Safety-critical Software Solutions at FREQUENTIS: A Technical Deep Dive into Air Traffic Management with Djamel Sahnine

Why this session matters for engineers

In “Safety-critical Software Solutions,” Djamel Sahnine (FREQUENTIS) doesn’t talk about yet another web service. He talks about software where every line of code is safety-critical. Air Traffic Management (ATM) is a domain where software protects lives. The session centers on how engineers can apply technology so air traffic controllers can consistently make the right calls: monitor, analyze, execute—on repeat.

Our editorial takeaway at DevJobs.at: “Making the skies safer” is not a slogan; it’s an architecture principle. Everything—from sensor fusion and video analytics to trajectory prediction, safety nets, digital voice, microservices, containers, and Kubernetes—is measured by whether it helps controllers work more safely, more reliably, and more clearly.

Problem space: Safety, situational awareness, and decision pressure

At its core, ATM is about guaranteeing separation between aircraft. That boils down to:

• delivering a precise, consolidated picture of air and ground traffic,
• providing decision support that surfaces risks early,
• and executing decisions with high confidence—especially via robust voice communication with pilots and adjacent sectors.

Sahnine frames ATM as an endless feedback loop in the controller’s mind: “Monitor – Analyze – Execute – and back again.” FREQUENTIS’ job is to deliver technology that is dependable in each step—without overloading, confusing, or distracting the human operator.

The triad: Monitor, analyze, execute

Monitor: Build a precise picture

Everything starts with a correct, trustworthy, and unified view of the situation:

• Sensing: FREQUENTIS develops radar sensors and ingests heterogeneous surveillance data.
• Data fusion: Algorithms merge multiple feeds into a single, accurate aircraft position for each flight.
• Context: Fusing with flight data enriches geometry with semantic intent and identifiers.

For airport operations, there is also video:

• Video surveillance and Remote Tower: Camera systems capture the airfield; complex video-processing applications curate the output.
• AI-driven processing: Algorithms detect objects and extract useful signals from video streams to “always provide the most useful information” to the controller.

The result is a consistent, trusted situational picture—in the air and on the ground—that underpins every decision.

Analyze: Provide decision support

Decisions rely on the now, but depend on robust forecasts and protective safeguards:

• Trajectory prediction: A “complex trajectory predictor” estimates where an aircraft will be in 5 seconds, 5 minutes, or an hour. Inputs include aircraft behavior models and meteorological factors (wind, pressure). Predictions are refreshed every second.
• Conflict detection: With trajectories available, safety nets can flag potential future hazards—specifically impending loss of separation. They also detect abnormal behavior (route deviations) and check whether pilot clearances have been executed properly.

Execute: Communicate without failure

Once a decision is made, it must be carried out and shared—reliably—with pilots and other controllers. In critical situations, the “ultimate” channel remains voice over radio:

• Digital voice communication: FREQUENTIS positions itself as the world leader in digital voice communication systems for ATM.
• High availability: These systems “cannot fall down.” Stability and fault tolerance are paramount.
• IP-based networks: FREQUENTIS introduced one of the first IP-based communication networks for air traffic control—“extremely robust,” leveraging microservices, containers, and modern data center technologies.

Situational picture in depth: Sensors, fusion, video, and context

Multi-source surveillance

Delivering a “single, reliable, and accurate position” is a nontrivial engineering challenge. The session highlights:

• Radar sensors capture raw positions.
• Diverse surveillance inputs are fused into a single track per aircraft.
• Flight data are linked to enhance raw geometry with intent and attributes on the controller’s screen.

Video and Remote Tower

The camera feed is not a dumb relay:

• Complex video-processing extracts and prioritizes signals.
• AI algorithms support object and event detection to present “the most useful information.”
• Remote Tower: Controllers can operate “miles away” and still maintain a reliable view of the apron and runway environment.

This imposes strict technical expectations on developers: low latency, synchronization with other sensors, graceful degradation, and robust pipelines are not nice-to-haves—they are safety requirements.

