Mission-Critical Communication Networks (MCCNs) are wireless networks whose malfunction can cause significant problems. The nature of MCCNs puts an extremely high standard on the Quality of Service (QoS). QoS assurance starts from monitoring and change/anomaly detection of network packets data. This problem has been primarily studied by the research community of communication networks, in which the existing methods fall short for providing a privacy-preserving, minimum-disruption, global monitoring tool. Another relevant research area is Multivariate Statistical Process Control (MSPC), in which generic methods have been developed for monitoring high-dimensional data streams. These methods do not account for the special data distribution and correlation structure of packet streams. Nor are they efficient enough to suit real-time analytics in MCCNs. We propose a method called Sparse Singular Value Decomposition (SSVD)-MSPC. SSVD-MSPC addresses the aforementioned limitations and additionally provides key capabilities toward QoS improvement, including monitoring, fault identification, and fault characterization. Extensive case studies are conducted for MCCNs that experience faults of different magnitudes and various temporal shapes. SSVD-MSPC achieves good performance across the different settings in comparison with existing methods.
|Original language||English (US)|
|Number of pages||13|
|State||Published - Dec 2 2018|
- Communication networks
- quality of service
- sparse learning
- traffic monitoring
ASJC Scopus subject areas
- Industrial and Manufacturing Engineering
FingerprintDive into the research topics of 'Integration of sparse singular vector decomposition and statistical process control for traffic monitoring and quality of service improvement in mission-critical communication networks'. Together they form a unique fingerprint.
Integration of sparse singular vector decomposition and statistical process control for traffic monitoring and quality of service improvement in mission-critical communication networks
Li, J. (Contributor) & Wang, K. (Contributor), figshare Academic Research System, Dec 2 2018
DOI: 10.6084/m9.figshare.6738593.v2, https://doi.org/10.6084%2Fm9.figshare.6738593.v2