AI Technology
Security Anomaly Detection
AI-powered anomaly detection across GPS telemetry, premises sensors, and access control data — identifying statistically unusual patterns that may indicate security incidents before they escalate.
How it works
Anomaly Detection establishes a behavioural baseline for each asset and location, then continuously monitors for statistically significant deviations from that baseline.
Step 1
Baseline Learning
The system observes 21 days of normal operational patterns — typical routes, stop durations, access times, and movement speeds — to establish a behavioural baseline per asset.
Step 2
Continuous Deviation Analysis
Real-time data is compared against the baseline using a statistical anomaly model. Deviations are scored by magnitude — small deviations ignored, large deviations escalated.
Step 3
Context-Aware Alerting
The system uses contextual intelligence to reduce false positives — for example, a vehicle stopping at a known fuel station is not flagged, but stopping at the same location at 02:00 is.
Data inputs
- GPS telemetry and stop patterns
- Access control event logs
- CCTV motion event data
- Historical behavioural baselines
- Operator-defined normal hours and zones
Outputs
Anomaly Alert
Real-time notification for statistically significant deviations
Anomaly Score
Severity score (1–10) for each detected anomaly
Evidence Package
Timestamped evidence bundle including GPS, video clip, and access log for each anomaly
Trend Report
Weekly summary of anomaly patterns for security review
Example insight
Anomaly detected: Vehicle LGS-BUS-007 stopped at an unregistered location on Lekki-Epe Expressway for 47 minutes at 23:12 — significantly outside baseline (average stop duration: 4 minutes). SOC reviewing.
See Security Anomaly Detection on your fleet
Talk to our security architects about vehicles, fleets, churches, schools, and enterprise assets.