The New Model for Air Traffic Control

The Operational Concept

Brighton and Hove Municipal Airport

The paper presented at the 50th United States Air Traffic Control Association Conference, held at the Gaylord Texan Hotel and Conference Centre, Grapevine, Texas, 31 October to 2nd November 2005 can be read or downloaded. The paper is in Microsoft Word 2003 format. The font is Times New Roman 11 point.

ATCA Paper November 2005 (84 KB)

It is an old maxim of mine that when you have excluded the impossible whatever remains, however improbable, must be the truth.

Sherlock Holmes, The Beryl Coronet, Sir Arthur Conan Doyle

The Current Operational Concept

Many people will be familiar with the general principles of Air Traffic Control using radar. Each Air Traffic Controller has a surveillance display showing in the form of a map the airspace for which he (or she) is responsible. The position of each flight is indicated and usually against each such marked flight is the callsign and the flight level. Some systems may display other data such as the destination airport.

Using skill and judgement gained over about three years in training the controller visualises the whole situation in three dimensions, plans suitable trajectories for each of the flights and then guides each of the flights through the airspace by means of voice instructions to the pilot over a radio link.

The whole airspace is divided into sectors and as each aircraft approaches the boundary of a sector the pilot is instructed to call the next controller on a new radio frequency. This process is repeated many times for each flight as it progresses from take-off to final touchdown.

The Germans, by contrast did not have the same close relationship between their serving officers and their scientists. When radar became a technical possibility, and this was realised at least as early in Germany as in Britain, the German services drew up specifications which the scientists and engineers then tried to satisfy. And very well they did so, within the limits imposed by the specifications. German radar was much better engineered than ours; it was much more like a scientific instrument in stability and precision of performance. The philosophy of using it, however, seemed to have been left to the German Services, and the Luftwaffe in particular made a philosophical mistake by focussing on the wrong objective.

R V Jones, Most Secret War, British Scientific Intelligence 1939-1945

Before take-off each flight notifies the regions it will be flying through of its intentions by means of an electronically communicated flight plan. The flight plan details, held on computer, are used to produce a paper flight progress strip for each controller whose airspace the flight is expected to pass through. The strips, measuring about one inch by eight inches are placed in plastic holders and held on a board in front of the controller. When a flight is under active control in each sector the responsible controller makes an accurate note on the strip of every instruction issued to the pilot.

The most important requirement in executing the control task is to ensure safety by providing a minimum separation between any two flights. Typical minimum separation standards are 5 nautical miles horizontally and 1000 feet vertically. The controller should then also ensure an orderly and expeditious flow of traffic.

A fuller description can be found on the Indian Air Traffic Controller’s site:

Air Traffic Controller’s Guild (India)

Limitations of the Current Concept

The manual approach just described has worked very well throughout the world for over half a century. Unfortunately, this approach also has a number of shortcomings as traffic levels increase and the search for a safe and effective means of automation has been under way for almost the same length of time.

It is expensive and difficult to train a controller. Up to 70 per cent of the people who start training will not pass all of the tests necessary to become a fully fledged controller at a major Air Traffic Control Centre (ATCC). ATCC buildings are extremely expensive to construct and equip. Controllers can only handle a few aircraft at a time before they become overloaded and liable to make mistakes. The process of rearranging the airspace to enable the employment of as many controllers as possible at a time has now certainly run its course. The Eurocontrol Central Flow Management Unit (CFMU) in Brussels now has to allocate take-off slots to flights in Europe to ensure that demand does not exceed capacity.

You should also read:

Tom Lusch’s Radar Investigations

The Search for an Automated Concept

Almost all authorities now acknowledge that the way forward lies in the direction of automation and a considerable amount of research has focussed on ways of providing more or better information to the controller. Often, the information that is considered to be most important relates to any predicted losses of separation (known as conflicts) between flights. Extensive use has been made of modern computer graphics and windowing systems to enable the controller to recognise such conflicts and then enable him (or her) to adjust the planned flight trajectories to eliminate them.

Solving Collision Course Problems

A solution to air traffic jams and a method of drastically reducing the possibility of mid-air collisions through use of the fastest computer system ever developed has been developed by Goodyear Aerospace Corporation of Akron, Ohio. The Staran 4 system, as it is called, can perform more than 40 million mathematical operations per second in predicting which aircraft are on collision courses and determining evasive action.

Flight International, 27th August 1970

Boeing’s Views on Advanced Avionics

Mr John E Steiner, Boeing vice president, technology and new programme development, told the British Guild of Air Pilots and Air Navigators in London recently that he had ‘worked up quite a head of steam’ in favour of a computer based four dimensional (three dimensions in space, plus time) air traffic control environment and aircraft equipped with advanced guidance and control systems. He said such a combination of ground based and airborne equipment was necessary to avoid acute traffic congestion in the 1980s.

Flight International, 7th November 1974

Coping with the ATC Workload

Computerised air traffic control is now firmly established and accepted by its operators. The new systems have however introduced scores of acronyms and led to notable changes in ATC techniques, leaving some users in fear of what is to come. In fact, the technical encroachment is probably only just under way, and the biggest advantages of automation have yet to be seen in the UK.

Flight International, 11th December 1976

Forthcoming Transition for Europe?

The computers, knowing all aircraft flight plans, current positions, heights and speeds by interrogating the FMS would be able to calculate the best four-dimensional conflict-free clearance for each aircraft as far ahead as possible and offer it on-screen to the controller for his approval. Having checked it, the controller could up-link it to the pilot’s FMS screen and draw his attention to it by normal VHF link.

