GB2430278A - Global neural network for conflict resolution of flights - Google Patents

Abstract

A network of integrated ground systems which monitor, allocate (in real time in response to requests, using a neural network) and control global air-space time (AST) and airport resources to automatically plan, monitor and keep clear paths in AST, communicate decision support information directly to pilots, controllers, airlines and airport operators, using direct channels of communications between the integrated ground systems and equipment on aircraft and terminals at airports and airlines.

Description

I. OBJECTIVES AND SIGNIFICANCE The integration and synchronization of

operations and of global procedures are vitally important for the development of a future global air traffic management system. A new global infrastructure requires supporting Integrated Operational Decision Support (LODS) technology for automated allocation and control of global air space-time and airport resources. In particular it is vitally important for controlling 4D end-to-end trajectories of flights and of global traffic flows according to global air space-time and airports resources available.

The JODS technology automites the global planning of conflict-free allocation of air space-time and airport resources and the clearances of 4D end-to-end trajectories flights. It provides global intelligent mechanisms of allocating and of monitoring the use of air space-time and airport resources and of controlling global traffic flows and four dimensional (4D) end-to-end flight trajectories through satellites, It ensures conflict-free planning for global air traffic and provides synchronjse integrated operational decision support to airlines, airport operators, air traffic controllers and pilots too. The automated intelligent control mechanisms provided with the integrated operational decisi(j support are important. They secure the safest control of air space-tjiiie of traffic and of' airport capacity use. They secure automated monitoring and controlling protection against hazardous and unexpected events.

The IODS systems manage knowledge and control air space-time and airport resources, 4D trajectories and traffic with embedded logic and technology of a global neural architecture of air space-time control structures, They automate the global air space-time design in learning clearancecategories of 4D end-to-end trajectories of flights and the synchronised communications of conflict-lice updates of these trajectories through satellites too. The IODS processes, control rnechanisnis and communications via satellites control global resources and traffic flows on behalf of all users of the JODS systems.

II. RELATION TO PUBLISHED WORK The background of the work on IODS technology for automation of Air Space-Time (AST) allocation and control is descri bed in the enclosed publications.

A brief outline of this published work is as follow. The Hierarchies of Air Space-Time (AST) knowledgeanddata control structures are designed to set up a new technology of managing traffic, knowledge and resources of great importance to conflict-free planning for global air traffic and a new air traffic knowledge management policy. It brings Integrated Operational Decision Support (IOD5) for pilots and controllers, It also brings the benefits of an automated and synchronised AST allocation and a global planning and controlling of traffic flows, 4D end- to-end trajectories of flights and of their automated revisions and conflict-free updates though direct data communications with on-board equipment of aircraft via satellites. It brings JODS parallel processes of continuous end-to-end monitoring of 4D trajectories of flights and of their automated clearances. Any predicted conflicts between trajectories are resolved off-line during a certain air space-time period ahead of current positions of aircraft reported and verified through ground positioning system and satellites. The above principles of future global air traffic management system we invented during our doctoral work. We claim a prior art on those major innovative principles of a future global system for future air traffic management since 2000.

The JODS technology brings the new global control mechanisms for keeping the use of air space-time and airports resources conflict-free via global network of JODS systems ibr airlines, airports, pilots and controllers and communications through satellites.

III. KEY ISSUES IN DEVELOPING THE NEW TECHNOLOGy The work Investigates the design of' a global network of dynamic Air Space-Time (AST) knowledgeanddata control sirructures and of intelligent neural control mechanisms for creating and updating clearance-categories of flights in response to series of requests in a future global air traffic management system. l'he global flows of AST structures organise and control clearance- categories of 4D end-to-efl(j trajectories of flights. Their neural mechanisms and parallel processes monitor and control global traffic flows and their conflict-free use of air-space-time and airport resources.

The important outcomes the new technology brings are the automated learning of decision support knowledge of most efficient clearances of 4D end-to-end trajectories of flights from series of requests and of automated efficient allocation and control of global air space-time and airport resources.

