Real-time anomaly detection for LTE and 5G signalling networks.
Analyse live signalling streams to identify anomalous behaviour using a general AI model. Deploy entirely on-premises, receive confidence scores in real time, and investigate results through an integrated browser-based dashboard.
Mobile networks generate vast quantities of signalling data every second. Hidden within that data can be behavioural patterns that differ significantly from normal subscriber activity.
The Melrose Networks Anomaly Engine continuously analyses signalling events, enriches them with operator-supplied network information and produces anomaly confidence scores that highlight behaviour worthy of investigation.
The engine does not attempt to classify devices or make operational decisions. Instead, it provides network engineers with a fast and objective indication of anomalous signalling behaviour.
The engine accepts signalling events from replay files or live data feeds.
Events are combined with operator-supplied reference information such as cell locations, site topology and network configuration.
A general machine learning model analyses mobility and signalling behaviour over time, producing an anomaly confidence score for each observed device.
Engineers use the integrated dashboard to review anomalous behaviour, examine contributing observations and determine whether further investigation is required.
The supplied model is trained for broad applicability across mobile networks.
It is not tuned to the characteristics of any specific operator network.
As a result, performance may differ between networks and may be improved through operator-specific model tuning using representative signalling data.
This approach allows operators to begin analysing their own network immediately while providing a straightforward path to higher detection performance where required.
The Anomaly Engine has been designed specifically for mobile network operators.
Typical applications include:
Deployment is intentionally straightforward.
The software runs entirely within the operator's own environment.
No cloud infrastructure is required.
No operator data leaves the network.
No telemetry is collected.
The integrated dashboard provides immediate visibility of network activity, including:
The engine exposes a documented REST API, allowing anomaly scores and supporting information to be integrated into existing operational tools, SOC platforms and analytics systems.
The software package includes:
The Anomaly Engine is supplied for self-service deployment.
Documentation is included with the software.
Melrose Networks can also provide commercial services including:
The Anomaly Engine identifies behaviour that differs from the baseline learned by the supplied model.
An anomaly score is not a classification and should not be interpreted as confirmation of any specific activity or device type.
Operational decisions should always be made using appropriate engineering judgement and, where necessary, additional sources of information.
The Melrose Networks Anomaly Engine is available at no charge for self-service use by mobile network operators.
The software can only be supplied to staff of a mobile network operator or to contractors working for mobile network operators. In all instances, the request to supply the software must be made by the mobile network operator.
To request a copy or discuss operator-specific model tuning and commercial support services, please contact sales@melrosenetworks.com.
Supports LTE EPC (S1AP)
Supports 5G NSA networks achored on LTE EPC (S1AP)
Supports the latest 5G Standalone networks (NGAP)
Request information.
Global 24x7.
