Anomalous Behavior Detection and Alerting for Congested Maritime Environments

Q1 received a substantive new answer clarifying that off-board/non-organic data feeds (e.g., airborne AIS/ADS-B) are acceptable as supplemental sources, but the system must remain effective without them and avoid requiring large persistent onboard databases. All other Q&As (Q2-Q5) remain unchanged.

**Key Changes to Q&A:** **New Q1 added:** Clarifies that off-board or non-organic maritime traffic data (e.g., airborne AIS/ADS-B collection) delivered via existing ship communications is a responsive "other data source," supporting PoL context when entering regions with no onboard history. **Questions renumbered:** Previous Q1-Q4 are now Q2-Q5 to accommodate the new Q1. **No answer changes:** All existing answers remain substantively identical; only question numbering shifted.

Q2 received a new answer clarifying that established PoL algorithms are acceptable if proposers demonstrate how their system-level solution addresses the topic's unique constraints (storage, explainability, confidence levels). Q3 received a new answer stating proposers can recommend any architectural approach (edge-processing, secure reach-back, or hybrid) and should base HMI designs on open standards and HSI best practices rather than classified SSDS specifications.

Q1 received a new answer clarifying that Phase I feasibility may use any representative/synthetic/public datasets proposers justify as sufficient—Government does NOT mandate specific contact-behavior cases. Proposers select and justify test scenarios based on their technical architecture and SSDS relevance.

**Q&A Changes Summary:** Added Q1 requesting Government guidance on minimum contact-behavior case coverage priorities (route deviation, AIS/ADS-B loss, kinematic discontinuity, loitering, abnormal closure, mixed traffic, composite-track cases) for Phase II decision-making. Existing Q2-Q4 preserved with Q3 (previously unanswered) remaining pending; Q4 answer confirmed synthetic data acceptable if sufficient quality/quantity for evaluation.

Status changed from Pre-Release to Open

Added 2 new Q&As: Q1 asks if established (non-novel) PoL algorithms are acceptable; Q2 addresses data link architecture vs. novel algorithms and requests HMI/cognitive workload guidelines for Phase I designs. Original Q1 (now Q3) on synthetic data use remains unchanged.

Q1 received an answer clarifying that representative synthetic data is acceptable for Phase I evaluation, provided it is sufficient quality and quantity for detailed product assessment and Phase II decision-making.

This Q&A addresses whether Phase I feasibility demonstrations for DON26BZ03-NV063 can use synthetic or open-source datasets to validate explainable anomaly detection logic, provided the proposed design shows integration with actual government track data in Phase II.

New opportunity: Anomalous Behavior Detection and Alerting for Congested Maritime Environments