Periodic Reporting for period 1 - DeepFlow (enabled Dynamic Air Traffic Flow Configuration for Flow-Centric Airspace Management)
Reporting period: 2024-09-01 to 2025-08-31
Context and Overall Objectives
The DeepFlow project addresses one of the key challenges facing the European Air Traffic Management (ATM) system, which is the need for a more flexible, efficient, and adaptive approach to Air Traffic Flow and Capacity Management (ATFCM). Despite continuous improvements in operational coordination and network management, current practices remain largely sector-based and constrained by static planning paradigms. As traffic demand becomes increasingly dynamic and unpredictable, these limitations hinder the system’s ability to manage capacity efficiently and respond to short-term imbalances. DeepFlow has been conceived to explore an alternative operational paradigm that shifts the focus from sector-based management to flow-centric operations, where air traffic is managed in terms of dynamic, data-driven flow structures that better reflect the actual movement and evolution of traffic across the network.
The main goal of DeepFlow is to develop and demonstrate a flow-based ATFCM and airspace management concept capable of enabling more efficient and flexible use of airspace and resources. The project is positioned to achieve Technology Readiness Level (TRL) 1, aiming to create the theoretical and methodological foundation for this innovative approach. Specifically, focuses on the development of a set of core methodologies and algorithms for flow pattern extraction, flow monitoring, congestion assessment, flow prediction, and dynamic flow regulation (including re-routing and adaptive measures). Together, these elements will form the basis of an integrated flow-centric framework designed to support capacity-on-demand and dynamic airspace configuration, two central pillars of the Digital European Sky vision.
Motivation and Strategic Relevance
The motivation behind DeepFlow stems from the growing operational pressures on the European ATM network. With air traffic expected to recover and increase steadily over the coming years, existing ATFCM processes face mounting difficulty in maintaining network performance, predictability, and sustainability. Current tools often lack the predictive and adaptive capabilities required to handle large-scale variations in demand and capacity, particularly in complex and congested areas. DeepFlow directly addresses these gaps by introducing a data-driven, algorithmically supported framework for understanding and managing air traffic as "evolving flows" rather than "isolated trajectories". This perspective is consistent with the SESAR 3 JU research and innovation needs outlined in the Digital European Sky programme, particularly in supporting scalability, resilience, and digital transformation of network management functions.
Pathway to Impact
DeepFlow’s pathway to impact is defined by its contribution to building the scientific and methodological underpinnings of the future flow-centric Digital European Sky. While the project’s maturity level remains low, its expected impact is high in terms of conceptual innovation and knowledge generation. The project establishes a new research direction that can be leveraged by subsequent SESAR projects at higher TRLs to develop operational prototypes and validation exercises. The methodologies for flow extraction, congestion metrics, and flow reconfiguration provide essential building blocks for predictive, adaptive network management solutions capable of improving airspace capacity, flight efficiency, and overall network resilience.
Moreover, DeepFlow actively fosters synergies with other European research initiatives, including FCA, HARMONIC, ISLAND, and KAIROS, ensuring that its findings complement and reinforce ongoing work in related domains. Collaborative exchanges and joint workshops have been organized to share progress, align methodologies, and explore potential integration pathways.
The DeepFlow project addresses one of the key challenges facing the European Air Traffic Management (ATM) system, which is the need for a more flexible, efficient, and adaptive approach to Air Traffic Flow and Capacity Management (ATFCM). Despite continuous improvements in operational coordination and network management, current practices remain largely sector-based and constrained by static planning paradigms. As traffic demand becomes increasingly dynamic and unpredictable, these limitations hinder the system’s ability to manage capacity efficiently and respond to short-term imbalances. DeepFlow has been conceived to explore an alternative operational paradigm that shifts the focus from sector-based management to flow-centric operations, where air traffic is managed in terms of dynamic, data-driven flow structures that better reflect the actual movement and evolution of traffic across the network.
The main goal of DeepFlow is to develop and demonstrate a flow-based ATFCM and airspace management concept capable of enabling more efficient and flexible use of airspace and resources. The project is positioned to achieve Technology Readiness Level (TRL) 1, aiming to create the theoretical and methodological foundation for this innovative approach. Specifically, focuses on the development of a set of core methodologies and algorithms for flow pattern extraction, flow monitoring, congestion assessment, flow prediction, and dynamic flow regulation (including re-routing and adaptive measures). Together, these elements will form the basis of an integrated flow-centric framework designed to support capacity-on-demand and dynamic airspace configuration, two central pillars of the Digital European Sky vision.
Motivation and Strategic Relevance
The motivation behind DeepFlow stems from the growing operational pressures on the European ATM network. With air traffic expected to recover and increase steadily over the coming years, existing ATFCM processes face mounting difficulty in maintaining network performance, predictability, and sustainability. Current tools often lack the predictive and adaptive capabilities required to handle large-scale variations in demand and capacity, particularly in complex and congested areas. DeepFlow directly addresses these gaps by introducing a data-driven, algorithmically supported framework for understanding and managing air traffic as "evolving flows" rather than "isolated trajectories". This perspective is consistent with the SESAR 3 JU research and innovation needs outlined in the Digital European Sky programme, particularly in supporting scalability, resilience, and digital transformation of network management functions.