The key issues in achieving this are as follows. The flight requests and traffic flows are distributed, and will be managed and controlled in a future global air traffic management system by distributed and integrated operational decision support (IODS) systems of departure and arrival airports, airlines and national control centres. The global networking architecture of ground LODS systems will be dedicated to automated control of allocation of global air space- time and airport resources to four dimensional (4D) end-to-end trajectories of flights and to traffic flows. The ground IODS systems will respond to flight requests with creating conflict-free 4D end-to-end trajectories of flights. They will use the embedded logic of learning of clearance- categories for allocating most efficiently global resources to requested flights. Their global control mechanisms and cornnlunication processes via satellites will synchronise the use of global resources. They will trionitor and control 4D end-to-end trajectories of aircraft. They will periodically verify the current positions of aircraft on these trajectories by automated reports of actual positions of aircratl through ground positioning system (GPS) and satellite communications and will revise the projected trajectories of flights ahead of the actual position of aircraft accordingly. They will automatically update their clearance- categories organised in global dynamic AST knowledge-and-data control structures.

Key computational issues in achieving an efficient automated allocation and control of global resources are as follow. First are the embedded logic and complexity in learning of clearance- categories of 4D end-to-end trajectories flights and their global dynamically and incrementally up-datable air space-time control structures from a series of distributed flight requests. Second are the parallel and distributed processes of the global control mechanisms in maintaining these dynamic control structures and their 41) end-to-end trajectories of flights conflict-free while the traffic flows progress.

Further important consequen(e of this new technology is that the networking architecture of dedicated IODS ground systems plans the most efficient allocation of global air space-time and airport resources to 4D end-to-end trajectories of flights of aircraft. In this way it also plans global traffic flows. It monitors and secures the conflict-free use of global resources by those end-to-end trajectories and triffic flows. It keeps conflict-free use of global resources from take off to landing of aircraft. Their predicted trajectories are monitored and revised according to automated reports of verified current positions of aircraft through GPS and satellites.

The IODS systems ensure synchronjsed revisions and automated clearances of flight trajectories to avoid any conflicts or cong:stions predicted during a certain air space-time period ahead from the current verified positions of aircraft. They monitor and control 4D end-to-end trajectories through global flows of AST control structures communicated via satellites. These structures control the conflict-free allocation of global resources in clearance-categories of flights and the conflict-free updates of 4D end-to-end trajectories. The IODS systems synchronise these updates of clearance-categories of flights in the global AST control structures through satellite communications and the updates of their cleared 4D trajectories of flights through direct data communications with on-board computer of aircraft via satellites too.

Important computational and complexity issues result from the dynamic nature of allocation of global resources and of controlling their conflict-free use by 4D end-to-end trajectories of aircraft in flight. The dynamic and real-time nature of series of flights and of control of dynamic allocation of global resources increases the complexity in learning from the series of distributed requests. The computational complexity is in the constraint-satisfaction and in the efficiencyoptimisation involved in thc automated learning of clearance-categories of 4D end-to-end trajectories of flights in response to the dynamic series of requests. It is also in keeping the conflict-free use of allocated air space-time and airport resources to global traffic flows and to 4D end-to-end trajectories of aircrafi in flight too.

The dynamics of allocation nd of use of global airspace-time and airport resources result in creating and in dynamically up-dating computational images of global air space-time and airport resources allocated to 4D endto-eid trajectories of flights and to global traffic flows. These computational images are organised in flows of global air space-time knowledge-anddata control structures of clearance-categories of flights These control structures are dynamic. They move and change together with the progress of aircraft in flight towards their destination. Those flight trajectories cleared within each clearance- category are continuously monitored. Positions of aircraft are checked and verified with actual positions through GPS and satellites. Projected 4D end-to-end trajectories of flights from the current verified position of aircraft are updated accordingly and their clearance-categories in the global AST control structures too.

The computational outcomes are through distributed processes of constraint satisfaction and of optimisation and of decentralised mechanisms of control in allocating global resources in response to a series of distributed requests.

Further outcomes the technology brings are through new global mechanisms of monitoring and of controlling 4D end-to-en(1 flight trajectories and traffic flows and of keeping the AST resources conflict-free and the use of airports capacity most efficient in the long the term.

Further important issue relates to a series of dynamic events triggering the learning processes and control mechanisnis of the neural architecture of AST structures in a global network of distributed IODS systems.

IV. SERIES of EVENTS Events triggering the learninV mechanisms in the neural architecture occur when new flights from departure airports to destination airports are requested. The embedded logic of the neural architecture learns how to accommodate these requests most efficiently for the individual flights requested by airlines and for the global planning, monitoring and controlling of conflict-free use of air space- time and airport resources by 4D end-t-end trajectories of flights.

