Tactical Separation Provision in High-Volume UTM: Evidence, Requirements, and Communications Architecture for Mixed Airspace
Near-term UTM deployments are appropriately centered on strategic deconfliction, which coordinates routes, corridors, and 4D intent volumes before departure to enable interoperability, policy compliance, and scalable access across operators. As operations scale toward high-density and disturbance-driven mixed airspace, however, strategic coordination may become insufficient as the primary in-flight safety mechanism. Existing UTM concepts already recognize tactical conflict management, but they do not yet clearly characterize when tactical separation provision becomes performance-critical, what communications and implementation constraints govern deployable performance, or how tactical coordination interacts with both strategic coordination and onboard collision avoidance. This report addresses that question through a scenario-driven, real-time high-volume stress campaign supported by more than 231,000 runs, approximately 120.5 million executed flight trajectories, and 15,695 combined simulation and field-testing hours, including experimental validation with a deployed V2V tactical communication stack. The campaign compares strategic-only coordination against both an upper-bound tactical baseline and a deployable tactical controller. Across the evaluated scenarios, strategic-only coordination exhibits two repeatable burden modes at scale: replanning storms, in which brief nonconformance triggers cascading intent updates across many airborne vehicles, and buffer inflation, in which expanding reservations to cover realistic uncertainty collapses throughput and underutilizes airspace. The upper-bound tactical baseline shows that local repair can preserve substantially greater continuity than strategic-only control under anomaly, spoofing, mixed-equipage, hotspot, and observability stress. The deployable controller further shows that tactical separation provision can localize short-horizon repair, while authenticated V2V supports neighborhood coordination and cooperative perception, and onboard collision avoidance remains the last-resort backstop under compressed reaction windows. At the same time, deployable tactical performance depends on specific communications capabilities, especially bounded freshness, authentication, and neighborhood-scale exchange, and its viable operating envelope remains density-and scenario-dependent in the hardest shared-resource and compound-disturbance regimes. Taken together, the results establish a large-scale, operationally grounded evidence base for when tactical separation provision becomes performance-critical in high-volume mixed airspace and how its deployable performance is shaped by communications, observability, and interaction density. The report contributes in three principal ways: it provides a multi-scenario empirical basis for identifying and measuring operational breakpoints at scale across strategic-only and tactical coordination regimes; it distinguishes tactical capability in principle from deployable tactical performance; and it presents, to our knowledge, the first characterization of an all-airborne, intent-first V2V tactical neighborhood communication stack and performance envelope, directly tied to density, contention, observability, and deployable tactical coordination outcomes in high-volume UTM. These contributions are enabled by a real-time, field-anchored evaluation stack that grounds the scenario-derived findings in closed-loop operational behavior. The report then derives practical requirements and validation artifacts for tactical separation provision, including communications and algorithmic requirements, a deployable tactical stack, a layered concept of operations, and a parameterized scenario suite to inform validation, certification, and safe scaling.
The authors acknowledge Qualcomm Technologies, Inc. as a collaborator on the project and as a provider of prototype V2V communication hardware used in the experimental evaluation. Their contributions supported the development and experimental validation of the communication layer examined in this report.
This project is additionally supported by National Artificial Intelligence Research Resource (NAIRR) Pilot award No. NAIRR240493. The authors acknowledge resources provided by TACC Vista (NVIDIA GH100 Grace Hopper Superchip), Hugging Face Spaces and Compute Grants, and PSC Neocortex (Cerebras Wafer Scale Engine AI Accelerator) for contributing to this research result.