Pathway to Impact
DeepFlow’s pathway to impact is defined by its contribution to building the scientific and methodological underpinnings of the future flow-centric Digital European Sky. While the project’s maturity level remains low, its expected impact is high in terms of conceptual innovation and knowledge generation. The project establishes a new research direction that can be leveraged by subsequent SESAR projects at higher TRLs to develop operational prototypes and validation exercises. The methodologies for flow extraction, congestion metrics, and flow reconfiguration provide essential building blocks for predictive, adaptive network management solutions capable of improving airspace capacity, flight efficiency, and overall network resilience.
Moreover, DeepFlow actively fosters synergies with other European research initiatives, including FCA, HARMONIC, ISLAND, and KAIROS, ensuring that its findings complement and reinforce ongoing work in related domains. Collaborative exchanges and joint workshops have been organized to share progress, align methodologies, and explore potential integration pathways.
During RP1, the project made significant progress in developing the foundations of a flow-centric operational concept aimed at transforming Air Traffic Flow and Capacity Management (ATFCM) into a more predictive, adaptive, and efficient process. The main scientific and technical efforts were dedicated to conceptualizing and designing the overall framework, supported by a structured methodology for modelling air traffic as dynamic flows rather than static sector-based entities. This approach represents a fundamental innovation within the ATM domain, introducing a data-driven paradigm capable of improving demand–capacity balancing, and enabling capacity-on-demand operations across the European airspace network.
The project achieved an important stage with the elaboration of the Initial Concept Outline, which defines the core elements of the flow-centric operational concept, its processes, and supporting services. This work was grounded in extensive collaboration with operational stakeholders (NM and FMP) and experts, ensuring that the proposed framework aligns with real-world ATFCM practices. In parallel, technical research advanced toward the development of key algorithmic components that form the scientific backbone of the DeepFlow framework. These included methodologies for flow pattern extraction, major flow monitoring, and congestion assessment, designed to detect and characterize the evolution of critical flow structures within the network. Early algorithmic developments demonstrated promising results in identifying flow patterns and congestion metrics, providing the first tangible steps toward implementing the DeepFlow vision.
Further work concentrated on the development of predictive and regulatory algorithms capable of forecasting and managing flow dynamics in response to evolving traffic and capacity conditions. Flow prediction methods were explored using data-driven and AI-based approaches to anticipate traffic densities, speeds, and potential congestion zones. Complementary to this, techniques for dynamic flow reconfiguration and regulation were designed to adaptively redistribute traffic across the network, contributing to a more balanced and resilient airspace system. These developments collectively established the technical basis for integrating predictive and regulatory capabilities into a coherent flow management framework.
The project also initiated the design and prototyping of the DeepFlow validation platform, which will integrate all algorithmic components and serve as the environment for testing and evaluating the proposed methodologies. Functional requirements and HMI specifications were defined, ensuring that the platform will effectively support operational validation in subsequent phases. In parallel, the Exploratory Research Plan was completed, defining the validation methodology and the structure of the upcoming laboratory exercises aimed at assessing the operational feasibility and potential performance of the flow-centric framework.
The project achieved an important stage with the elaboration of the Initial Concept Outline, which defines the core elements of the flow-centric operational concept, its processes, and supporting services. This work was grounded in extensive collaboration with operational stakeholders (NM and FMP) and experts, ensuring that the proposed framework aligns with real-world ATFCM practices. In parallel, technical research advanced toward the development of key algorithmic components that form the scientific backbone of the DeepFlow framework. These included methodologies for flow pattern extraction, major flow monitoring, and congestion assessment, designed to detect and characterize the evolution of critical flow structures within the network. Early algorithmic developments demonstrated promising results in identifying flow patterns and congestion metrics, providing the first tangible steps toward implementing the DeepFlow vision.
Further work concentrated on the development of predictive and regulatory algorithms capable of forecasting and managing flow dynamics in response to evolving traffic and capacity conditions. Flow prediction methods were explored using data-driven and AI-based approaches to anticipate traffic densities, speeds, and potential congestion zones. Complementary to this, techniques for dynamic flow reconfiguration and regulation were designed to adaptively redistribute traffic across the network, contributing to a more balanced and resilient airspace system. These developments collectively established the technical basis for integrating predictive and regulatory capabilities into a coherent flow management framework.
The project also initiated the design and prototyping of the DeepFlow validation platform, which will integrate all algorithmic components and serve as the environment for testing and evaluating the proposed methodologies. Functional requirements and HMI specifications were defined, ensuring that the platform will effectively support operational validation in subsequent phases. In parallel, the Exploratory Research Plan was completed, defining the validation methodology and the structure of the upcoming laboratory exercises aimed at assessing the operational feasibility and potential performance of the flow-centric framework.