The learning mechanisms are also triggered by events such as predictions of conflicting trajectories of flights ahead of space-and-time from the actual dynamically verifiable and updateable positions of the airraft on the monitored trajectories.

The above events form a serits of distributed requests for clearances of flight trajectories. These events request conflict-free pl inning of' allocation of global air space-time and airport resources.

These requests form a population of 4D end-to-end flight trajectories which the architecture of the global neural network of AST control structures has to accommodate most efficiently within the global air space-time and airport resources available and also to keep them conflictfree from take off to landing of aircraft.

V. STATEMENT OF WORK

This work concerns the new IODS processes and technology of the new global mechanisms for monitoring and for controlling efficient allocation of conflict-free air space-time and airport resources in response to distributed requests of flights. It includes the design of the processes and the control structures embedded in global neural mechanisms for allocation of global resources and control of global traffic via satellites. It is also concerned with the logic of learning clearancecategorjes of flights requested within global networking architecture of distributed LODS systems.

The global neural network for synchronised Air Space-Time (AST) allocation and control aims to operate within a new global infrastructure for future air traffic management created by the networking architecture of JODS systems for airports, airlines and air traffic control and direct data communications with the on-hoard equipment of aircraft through satellites.

The work investigates the computational and complexity issues related to the embedded logic of learning clearancecategories of 4D end-to-end flight trajectories and of creating global AST knowledgeandda control structures, It investigates the decentralised and distributed intelligent mechanisms for monitoring nnd for controlling 4D end-to-end trajectories of flights and their automated clearances organisc,d in the dynamically up-datable AST knowledgeanddat control structures The work designs the new global processes of the JODS technology for securing conflict-fee use of global AST resources and of airports capacity. It develops also global intelligent mechanisms and their parallel and dislrihui;cd processes for continuously monitoring and controlling 4D end- to-end trajectories of aircraft in flight through satellites. The JODS technology creates AST control structures for synchionising revisions and updates of those dynamic 4D end-to-end trajectories off-line any predicted conflict along those projected trajectories ahead of the current positions of aircraft. These control structures are created by parallel and distributed learning processes automating the knowledgeicquj5j0 and reasoning (KAR) for allocating air space- time and airport resources to requested flights. The processes create clearancecategories of their 4D end-to-end trajectories Those categories and their cleared trajectories of flights are periodically revised for keeping the flights and their use of global resources conflict-free. Thus the JODS technology ensure; the continuous conflict-free use of global resources and most efficient use of air space-time and airports capacity.

A. IODS Distributed and Pa rallel Processes The JODS technology for future global air traffic management proposes a future global networking infrastructure of ground JODS systems for synchronised and automated management of global resources traffic fiDws and end-to-end trajectories of flights and knowledge. These infrastructure incorporate five management layers of components (Fig. 1) of the global networkiig architecture of LODS systems. Each of these layers forms a concentric circle. Each circle contains hierarchical components and their protocols within the global networking architecture of the IODS systems.

The outermost layer represet; the user-interface components managing sequences of distributed flight requests. The user interface of each ground JODS system communicates with the on- board equipmeiit of aircraft through satellites and with the parallel and distributed KnowledgeAcquisitioti and Reasoning (KAR) processes of the Planning of ConflictResolutions components higher up in the hierarchy (Fig. 1).

The R processes are associated with the internal components of the Planning of Conflict- Resolutions components, nanLely the Hierarchy of Objects and Constraints (HOC) components of the architecture. The HOC components and their parallel and distributed KAR processes (Fig. 2) allocate the air space-time and airport resources in response to a series of distributed flight requests and create their clearancecategorjes of 4D end-to-end trajectories. They organise these in AST knowledgeanddatt control structures. These AST control structures and their clearancecategories are revised with the progress of the traffic flows. Their revisions and updates of trajectories of flights are communicated through satellites to the on-board computer of aircraft and to other JODS sysl.en-is concerned with the updates.

The KAR processes communicate with the AST management components managing global Air Space-Time (AST) and airport resources and their parallel and distributed processes. The latter processes monitor AST kno\vledgeand control structures of clearancecategorjes of 4D end-to-end trajectories of flights during a certain air space-time period ahead of the current c7 position of aircraft on that trajectories from their departure to their destination. These structures are associated with the innermost hierarchical layer of the global architecture of IODS systems.

