Connectivity-Gap Monitoring Latency Index (CGMLI)

GMI-MAR-015Maritime IndustriesIntegration-Required

Definition

Time delay between anomaly onset aboard and ROC detection, decomposed into transmission and model detection latency.

The Measurement Gap

What This Metric Measures

CGMLI measures the time delay between an anomaly developing aboard a vessel and the ROC receiving data sufficient to detect it. The metric separates total latency into two components: transmission latency (how long until the data reaches shore via satellite) and detection latency (how long the AI takes to flag the anomaly once data arrives). This decomposition reveals whether monitoring gaps are caused by connectivity infrastructure or AI model limitations.

The Decision It Supports

Are monitoring gaps caused by AI model limitations or by data transmission constraints, and which vessels on which routes have the worst gaps? If total latency is 8 hours and 7 of those hours are transmission delay, investing in better AI models will not help. If transmission is fast but detection takes 6 hours, the AI needs improvement. CGMLI gives the fleet technical director the diagnostic to allocate investment correctly.

Why Legacy Metrics Cannot See This

Traditional monitoring metrics measure alarm response time from the moment the alarm reaches the operator. They never measure the time between the anomaly developing aboard and the alarm being generated shore-side. A shore-based system can show perfect response times while anomalies go undetected for hours because the data hasn't arrived yet. CGMLI measures the gap that exists before the monitoring system even knows there is something to monitor: the structural blind spot created by maritime connectivity constraints.

Formula

Primary Formula

CGMLI_transmission = T_data_received - T_anomaly_onset
CGMLI_detection = T_AI_flagged - T_data_received
CGMLI_total = T_AI_flagged - T_anomaly_onset

Variable Definitions

  • T_anomaly_onset = timestamp when the anomaly condition began aboard (retroactively determined from VDR/sensor logs after the vessel reaches connectivity or uploads data)
  • T_data_received = timestamp when the data packet containing the anomaly indicators arrived at the ROC
  • T_AI_flagged = timestamp when the ROC AI system flagged the anomaly for operator attention

Units: Hours. Track all three components separately. Segment by vessel, route type (ocean transit vs. coastal vs. port), and satellite provider to identify systematic patterns.

Edge Cases

Anomalies developing during complete communications blackouts (polar routes, satellite coverage gaps) produce CGMLI_transmission equal to the full blackout duration. These represent the highest-risk events and should be tracked separately. Anomalies detected by onboard systems and reported by crew before ROC detection should measure the delta as a human-vs-AI timing comparison. Minimum sample of 20 detected anomalies per quarter is required.

Proposed formula, pending validation pilots.

At a Glance

Unit
Time (hours)
Trend Direction
Lower
Reporting Frequency
Rolling monthly
Category
Autonomous Operations and Human-AI Authority
← Warranty and Regulatory Interval Breach Rate (WRIBR)MASS Handoff Decision Quality Index (MHDQI) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment