Explained: How the FAA’s Communications Backbone Connects U.S. Air Traffic Control

NEW YORK — The network behind U.S. air traffic control carries far more than telephone calls. It moves radar information, flight data and other operational traffic between the systems and facilities that controllers rely on. Its role helps explain why a communications problem can affect several functions at once.

Friday’s nationwide ground stop and continuing airline recovery brought that infrastructure into public view. Transportation Secretary Sean Duffy attributed the disruption to overloaded older switches and routers producing inconsistent information.

At the same briefing, Homeland Security Secretary Markwayne Mullin offered a preliminary explanation involving patches and the addition of new equipment and technology, saying the newer technology was “overwhelming the older system.” His account did not identify the specific changes or establish how they contributed to the overload.

These preliminary accounts do not establish the complete sequence that produced the disruption; the official materials reviewed for this explainer do not establish the complete sequence that produced it.

What the Backbone Carries

The Federal Aviation Administration (FAA) calls its interconnecting network the FAA Telecommunications Infrastructure (FTI). In its August 2022 architecture document, the agency describes FTI as supporting its wide-area communications needs. In practical terms, it provides connections across geographically separated FAA facilities and systems.

L3Harris, which operates the network, described its role in a May 11 account of telecommunications modernization. The company said the infrastructure supports voice, video, radar, weather and flight data, and is designed to continue serving older systems as other parts of air traffic control are modernized.

That transport role matters when interpreting an outage. A radar sensor, the network carrying its information and the equipment displaying that information perform different jobs. A communications interruption can interfere with delivery without establishing that the sensor itself has stopped working. Equally, a problem with voice traffic carried through the network does not, by itself, identify a fault in an aircraft’s radio.

In that November 13 explanation, published by the Department of Transportation, Duffy described instability in radar information and communications at en route centers and terminal radar approach control facilities. He explicitly said the FAA had not lost all radar or telecommunications. The departure halt was a precaution in response to that instability, rather than evidence that every component had failed.

Older Technology and IP Already Coexist

References to legacy equipment can give the impression that the entire network uses one obsolete technology. The FAA’s own architecture document gives a more specific picture: FTI already listed an Operational IP network among its services in 2022. IP stands for Internet Protocol. The presence of older equipment does not make every connection analog.

A November 2023 independent safety review hosted by the FAA identified a particular transition problem. It warned that the network depended on time-division multiplexing, an older telecommunications technology whose support from telephone providers was disappearing. The review described the planned replacement, FAA Enterprise Network Services, as an IP-based system.

During that transition, the review warned, an unplanned failure involving time-division multiplexing equipment could disconnect individual installations or even an en route center from the wider network. It said the FAA’s July 2023 operational risk assessment had placed FTI risk at its highest level.

The review documented transition risks years before Friday. Connecting its warning to this outage requires evidence identifying the affected equipment and the failure sequence.

Backup Design and Friday’s Evidence

L3Harris describes multiple telecommunications carriers and built-in redundancy as features of FTI. In its May modernization account, the company also described installing new infrastructure while the existing network remained in service, testing it in parallel and switching over during periods of lower traffic with backup systems available.

These are the contractor’s descriptions of its architecture and deployment methods. They provide a basis for examining how the network is meant to stay available during maintenance and failures. They are not a record of which backup paths activated on November 13 or how those paths performed.

The official statements reviewed do not provide that sequence. An explanation of Friday’s resilience therefore needs evidence about the affected primary and alternate paths, whether they depended on common equipment, and whether traffic was transferred and how alternate paths performed. The nationwide departure halt alone does not establish that every backup failed.

Why SMART Is a Separate System

The FAA’s Strategic Management of Airspace, Routes and Trajectories (SMART) tool addresses another part of air traffic management. Its September launch announcement describes software that combines information about weather, traffic and capacity to help specialists make scheduling recommendations. FAA staff review the suggestions, and local facilities can accept or decline them.

As Airways explained in its coverage of SMART’s expansion, that planning capability is distinct from completion of the broader infrastructure overhaul. SMART does not replace controllers or take control of aircraft.

At Friday’s briefing, Duffy said SMART was disconnected from the affected system and could not have caused the disruption. That is his stated exclusion of the tool. It does not establish what initiated the older equipment’s overload.

Modernization Requires Coordinated Changes

The Government Accountability Office’s September 2026 modernization assessment illustrates the scale of work around these networks. It found that the FAA lacked a reliable lifecycle cost estimate and an integrated master schedule for the first phase of its new modernization initiative. As of May, the agency had 11,389 individual project schedules that GAO said were not integrated.

The Department of Transportation partially concurred with GAO’s recommendations, pointing to program schedules and a contractor scheduling tool. GAO maintained that the arrangements did not adequately coordinate project dependencies.

For network modernization, that coordination has practical importance: equipment changes must fit around other installations and continuing operations. The oversight findings concern how the work is planned. They do not connect a scheduling deficiency to Friday’s outage, or establish that a particular funding package would have prevented it.

In Airways’ Pod-Ed on artificial intelligence (AI) and legacy systems, DataArt’s Tim McMullen argues for building on established platforms and preserving human understanding of business requirements as AI changes software development. That discussion concerns airline technology generally; it does not diagnose the FAA outage.

What a Technical Account Still Needs to Establish

The official material reviewed supports Duffy’s attributed description of older equipment producing inconsistent information. A reconstruction still needs to identify the affected components and facilities, establish what triggered the overload, and trace how the disruption reached multiple locations.

It also needs to document how alternate connections behaved and which corrective actions restored stability. That evidence would allow readers to distinguish a failed component, a problem shared by several systems and a limitation in the network’s ability to contain a fault. Until those details are established, the backbone’s documented role and historical vulnerabilities provide context for Friday’s disruption, rather than a completed explanation of its cause.

Bottom Line

Besides editing, I work in IT and software development, including for Airways, and regularly connect newer tools to existing systems. At Airways, we approach modernization by first understanding the architecture already in place: its dependencies, the data that must remain reliable and the behavior a change must preserve. Our editorial system, for example, is built around guidelines developed and refined over the past decade.

AI-assisted development carries the same responsibilities. Generated code can be deterministic; it still needs clear requirements, technical review and testing against the systems it will actually touch. Understanding the legacy environment, documenting that knowledge and planning a safe rollback are part of the engineering work.

For Friday’s disruption, I would want a technical account to establish whether a change contributed, how any such change was tested and how the existing system responded. AI’s role, if any, has to be established through that evidence.