Sentinel, a research programme

Goals

When a call route degrades or suspicious traffic appears, the usual responses rely on fixed thresholds and on backup routes defined in advance. The programme studies whether finer methods can do better, while staying within each carrier’s rules.

• Spot a traffic or quality anomaly earlier.
• Choose a suitable response among the options actually permitted.
• Measure what each response adds compared with current methods.

The three hurdles

1. Detect with little data. New traffic or a new country has no history. An anomaly has to be recognized without months of measurements.
2. Decide under constraints. A routing decision has to respect number authentication and the blocks set by carriers. It never works around them.
3. Prove progress. A result is only worth something if it beats simple baselines: fixed thresholds and static backup routing. Without that comparison, there is no proof.

Approach

Trials start on simulated data and replayed scenarios. Any measurement on real traffic would need a contractual framework and notice to the parties concerned. Results, positive or not, would be documented before any product decision.

Limits

• The programme is exploratory. It may end in a negative conclusion.
• It does not work around any authentication or route block.
• It promises no fraud reduction, no availability and no timeline.
• It depends on the data and capabilities that partner carriers agree to share.