(Fig. 1).

The architecture of IODS systems integrates two main components and the collaboration and communications between thei-- parallel and distributed processes. They are those of management of global AST and airport resources through parallel and distributed monitoring processes (Fig. 3) of AST knowledge-anddjta control structures and those of management of Hierarchies of Objects and Constraints (H( )C) (Fig. 2) KnowledgeAcquisition and Reasoning (KAR) for Planning of Conflict Resolutions (PCR) . The collaboration between their processes (Fig. 3) results in allocating conflict- Free AST and airport resources to clearance-categories of 4D end- to-end trajectories of flights and keeping these trajectories conflict- free from take off of aircraft to their landing.

Those parallel and distributj processes (Fig. 3) automate the efficient allocation of the air space-time and airports resources according to the technical capabilities of the various aircraft and to the efficiency requirerrients of end-to-end flight trajectories and of global air spacetime and airports capacity use. They generate the decision support knowledge for efficient clearance- categories of flights (4D end-to-end trajectories) and for keeping them conflict-free and most efficient within the resources available. The IODS technology also communicate the automated clearance up-dates of those trajectories via data communications with the on-board computer of aircraft through satellites (Fig. 2).

The global networking architecture of IODS systems synchronizes the communication processes of the updates of AST knowlcdge-anddata control structures, It does this via satellites. It also synchronises and secures corn municatjons from ground 1ODS systems to the relevant terminals of the airlines, airport operitors, and air traffic controllers concerned with the updates of clearances of 4D end-to-end trajectories of flights and of their AST control structures.

The KAR processes (Fig. 2) perform automated reasoning about the most efficient design of allocation of available air space-time and airport resources to requested flights and create their clearance-categories in AST knowledge-anddata control structures of conflict-free and optirnised 4D end-to-end trajectories of flights.

The KAR processes also maintain the AST knowledge-and-data control structures, revise the clearance-categories flights and their 4D end-to-end trajectories in order to keep them conflict- Free and efficient within the dynamic availability of global resources. The HOC components (Fig. 2) create and maintain a flow of AST knowledge-and-data structures of conflict-free and optimised 4D trajectories of flights, given the available resources, through their parallel and distributed KAR processes (Fig. 4).

Each of these clearance-calegories and their AST structures is controlled by monitoring processes of the AST manarnent components (Fig. 3). They test a predicted 4D end-to-end trajectory for conflict during a certain air space-time period ahead of the current dynamically up- datable position of aircraft via automated reports through GPS and the satellites.

An individual AST monitoring process (Fig. 3) monitors an individual 4D end-to-end trajectory cleared within an AST knowldge-and-data control structure of clearance- categories of conflict- free and optimised flights. It verifies and updates the recorded current position of the aircraft on the monitored trajectory with reported real- time position of aircraft via GPS and communications through satellites. It revises the 41) trajectory and performs a trajectory prediction from the reported real position of aircraft. It performs tests for conflicts along the updated 4D end-to-end trajectory predicted ahead of the currently verified and updated real position of the aircraft.

The KAR process (Fig. 4) provides a trajectory maintenance service by automating the clearance of the predicted trajectory ahead of the reported verified real position of the aircraft. Thus it maintains the conflict-free revision of the 4D end-to-end trajectory of aircraft. The KAR process generates knowledge of conflict-free and most efficient alterations of 4D end-to-end trajectories within the current or new clearance-category. It up-dates the old and the new clearance- categories and their global AS IF knowledge-and-data control structures.

When an AST conflict is predicted during an AST period ahead the AST monitoring process (Fig. 3) communicates with the relevant KAR process (Fig.3). This process plans conflict- resolution alterations of the flight-plans and their AST structures off- line during this AST period in advance before any conflict can develop in real time. These alterations are communicated through the user-interface of the system to the terminal of the controllers, airlines and airport operators and to the on-board computer of the aircraft.

An individual KAR process (Fig. 4) learns initially concepts through parallel concept-formation processes. It generates the decision support knowledge about the most efficient clearances- categories of 4D end-to-end trajectories of flights given the resources available by exploring the alternative concepts it creates from a series of requests and also by incrementally specialising those concepts. The most efficient concepts accommodating most efficiently the series of flight requests are accepted as clearance-categories of these flights.

