Modernizing the National Airspace System is often discussed in terms of new technologies: cloud computing, artificial intelligence, machine learning, automation, digital twins, and increasingly sophisticated decision-support tools.
However, beneath nearly every one of those capabilities is a less visible requirement: high-quality data.
Modern air traffic management depends on information arriving from many different sources, including surveillance systems, flight plans, airline operational systems, weather services, airport systems, and FAA information services. As aviation becomes increasingly connected and information-driven, the challenge is no longer simply accessing data. It is making sure that data is timely, consistent, complete, correctly associated with the right flight, and trustworthy enough to support an operational decision.
It is safe to say that data quality is one of the fundamental challenges of air traffic modernization.
Modern Air Traffic Management Runs on Shared Information
The FAA’s modernization strategy increasingly depends on information moving between systems and stakeholders.
System Wide Information Management (SWIM), for example, serves as the digital data-sharing backbone of NextGen, allowing aviation systems and users to access information through common services rather than relying solely on customized point-to-point interfaces.
Modern air traffic management is also becoming increasingly trajectory-based, using four-dimensional flight trajectories (latitude, longitude, altitude, and time) to support more strategic planning and coordination across the NAS. Those trajectories are continually updated as operating conditions change and shared among systems and stakeholders.
This creates tremendous opportunities for more predictable and collaborative operations. It also means that the quality of the underlying information matters enormously.
A decision-support system cannot reliably calculate demand if flight records are duplicated. A surface-management tool may produce misleading estimates if departure times are stale. An airline and an FAA system may appear to disagree about a flight simply because their records have not been properly correlated. And a machine-learning model trained on incomplete or inconsistent historical information may carry those weaknesses into its predictions.
In other words, more data does not automatically produce better decisions.
What Does “Data Quality” Mean in Aviation?
There is no single data-quality problem. Operational aviation systems must account for several dimensions of quality at once.
- Completeness: Are all of the necessary fields and records available? Missing flight plans, surveillance observations, airport events, or operational updates can leave systems working from an incomplete picture.
- Timeliness: How old is the information? In a dynamic environment such as an airport surface or congested terminal area, information that was accurate several minutes ago may no longer reflect current conditions.
- Consistency: Do different sources agree? Airline, FAA, surveillance, and third-party systems may describe the same operation differently or update information at different times.
- Identity and correlation: Do systems know that two records actually represent the same flight? Reliable flight matching becomes particularly important when data is being aggregated across multiple systems.
- Accuracy: Does the information represent what is actually happening in the operation?
- Context: Even technically correct information can be misleading without understanding where it came from, how it was generated, and how it should be interpreted operationally.
These challenges are not entirely new. NASA researchers studying traffic-flow-management data more than two decades ago examined problems such as missing flight-plan information, departure delays, altitude errors, deviations from filed routes, and dropped track data, and noted that the quality of decision support is inherently influenced by the quality of its inputs.
What has changed is the scale of the data ecosystem and how much modern capabilities depend upon it.
Integration Is Not the Same as Data Fusion
Connecting systems is an important first step. But moving information between systems does not necessarily make that information coherent.
Consider a modern flight record. Information about the same aircraft may arrive through multiple FAA services, airline feeds, airport systems, surveillance sources, and third-party providers. Each source may update at a different frequency. Identifiers may differ. One source may contain more recent information while another contains details unavailable elsewhere.
Simply displaying all of those feeds leaves the operational user, or every downstream application, to resolve those differences independently. This is why data fusion plays such an important role in modern aviation systems.
Mosaic ATM has worked with aviation data integration and fusion for decades. The Mosaic Fuser connects to multiple data feeds, consumes and parses their information, mediates between disparate sources, and creates coherent, current records organized around individual flights. The technology has supported both research and operationally relevant applications, including NASA’s Airspace Technology Demonstration 2 (ATD-2) and Digital Information Platform (DIP) work.
During ATD-2, for example, NASA and its partners needed information from SWIM, airlines, and other sources to support surface traffic management. Mosaic developed a customized data-processing and fusion capability that synthesized those sources into consistent flight information for downstream systems.
That work illustrates an important distinction: data integration makes information available; data fusion makes it usable.
