Definition
Average passenger connection quality impact attributable to AI-optimized gate assignments: measuring whether the gate optimization system's decisions create or destroy connection feasibility and comfort for passengers with tight itineraries.
The Measurement Gap
Gate assignment optimization systems are designed to maximize aircraft gate utilization, minimize taxi conflicts, and reduce ground delay: all legitimate operational goals. Connection quality for passengers is rarely included in the gate optimizer's objective function. As a result, the system may assign a gate that is operationally efficient but significantly increases walk distance for a bank of tight connecting passengers, changes the terminal and adds a bus transfer, or reduces available connection time in ways that create misconnects.
Traditional gate management metrics measure gate utilization, ground delay minutes, and taxi conflict rates: all supply-side operational measures. There is no standard metric for what gate optimization decisions do to passenger journey experience. Misconnect rates capture the downstream outcome, but they conflate gate assignment effects with delay effects and don't isolate the gate assignment contribution.
Decision supported:
Is the gate optimizer making passengers' lives better or worse? Should connection quality constraints be added to the gate optimization objective function? GACIS answers both questions with a single computable metric. A persistently negative GACIS at a hub station (the optimizer is consistently degrading connections relative to the baseline gate plan) is a direct signal that the optimizer's objective function needs a connection quality constraint. The constraint doesn't have to eliminate all gate reassignments; it only needs to prevent reassignments that reduce any connecting passenger's available connection time below a minimum feasibility threshold.
Formula
Plain language: For each connecting passenger affected by an AI gate reassignment, score their connection quality under the AI-assigned gates (walk distance, available connection time, terminal change, bus transfer requirement) and compare to what their connection quality would have been under the original planned gate assignment. Average the difference across all affected connecting passengers.
Notation:
GACIS = (1/P) × Σ [ ConnectionQuality_AI_i − ConnectionQuality_baseline_i ]
Variables:
| Variable | Definition | Source | Unit |
|---|---|---|---|
| P | Number of connecting passengers affected by gate changes in the measurement period | Derived from PSS connection data + gate change records | Count |
| ConnectionQuality_AI_i | Connection quality score for passenger i under AI-assigned gates: f(available_connection_time, walk_distance, same_terminal_flag, bus_transfer_flag), normalized 0–1 | Derived | Score |
| ConnectionQuality_baseline_i | Same score under the originally planned gate assignment | Derived | Score |
Units: -1 to +1. Positive = AI gate assignment improved connection quality on average. Negative = degraded it. Zero = neutral (AI reassignments had no net connection impact). The connection quality function weights should be calibrated against misconnect rate data: if walk distance is a better predictor of misconnect than available time at a specific hub, weight accordingly.
Proposed formula, pending validation pilots.