Decision support in depth: Prediction, safety nets, and conformance

Trajectory engine

Predictions are “refreshed every second,” blending multiple data domains:

• State: current surveillance-derived kinematics
• Models: aircraft behavior/performance
• Environment: meteorological factors such as wind and pressure

The goal is a precise temporal projection, feeding conflict detection and giving controllers enough lead time to act.

Safety nets

Safety nets serve several safety functions:

• Predict conflicts: flag potential loss of separation before it happens
• Detect deviations: identify abnormal states or route deviations
• Conformance monitoring: verify that pilot instructions have been executed correctly

Common denominator: buying time. Early warnings expand the controller’s maneuvering space.

Human-machine interfaces: User-centric, not marketing-driven

Sahnine’s stance on UI is unambiguous: interfaces must not “confuse” but deliver the right information, at the right time, in the right form.

FREQUENTIS provides a “plurality of user interfaces,” including:

• Air situation radar displays
• Ground radar displays
• Electronic flight strips
• Weather and meteorological displays

Design tenets emphasized in the session:

• Safety First and Human Factors drive implementation and interaction design.
• UI is not for pushing users into actions they do not want—it’s for enabling sound decisions under pressure.

For engineering teams, that means UI/UX is a safety function. Information density, alert design, color semantics, and input mechanics must respect cognitive load and human factors.

Flow management and optimization: Order instead of chaos

Left unmanaged, traffic gets “chaotic.” FREQUENTIS builds flow management and optimization applications:

• Goal: determine and enforce the “most optimal” aircraft flow and sequence.
• Outcomes stated in the talk: improved safety, reduced CO2 emissions, and better passenger experience via dramatically reduced delays at airports and in the air.
• Technically: complex algorithms for sequencing and optimization, tightly integrated with prediction and conflict logic.

Again, prediction quality and UI transparency determine how much benefit controllers can realize in operations.

UTM: Bringing order to drone traffic

Drone usage “is on the rise.” Without control, chaos is likely. FREQUENTIS develops UTM (Unmanned Traffic Management) systems that:

• organize and monitor drone traffic—in cities and around airports,
• and, per the session, FREQUENTIS counted among the first companies to believe in and build such systems.

For engineers, this signals the transfer of ATM principles (situational awareness, prediction, safety nets, communication) to new vehicles and flight envelopes.

Digitalizing aeronautical information: Data, models, and distribution

ATM “publishes a huge amount of information,” including:

• airspace information,
• routes in the air,
• and airport information.

This must be digitized, managed, and distributed:

• Applications, databases, and tools model, maintain, and handle this information,
• and distribute it “seamlessly” to various partners.

The architectural implications: strong data models, versioning, quality assurance, consistency across systems, and a reliable distribution layer.

Execution with zero compromise: Voice as the lifeline

When stakes are highest, voice prevails. The session sets clear priorities:

• “Voice communication is the ultimate way to communicate with the pilot—it cannot fall down.”
• FREQUENTIS is, as stated, “the world leader in digital voice communication systems” for ATM.
• The company introduced one of the first “IP-based communication networks” for ATC—“extremely robust,” and built on microservices, containers, and data center practices.

This is the crucial bridge: mission-critical and cloud-native do not contradict each other—if engineered correctly.

Mosaics: A platform uniting mission-critical and cloud-native

“Mosaics” is the platform FREQUENTIS uses to deploy, run, and manage the applications described in the talk. Its core ambition:

• Reconcile two worlds: “mission-critical, high availability” and “cloud native, CI/CD, containers, Kubernetes.”
• Provide the same guarantees of availability and trust, while leveraging innovative technologies.

For dev teams, this is a strong architectural stance: cloud technologies are curated and integrated to meet safety requirements. Microservices and containers are purpose-driven—not fashion-driven.