A clearance-category of a 1li1ht consists of hierarchical objects of an AST knowledge-and-data control structure (Fig 5) such as the air traffic route and the stream of aircraft that most efficiently accommodate the 4D end-to-end trajectory of the flight in respect of its own efficiency and the efficiency of its air space-time and airport capacity use. The KAR process (Fig.4) organises this knowledge of the most efficient clearancecategories of flights in hierarchical AST knowledge.-and-data control structures of conflict-free and optimised flights.

Fhus it automates the desigi: of 4D end-to-end trajectories of flights, the allocation and the control of global air space-time and airport resources and the planning of global traffic flows.

The hierarchical AST strategies and integrated constraints-satisfaction and efficiency- optirnisation reasoning operators guide the concept-learning at hierarchical levels of the automated acquisition ol an AST knowledge-and- data structure (Fig. 5) from series of flight requests.

B. Distributed Flight Requests We make the following distinction between two types of requests. Those of flight-plans and those of flight-paths.

We define flight-plans as 4D end-to-end trajectories of flights requested in advance or at the time of take-off of aircraft. In both cases the clearances of flight-plans(4D end-to-end trajectories) are requested from agents of the on-board computer of aircraft or of terminals of airlines prior to departure.

Flight paths we define as 4D end-to-end trajectories of aircraft in flight. The requests for conflict-resolution alterations of flight-paths are made by agents of the AST management components (Fig. 3) and their processes of monitoring the AST control structures containing the clearance-categories of the 4D end-to-end trajectories of the flightpaths (Fig. 2).

The KAR processes (Fig. 2 and 4) deal with the above two types of requests. They maintain and update incrementally the AST knowledgeanddata control structures of conflict-free and optimised 4D end-to-end trajectories of flights. Thus they also generate the decision support knowledge for sustaining the efficiency of the AST use and of the 4D endto-end trajectories of flights in the long term.

C. Global Networking Infrastructure The IODS technology, processes and control mechanisms provide intelligent, interactive and integrated (la) operational decision support (ODS) to pilots via direct data link communications though satellites with on-board computer of aircraft and to controllers, airlines and airport operators via synchronized and secure communications within the global I3ODS infrastructure of the networking architectures of dedicated ground IODS systems.

The global I3ODS infrastructire of networking LODS systems communicates to each other AST know1edgeand control structures via satellites. Thus they control the conflict-free use of the global resources and secure the conflict-free planning for global traffic. They monitor and control air space-time and airport capacity use and global traffic flows through satellites.

They accomplish the latter by the integrated management of automated monitoring and automated control of allocation of global air space-time and airport resources. They secure global planning of confiict-resolutins of 41) end-to-end trajectories of aircraft off-line before any conflict can develop in real-time. The collaboration between IODS parallel AST and KAR processes (Fig. 3) and their secure communications provides the knowledge for the global planning of the conflict-free Lfld efficient use of AST and airports capacity. The IODS systems secure most efficient clearances of 4D end-to-end trajectories of flights and traffic flows according to the global resources available. They also keep the 4D end-toend trajectories conflict-free and most efficient in the long term. This they accomplish by the planning of conflict-resolutions off-line during a certain AST period in advance before signs of AST conflict can develop in real-time.

The satellite communications of the global I3ODS infrastructure of networking architectures of dedicate ground JODS systems and the on-board computers of aircraft will deliver the synchronised and automated updates of conflict-free 4D end-to-end trajectories of flights to the pilots via the direct data link between IODS systems and the on-board computer of aircraft.

Those automated updates they will deliver to the relevant terminals of airlines, controllers and 17) airport operators through satellite communications between the ground JODS systems. Thus they will keep the use of global resources by the 4D end-to-end trajectories of flights conflict-free.

Through their global AST control structures of clearance-categories they will allocate traffic flows according to global resources available too.

The JODS technology brings major innovations for meeting the challenges of today and of tomorrow.

The IODS automated parallel and distributed design processes and mechanisms of planning of allocation of global air space-time and airport resources to 4D end-to-end trajectories of flights together with the IODS automated monitoring and controlling mechanisms of their conflict-free and efficient use of global resources produce the technical effect of global automation of allocation and of control of global resources and of traffic flows.

