorithms namely the K-Means algorithm and the DBSCAN to establish correlations between alarms in order to identify root causes of problems and reduce the time to troubleshoot. To achieve this objective, log files were exploited in order to obtain the root causes of the alarms, and then K-Means algorithm and the DBSCAN were used firstly to evaluate their performance and their capability to identify the root cause of alarms in optical network.展开更多
以某电信公司的网络设备告警日志作为研究基础,提出随机森林嵌入(random trees embedding,RTE)和极端梯度提升(extreme gradient boosting,XGBoost)的组合模型,并将其应用于对网络业务运行状态的故障预测。针对特征中噪声大的问题,采用...以某电信公司的网络设备告警日志作为研究基础,提出随机森林嵌入(random trees embedding,RTE)和极端梯度提升(extreme gradient boosting,XGBoost)的组合模型,并将其应用于对网络业务运行状态的故障预测。针对特征中噪声大的问题,采用RTE算法对告警日志数据进行特征转化,借助Bagging抽样方法减少噪声数据的影响;结合XGBoost算法的多分类功能建立预测模型。实验结果表明,相对于XGBoost算法、随机森林算法以及基于概率统计的贝叶斯算法,RTE+XGBoost组合模型的预测准确率分别提升了1.23%、3.44%、4.43%。展开更多
基金supported by the National Key Research and Development Program of China under Grant No.2019YFB-2103202.
文摘The complexity of alarm detection and diagnosis tasks often results in a lack of alarm log data.Due to the strong rule associations inherent in alarm log data,existing data augmentation algorithms cannot obtain good results for alarm log data.To address this problem,this paper introduces a new algorithm for augmenting alarm log data,termed APRGAN,which combines a generative adversarial network(GAN)with the Apriori algorithm.APRGAN generates alarm log data under the guidance of rules mined by the rule miner.Moreover,we propose a new dynamic updating mechanism to alleviate the mode collapse problem of the GAN.In addition to updating the real reference dataset used to train the discriminator in the GAN,we dynamically update the parameters and the rule set of the Apriori algorithm according to the data generated in each epoch.Through extensive experimentation on two public datasets,it is demonstrated that APRGAN surpasses other data augmentation algorithms in the domain with respect to alarm log data augmentation,as evidenced by its superior performance on metrics such as BLEU,ROUGE,and METEOR.
文摘The evolution of telecommunications has allowed the development of broadband services based mainly on fiber optic backbone networks. The operation and maintenance of these optical networks is made possible by using supervision platforms that generate alarms that can be archived in the form of log files. But analyzing the alarms in the log files is a laborious and difficult task for the engineers who need a degree of expertise. Identifying failures and their root cause can be time consuming and impact the quality of service, network availability and service level agreements signed between the operator and its customers. Therefore, it is more than important to study the different possibilities of alarms classification and to use machine learning algorithms for alarms correlation in order to quickly determine the root causes of problems faster. We conducted a research case study on one of the operators in Cameroon who held an optical backbone based on SDH and WDM technologies with data collected from 2016-03-28 to “2022-09-01” with 7201 rows and 18. In this paper, we will classify alarms according to different criteria and use 02 unsupervised learning algorithms namely the K-Means algorithm and the DBSCAN to establish correlations between alarms in order to identify root causes of problems and reduce the time to troubleshoot. To achieve this objective, log files were exploited in order to obtain the root causes of the alarms, and then K-Means algorithm and the DBSCAN were used firstly to evaluate their performance and their capability to identify the root cause of alarms in optical network.
文摘以某电信公司的网络设备告警日志作为研究基础,提出随机森林嵌入(random trees embedding,RTE)和极端梯度提升(extreme gradient boosting,XGBoost)的组合模型,并将其应用于对网络业务运行状态的故障预测。针对特征中噪声大的问题,采用RTE算法对告警日志数据进行特征转化,借助Bagging抽样方法减少噪声数据的影响;结合XGBoost算法的多分类功能建立预测模型。实验结果表明,相对于XGBoost算法、随机森林算法以及基于概率统计的贝叶斯算法,RTE+XGBoost组合模型的预测准确率分别提升了1.23%、3.44%、4.43%。