Engineering at FREQUENTIS: Tech, methods, and values

The talk lists the engineering stack actively used:

• AI for object recognition and video processing
• Microservices architectures, containers, Kubernetes
• Agile methods
• Cloud-native data center approaches

Just as important are the values shaping the code:

• Safety First: Safety considerations lead every design and implementation choice.
• Human Factors: Ensure the controller gets the “right information in the right way at the right time.”
• Community: ATM is a “big family.” Projects span all continents, with a worldwide engineering community sharing experience and knowledge.

Practical lessons for engineers

From our vantage point, “Safety-critical Software Solutions” yields several actionable takeaways:

1. Treat safety as a first-class requirement

• Make safety explicit in your acceptance criteria and architecture. In safety-critical domains, it’s not a side constraint—it is the core.

1. Prefer data fusion over single-sensor perfection

• Build robust situational pictures by integrating heterogeneous sources with quality metrics and consistency checks.

1. Shift left with prediction

• Trajectory prediction and safety nets turn reaction into anticipation. Update continuously, include domain models (aircraft, weather), and define conflicts crisply.

1. UI is a safety mechanism

• Design interfaces to reduce cognitive load, prioritize relevance, and minimize operator error. Alerting and interaction patterns must be human-centric.

1. Video is a data source, not just a feed

• Process video algorithmically to surface useful signals. Relevance beats raw volume.

1. Flow optimization is a system discipline

• Interlock prediction, sequencing, and UI to realize operational benefits (safety, punctuality, emissions).

1. Join cloud-native with mission-critical rigor

• Microservices, containers, Kubernetes, and CI/CD are usable—if reliability, fault tolerance, and observability meet mission-critical standards. A platform like “Mosaics” embodies this.

1. Protect the last line of defense

• Architect around the inevitability that voice is the lifeline. Nothing should compromise it.

1. Take data governance seriously

• Aeronautical information needs strong modeling, versioning, and distribution. Data quality is safety quality.

1. Values as architectural constraints

• Safety First and Human Factors belong in your Definition of Done, code reviews, and architecture guardrails.

Key quotes and messages

• The mission is to “make the skies safer,” and technology choices are judged by that yardstick.
• “Monitor – Analyze – Execute” is the controller’s continuous loop.
• “Safety Nets” are the protective layer that identifies future conflicts before they occur.
• Voice communication is the “ultimate” channel in critical situations—it must not fail.
• “Mosaics” unites high availability with cloud-native practices.

Conclusion

“Safety-critical Software Solutions” with Djamel Sahnine demonstrates how modern engineering works under hard safety constraints. Sensor fusion, AI-powered video processing, trajectory prediction, and safety nets provide the basis for decisions. User-centered interfaces make the information effective. Robust voice communications and IP-based networks secure execution. And a platform like “Mosaics” shows that microservices, containers, Kubernetes, and CI/CD can coexist with mission-critical availability—when safety and human factors are the guiding principles.

If you build safety-critical systems (or intend to), this session provides a clear compass: every technical choice must answer the same question—does it make the sky safer?

• 

  

  Drones Swim

  Thomas Lutz von Frequentis zeigt in seinem devjobs.at TechTalk welche Rahmenbedingungen geschaffen werden, um in Zukunft mit einer hohen Zahl an Drohnen umgehen zu können.

  Watch now
• 

  

  CCR

  Richard Draxelmayr von Frequentis spricht in seinem devjobs.at TechTalk darüber, welche Vorteile eine gute Review-Kultur im Devteam hat.

  Watch now

• 

  

  Günter Graf, VP New Business Development von Frequentis

  Der VP New Business Development von Frequentis Günter Graf umreißt im Interview die Inhalte des Unternehmens, was der Fokus bei New Business Development ist und erläutert auch, warum gutes Domain Knowledge den Unterschied macht.

  Watch now
• 

  

  Karl Wannenmacher, Head of Public Safety Products von Frequentis

  Der Head of Public Safety Products von Frequentis Karl Wannenmacher erzählt im Interview über die Organisation des Devteams der Abteilung Public Safety, wie dort das Recruiting geschieht und welche technischen Challenges es gibt.

  Watch now