D. Agents, Communications, Interactions and Services The JODS components and their parallel and distributed processes communicate to each other through agents managing the processes and their communications. The agent-managers of the user interface components of JODS systems manage

the series of flight requests. Their interface agents secure communications with the on-board computer of aircraft via direct data link through satellites and the terminal of the controllers, airlines and airport operators via synchronised and secure links. The agent-managers prioritise flight-plans requested for clearance, and communicate with the agents managing the AT Hierarchies of Objects and Constraints components and their parallel KAR processes. Each individual KAR process generates the Air Spae-Tirne knowledge-and-data structure of clearances of flights (4D end-to-end trajectories) requested during a certain AST period. The result of all parallel KAR processes is a flow of AST structures of conflict-free and optimised flights (4D end-to-end trajectories) given the resources available.

l'hc agent-managers of the 1-IOC components attend to the management of the parallel KAR Processes, and of communications between agents managing the individual KAR processes. He also attends to the management of the communications between the agents managing the individual KAR processes and the agents monitoring the AST control structures.

The agents managing the AST Management components also manage their parallel AST monitoring processes, and the communications with the agents managing each individual process. The agent also secLires the communications between its agents managing the AST Monitoring processes and the Igents managing KAR processes of the HOC component.

An agent managing an monitoring process communicate with an agent managing a KAR process.

These communications inclucLe the request for the maintenance service asked for by the agent managing the AST monitorin; process. This is provided by the agent managing the KAR process of the HOC component.

An agent managing a KAR process also communicate with the agents of the user interface components. These communiations include the request for the flight clearance service asked for to - by the interface agent on behalf of an user. This is provided by the agent managing the KAR process.

The agent managing the KA R process plans 4D end-to-end trajectory of a flight, clears and secures up-dates of the trajectory and its old and/or new clearancecategorjes and of their AST control structures containing these clearance-categories.

An agentmanager of an individual KAR process (Fig. 4) deals with the above two types of requests by dynamically creating interactive groups of agents and their parallel concept- formation processes. These gents create possible disjunctive and overlapping concepts of categories of clearances of dynamically selected samples of requests, while collaborating in a search for the most efficient clearances of the series of requested flights within the available global air space-time and airj:ort resources. They create and update dynamically categories and their fuzzy conceptual descriptions of objects from parameter-values of 4D end-to-end trajectories associated with them. They evaluate the measures of efficiency of these categories.

They order them by the efficiency of individual trajectory and of their AST use. The concepts which predict the clearances of sets of 4D end-toend trajectories with the best measures of efficiency are accepted as updates of the global AST knowledgeandda1a control structures.

The agentmanagers of the KAR processes secure the synchronised updates of the hierarchical AST conirol structures in response to clearance requests.

The aircraft agents or the airline agents who have requested flight clearance have the capacity to influence the decision of an agent managing the relevant KAR process, by providing him with certain efficiency parameters logether with their requests through the interface agents. The agent managing the KAR process tikes into account efficiency measures and functions for evaluating the possible clearancecateories according to the requested efficiency and meanwhile communicates them in order of priority to the agents requesting them. The latter agents can make their final choice according to these objectives.

E. Global Neural Control Mechanisms The global neural network mechanisms for synchronised air space-time allocation and control are intended to be embedded in a future global infrastructure of IODS systems. Their building blocks are those parallel and distributed processes (Fig. 3) in the global network of IODS systems. Their embedded learning logic derives knowledge from the series of dynamic and distributed events described in section 3. The global distributed neural architecture embedded in IODS systems generates and maintains flows of AST control structures of clearancecategories of 4D end-to-end trajectories of flights.

Air Space-Time knowledgeand..d control structures are created and used by neural control mechanisms in the global networking architecture of LODS systems. Through the embedded logic of dynamic neural mechanisms and processes the JODS systems learn knowledge from a series of distributed requests. They create and monitor clearance-categories of 4D end-to-end trajectories of flights organisd in global and dynamically up-datable AST control structures.

Through those categories they monitors and controls the global air spacetime and airports resources, traffic flows and tJjectories of flights. They control and update those categories and their global AST control structures incrementally and synchronise those updates of control li structures and of trajectories through direct data communications with onboard computers of aircraft through satellites. Thus they keep the use of global resources conflict-free while the aircraft progress along their cleared trajectories projected towards their end destinations.