AI Makes Data Quality Even More Important
Artificial intelligence and machine learning can uncover patterns that would be difficult to identify manually, but their performance is inseparable from the information they receive.
NASA’s Digital Information Platform (DIP) research demonstrated how cloud infrastructure, integrated aviation data, and machine-learning-based predictive services could support more efficient and increasingly information-centric airspace operations. DIP explored capabilities ranging from data integration and traffic prediction to digital trajectory management, providing an important research environment for evaluating how advanced data services could support future aviation decision-making.
Mosaic ATM supported this research through work on aviation data integration and machine-learning-based airport surface prediction. These efforts included transforming surface-management capabilities into scalable predictive services capable of estimating factors such as airport configuration, runway assignments, surface movement times, demand, and potential delays.
The underlying challenge remains highly relevant as the FAA advances newer operational capabilities built around integrated data, predictive analytics, and strategic trajectory management: an AI or machine-learning model does not inherently know when an upstream data feed has changed, when information is missing, or when two records represent the same physical flight. Those problems must be engineered around.
As aviation increases its use of predictive analytics and AI, data validation, provenance, monitoring, reconciliation, and operational evaluation therefore become part of the AI problem itself, not separate supporting activities.
Data Quality Must Also Be Maintained Over Time
Another easily overlooked challenge is that aviation data environments do not remain static.
Interfaces evolve. Message formats change. New systems come online. Operational procedures change. New information sources become available, and existing feeds can change behavior. A system that handles today’s data correctly therefore cannot simply be deployed and forgotten.
As the FAA modernizes its traffic management systems and moves toward more integrated, information-driven operations, capabilities such as data matching, reconciliation, validation, and conversion remain essential. These functions help ensure that information coming from different sources can be combined into a consistent and usable operational picture.
Mosaic encounters similar challenges in its own data systems. Maintaining the Mosaic Fuser, for example, requires adapting parsers and processing logic as source data streams evolve.
Mosaic’s Track-Nexus research addresses a related challenge in distributed tracking environments, where different nodes can develop duplicate, stale, or conflicting track information because of asynchronous updates and network latency. Track-Nexus explores continuous comparison and reconciliation of distributed track states to maintain a more consistent operational picture.
Different applications require different architectures, but the underlying principle is the same: trustworthy information requires continuous management.
Data Quality Is Ultimately About Operational Trust
The goal of improving aviation data is not simply to produce cleaner databases. It is to help people and systems make better decisions.
For an air traffic controller, traffic manager, airline operations center, airport operator, or automated decision-support service, confidence in a tool depends in part on confidence in what that tool knows.
If users repeatedly encounter conflicting flight information, stale predictions, unexplained discrepancies, or missing operational events, they may stop trusting the system, even when its underlying algorithms are technically sophisticated.
Conversely, when information is reconciled, traceable, timely, and presented within the right operational context, increasingly advanced tools can become practical parts of real-world operations.
This becomes especially important as the aviation industry moves toward increasingly trajectory-based and predictive air traffic management, digital twins, advanced analytics, and AI-assisted decision support, these capabilities depend on a reliable, shared understanding of both the current state of the NAS and how that state is likely to evolve.
Building the Data Foundation for the Next Generation of Air Traffic Management
The future NAS will almost certainly contain more data, more connected systems, more automation, and more predictive capabilities than the system operating today.
That does not make data quality less important. It makes it foundational.
Modernization therefore cannot focus solely on building the next application or deploying the next algorithm. It must also address the less visible engineering required to ingest, clean, correlate, reconcile, monitor, and maintain the information those technologies depend upon.
Through work spanning aviation data fusion, ATD-2, NASA’s Digital Information Platform, machine-learning-based airport surface prediction, distributed track synchronization, and commercial aviation data services, Mosaic ATM has developed deep expertise at this intersection of aviation operations and data engineering.
As air traffic management becomes increasingly digital, the quality of the decisions we can automate, predict, or optimize will continue to depend on the quality of the information underneath them.
The technology shaping the future of aviation may be increasingly sophisticated. But before a system can make a smarter decision, it first has to know it can trust the data.
