With rising capacity demand in mobile networks, the infrastructure is also becoming increasingly denser and complex. This results in collection of larger amount of raw data(big data) that is generated at different lev...With rising capacity demand in mobile networks, the infrastructure is also becoming increasingly denser and complex. This results in collection of larger amount of raw data(big data) that is generated at different levels of network architecture and is typically underutilized. To unleash its full value, innovative machine learning algorithms need to be utilized in order to extract valuable insights which can be used for improving the overall network's performance. Additionally, a major challenge for network operators is to cope up with increasing number of complete(or partial) cell outages and to simultaneously reduce operational expenditure. This paper contributes towards the aforementioned problems by exploiting big data generated from the core network of 4 G LTE-A to detect network's anomalous behavior. We present a semi-supervised statistical-based anomaly detection technique to identify in time: first, unusually low user activity region depicting sleeping cell, which is a special case of cell outage; and second, unusually high user traffic area corresponding to a situation where special action such as additional resource allocation, fault avoidance solution etc. may be needed. Achieved results demonstrate that the proposed method can be used for timely and reliable anomaly detection in current and future cellular networks.展开更多
Nowadays,the fifth-generation(5G)mobile communication system has obtained prosperous development and deployment,reshaping our daily lives.However,anomalies of cell outages and congestion in 5G critically influence the...Nowadays,the fifth-generation(5G)mobile communication system has obtained prosperous development and deployment,reshaping our daily lives.However,anomalies of cell outages and congestion in 5G critically influence the quality of experience and significantly increase operational expenditures.Although several big data and artificial intelligencebased anomaly detection methods have been proposed for wireless cellular systems,they change distributions of the data and ignore the relevance among user activities,causing anomaly detection ineffective for some cells.In this paper,we propose a highly effective and accurate anomaly detection framework by utilizing generative adversarial networks(GAN)and long short-term memory(LSTM)neural networks.The framework expands the original dataset while simultaneously keeping the distribution of data unchanged,and explores the relevance among user activities to further improve the system performance.The results demonstrate that our framework can achieve 97.16%accuracy and 2.30%false positive rate by utilizing the correlation of user activities and data expansion.展开更多
基金supported in part by the National Natural Science Foundation of China under the Grants No.61431011 and 61671371the National Science and Technology Major Project under Grant no.2016ZX03001016-005+1 种基金the Key Research and Development Program of Shaanxi Province under Grant No.2017ZDXM-G-Y-012the Fundamental Research Funds for the Central Universities
文摘With rising capacity demand in mobile networks, the infrastructure is also becoming increasingly denser and complex. This results in collection of larger amount of raw data(big data) that is generated at different levels of network architecture and is typically underutilized. To unleash its full value, innovative machine learning algorithms need to be utilized in order to extract valuable insights which can be used for improving the overall network's performance. Additionally, a major challenge for network operators is to cope up with increasing number of complete(or partial) cell outages and to simultaneously reduce operational expenditure. This paper contributes towards the aforementioned problems by exploiting big data generated from the core network of 4 G LTE-A to detect network's anomalous behavior. We present a semi-supervised statistical-based anomaly detection technique to identify in time: first, unusually low user activity region depicting sleeping cell, which is a special case of cell outage; and second, unusually high user traffic area corresponding to a situation where special action such as additional resource allocation, fault avoidance solution etc. may be needed. Achieved results demonstrate that the proposed method can be used for timely and reliable anomaly detection in current and future cellular networks.
基金supported by National Natural Science Foundation of China under Grant 61772406 and Grant 61941105in part by the projects of the Fundamental Research Funds for the Central Universitiesthe Innovation Fund of Xidian University under Grant 500120109215456。
文摘Nowadays,the fifth-generation(5G)mobile communication system has obtained prosperous development and deployment,reshaping our daily lives.However,anomalies of cell outages and congestion in 5G critically influence the quality of experience and significantly increase operational expenditures.Although several big data and artificial intelligencebased anomaly detection methods have been proposed for wireless cellular systems,they change distributions of the data and ignore the relevance among user activities,causing anomaly detection ineffective for some cells.In this paper,we propose a highly effective and accurate anomaly detection framework by utilizing generative adversarial networks(GAN)and long short-term memory(LSTM)neural networks.The framework expands the original dataset while simultaneously keeping the distribution of data unchanged,and explores the relevance among user activities to further improve the system performance.The results demonstrate that our framework can achieve 97.16%accuracy and 2.30%false positive rate by utilizing the correlation of user activities and data expansion.