The size and the shape of each of those flight clearance-categories and the neural architecture of AST control structures as a whole depends on the series of dynamic events of flight clearances requested in regions in global iir space and time.

Each AST control structure contains hierarchical layers of objects representing learning "neurons' and chains of objects representing the clearance-categories. The "neurons" explore the dynamic series of requests serching for relevant parameter values and matching the requested efficiency of flights within c sting or new clearancecategories They learn clearance- categories of flights and the knowledge about the efficiency with which these categories clear individual requested flights. The layers of neurons of categories encapsulate priorities of satisfying constraints in forming concepts of clearances of flights, and in validating decisions about allocating AST resources.

Layers of neurons satisfy constraints in generating parameters of clearances of 4D end-to-end trajectories of flights. Chain of neurons hierarchically inherit and pass down parameters of clearance-categories of 4D cnd-to-end trajectories of flights. Thus they generate validating proofs of clearanceparaineters for those 4D end-to-end trajectories of cleared flights within individual clearancecategories They prioritise those categories by efficiency they supply for flights and for the global AST and airport capacity use.

Major outcomes from the wrk are those [ODS parallel and distributed processes and their embedded logic and global control mechanisms in learning clearance-categories from distributed requests in a global neural architecture of AST knowledge-and.data control structures for global air space-time and airport resources allocation and control.

The work investigates the con1putatioial logic of the JODS parallel and distributed processes of learning concepts of clearances of 4D end-to-end trajectories of flights and of verifying their clearancecategories within a global neural architecture. The concepts are formed based on sampling of measurements of parameters of requested flights selected dynamically from the series of requests in regions of global air space and time. Through this logic and control mechanisms the IODS parallel and distributed processes plan and allocate AST and airport resources to 4D end-to-end trajectories of flights, clear these within verified clearance- categories and monitor and control these categories and their end-to-end flight trajectories.

The concepts of clearancecategorjes are established based on an estimation of the central tendencies of measurements of parameters of flights. The learning logic of parallel processes aid of their control mechanisms measures the variance and central tendencies of dynamic samples of the series of requests and creates and updates concepts of categories dynamically and incrementally Through this logic and control mechanisms the parallel processes learn knowledge from the parameters of measurements taken from the dynamic samples of requests and use it as an a priory knowledge in dynamically specializing clearance-categories and in updating their AST control structures. The categories are measured by the efficiency they secure for 4D end-to-end trajectories of cleared flights and the efficiency of their use of global air

IL

space-time and airport resourL:eS. The dynamic specialisation of categories reflects with a higher accuracy the central preferences of the series of requests and improves the efficiency of using global AST and airports capacity.

Further work regards the complexity analysis of the new global control mechanisms of the IODS technology. It regards the computational complexity of the learning logic in a global neural networking architecture of AST structures and their global mechanisms of allocation of AST and of airport resources and of control of global traffic flows though satellite communications too.

The work extends the previous work on how hierarchies of objects and constraints reduce complexity. The design of AST structures and how it reduces the complexity of learning clearancecategorjes are puhli:hed in IJCNN 2003. It is included in patent application regarding new processes and global ccntrol mechanisms for automation of global AST allocation and control of traffic through satclites.

The work includes the embed(led logic of learning dynamic flows of AST control structures from distributed series of flight requests and of controlling global air space-time and airport resources.

It investigates the complexity in planning and in verifying conflict-free allocation of global dynamic AST and airport resources to 4D end-to-end trajectories of flights in response to series of distributed requests.

In the context of a global architecture the work considers the complexity in creating and maintaining dynamic flows of AST control structures of clearance-categories accommodating distributed dynamic series of flight requests within the global AST and airports resources available. The work also considers the conîplexity of processes of monitoring and of controlling traffic flows, four dimensjonl trajectories of flights, and of planning and of verifying conflict- free use of AST and airport resources via IODS technology and the direct data communications between ground LODS systems and on-board equipment of aircraft through satellites.

Further papers will be published regarding the embedded logic and the reductions of complexity of global control mechanisms in allocation of' air space-time and airports resources in response to distributed series of requests in the global neural architecture of AST control structures.

