As networks evolve and alarm and performance data volumes increase, it is quickly becoming impractical for providers to manually monitor and process network alarm and performance data. Without a solution, network and service performance will degrade and customer experience will be impacted.
Artificial Intelligence and Machine Learning can automate the analysis of this data to reduce alarm noise and predict future failures and performance issues. Many solutions, however, require advanced and ongoing configuration of these modules when what is needed is a hands-on solution that analyzes network data on its own, reduces noise and predicts failures and learns over time.
Centina vSure® is pioneering AI and Machine Learning for Service Assurance. AI and Machine Learning in vSure focuses on:
Learning what’s normal and identifying abnormalities
With thousands or millions of datapoints being processed per second, its beyond human ability to scan through all of the data to find the few things which require attention. vSure uses machine learning to study this data and to detect when abnormal or anomalous events occur. vSure can automatically adapt depending upon the type of data being collected.
Anomaly detection of fault data
vSure uses advanced, unsupervised machine learning algorithms to learn which patterns of alarms normally appear. Armed with this knowledge, vSure is able to accurately cut noise and accurately pinpoint the problems that require action to resolve.
This drastically reduces the number of alarms – by as much as 90%.
And the best part – This works without any configuration and gets better over time as vSure learns about your network.
Anomaly detection of performance data
vSure statistically models all performance data collected to learn what’s normal for your network over time. With this modeling, vSure can detect when values deviate from normal and can generate alarms and notifications. In addition, when the alarm is generated, vSure can invoke Workflow Automation to remediate a problem or optimize a performance issue.
Learning data trends, including seasonality, and predicting future performance
vSure performance models are also used to understand seasonal trends of performance and to predict future performance and failures. vSure provides dynamic and interactive views of not only past but also future performance.
Predictive alarm thresholding turns future predicted performance into actionable events that can generate notifications and automations. For example, predicting that utilization will hit 100% capacity in 7 days so that proactive notifications and actions can be taken before the issue occurs.