VI. SCIENTIFIC CONTRIBUTIONS.

The research paper presented here is an extension to an entirely original peace of work on a new IODS technology for management of traffic, knowledge and global resources via satellites. It presents a new global neural networking architecture for learning, monitoring and controlling 4D end-to-end trajectories and their clearances organised in global AST control structures. It is a further development of the work on a new air traffic knowledge management and conflict-free planning for global air traffic by Integrated Operational Decision Support Systems automating the allocation of air space-tin-e and airport resources and the control of global traffic flows via neural network of AST control structures arid communications through satellites. The IODS intelligent processes and global decentralised control mechanisms learn, monitor and control clearance-categories of 4D end-to-end trajectories of flights from series of dynamic requests and provide integrated operational decision support to airport and airlines operators, pilots and controllers. They communjcat with the on-board computer of aircraft in flight through satellites and they control the 4D end-to-end trajectories of aircraft and the global traffic flows too.

The work makes major contributioi in a highly specialised domain and in an interdisciplinary area of intelligent and complex systems, machine learning, neural networks, systems design and integration, automation and c)ntrol, computational analyses and complexity reductions, parallel and distributed architectures kind processes networks and satellite communications systems and operational research.

It makes a significant contrihjtion in the areas of intelligent systems and complex systems and decisioi support. It contributes the design of a new global networking architecture of IODS systems and their parallel and distributed processes and intelligent global control mechanisms for knowledge, traffic and resources managemeit and control in a future global air traffic management system. It contributes a new IODS embedded logic and mechanisms of learning automated clearances of 4D end-to-end trajectories from distributed clearance requests. It contributes a global IODS architecture of a neural network of AST control structures for the allocation of conflict-free AST and airport resources to 4D end-to-end trajectories. The architecture controls confljct-jiee use of global resources by those trajectories and global traffic flows though flows of global AST control structures and their communications through satellites.

The work contributes in the area ol learning and complexity analyses and reductions. In particular regarding the complexity of learning clearaneecategorjes from a distributed series of flight requests. The further ontrjhutjons concern the computational complexity analysis of parallel and distributed learning processes and intelligent control mechanisms in a global neural network of AST control struci ures. These will be published in further papers.

VII. BROADER IMPACT.

The work makes major contribution to operational research regarding future global air traffic management system. It desins the new global processes and control mechanisms of the integrated operational deeisi(:11 support technology for aerospace and air traffic management systems and for control of allocation and of conflict-free use of global air space- time and airport resources.

The work makes contribution to engineering of integration and synchronization of systems and global operations through integrated, operational decision support for airports, airlines, pilots and controllers It puts forward the conflict-free planning policy for global air traffic. Its puts forward the automation of allocation of global resources and of control of global traffic via 4D end-to- end trajectories of flights through IODS technology and communications through satellites. lit

Claims (3)

1. Claims: 1) 1) New intelligent (neural) mechanisms and their distributed

   and parallel processes that a) monitor, allocate and control resources.

   b) plan, monitor and keep clear paths (4D end-to-end trajectories) without interventions (exclude controllers from the loop).

   c) communicate with users: pilots, controllers, airlines and airports operators simultaneously and directly.

   d) establish simultaneous and direct new channels of communications between V. integrated ground systems, VI. integrated ground systems and on-board equipment of aircraft, VII. integrated ground systems and terminals of airports, VIII. integrated ground systems and terminals of airlines.

   e) allocate resources, in response to requests and control their conflictfree use.

   0 generate decision support and control knowledge and communicate these to all concerned systems/users simultaneously and directly.

   g) synchronise global operations.

   h) optimise the use of air space-time resources and airports capacity.

   i) plan and control global traffic flows according to resources available.

   j) plan conflict-free use of resources off-line versus conflictresolutions in real-time.
2. 2) New intelligent (neural) mechanisms and their distributed and parallel processes according claim 1 embedded in new global control mechanisms for k) avoiding congestions and so called boftlenecks in the current practice caused by sectors with low capacity.

   I) obviating the need of division of airspace in sectors.

   m) planning conflict-free 4D end-to-end trajectories and traffic flows between departure and destination airports.

   n) keeping the use of global air space-time conflict-free and of airports capacity efficiently used without any congestions..
3. 3) New intelligent (neural) mechanisms and distributed and parallel processes according claim 1 implementing global control mechanisms according claim 2 embedded in future global networking infrastructure of new integrated ground systems for global air traffic management and control.