Flight International, 6th -12th November 1991

In so far as controller productivity can be measured these approaches tend to yield improvements of no more than ten or twenty percent relative to any comparable manual baseline and no single coherent approach has yet emerged.

Almost all approaches require significant further controller training to interpret and handle the user interface and are vulnerable to system failure under high workload conditions when the controller can least afford to loose the support of the computer tools he is using.

It seems that any attempt to predict air traffic situations and display them to the controller is doomed to failure because of the huge number of variables in four dimensions and the impossibility of the controller easily making his thoughts known to the computer through any known Human Computer Interface (HCI).

An Alternative Approach

A key problem with all conventional approaches seems to be that the computer network supporting the controllers could only ever be as good as the data made available to it. The network will always be ‘behind’ the controller and will never really ‘know’ what is going on. The network will consequently sometimes be completely wrong. If, during a busy period, a controller falls even slightly behind with data entry the resulting unreliability of the system outputs will usually ensure that the controller becomes completely overloaded and that the man-machine combination will therefore become ineffective.

Any yet, fundamentally, all of the information available to the controller has already been made available to the computer network through the Flight Data Processing and the Radar Data Processing Systems. It is just not used in the most productive way.

The starting axiom in the New Model is therefore that the computer system should itself construct the most probable complete tactical solution to any given air traffic situation and offer it to the controller for his (or her) agreement. The computer system is then able to anticipate the events that may occur rather than having to simply react to inputs after a delay. The data held by the computer must also be considered to be correct until the computer is told otherwise by a user. In system and database terms the computer is said to be a CANONICAL reference. This data philosophy has been well proven in many different applications including some for ATC.

Two other simple axioms make up the complete New Model Concept. One, the system must continuously update its advice to the tactical controller to reflect the most up to date data available and two, the interface to the controller must use standard RT (radiotelephony) phraseology.

If you want to become an Air Traffic Controller you must be able to handle with lots of stress and you are able to handle with an enormous responsibility. Further you need a certificate that is the equivalent of A-levels and you must be very authoritarian. When you have those abilities and you’ve come through the pre-selections, you have a chance to become Air Traffic Controller if you can prove yourself at the training during 3 to 5 years.

E J G Spaan, The Netherlands, via the internet

Education is essential in getting a high-paying job. In fact, all but two of the 50 highest paying occupations require a college degree. Air traffic controllers and nuclear power reactor operators are the only occupations of the 50 highest paying that do not require a college degree.

Tomorrow’s Jobs, Bulletin 2540-1, February 2002, US Department of Labor, Bureau of Labor Statistics

In the most basic implementation of the New Model Concept the only addition to the conventional provision of a surveillance display, a flight data display and a support information display is therefore a window showing a list of the RT instructions recommended by the system. Each instruction has a timer counting down to the moment when it should be issued and the list is ordered according to this timer from soonest to least soon.

RT Advisory Form Example

The RT Advisory List Window from The New Model Demonstrator

In this example FLIGHT08 should be turned (Direct or DCT) to the Midhurst VOR and FLIGHT02 should be cleared to descend to Flight Level 80 as soon as possible. Handover of FLIGHT09 from the previous sector can be expected in 15 seconds and FLIGHT07 should be cleared to descend to Flight Level 70, also in 15 seconds from now. FLIGHT04 should be turned to BOGNA and allowed to climb to Flight Level 170 in 45 seconds time. In one and a half minutes FLIGHT10 can be expected in the sector and FLIGHT07 should be turned to Goodwood VOR. Such instructions are always advisory. They are there to help the controller and never demand his (or her) compliance. The A button can be clicked if the controller accepts the instruction or the M button can be clicked if the controller wishes to modify the instruction. Modification could be done within the same form but for simplicity in the demonstrator the text is loaded into a separate form for editing. This form is shown below.

The Advisory Editing Form

The Edit Advisory Window from The New Model Demonstrator

It should not be surprising that the concept of advisory information has been investigated before. Eurocontrol and the NLR (Dutch National Aerospace Laboratory) have prototyped the approach in the CORA-2 (Conflict Resolution Assistant Tool) and found it to be fundamentally workable. The following presentation was given in November 2001 at the IFATCA Workshop by Seppo Kauppinen.

The Fundamentals of an Advisory Concept in CORA-2 (PDF 556 KB)

However, the New Model Concept considerably consolidates the work done here and answers the explicit and implicit questions posed in this presentation.

A Change of Perspective

The website for the Indian Guild of Air Traffic Controllers declares:

The primary objective of Air Traffic Control is to prevent collisions between aircraft.

The New Model considers the avoidance of collisions to be a consequence of a much more fundamental objective:

The primary objective of Air Traffic Control is to allocate resources safely and efficiently.

It should be apparent that when airspace and runways, together with any appropriate buffer zones in time and space, are seen as resources that this objective will always result in the avoidance rather than the prevention of collisions. This view of the primary objective is part of the key to understanding how an automated support tool must work.

The consequences of these relatively innocuous changes of stance in terms of conception and perspective turn out to be astonishingly far reaching. These consequences are discussed shortly but first it may be helpful to consider two systems that already embody the espoused concepts.

Vehicle Navigation by Satellite

The simplest system that illustrates the proposed principle is the satellite navigation system fitted to some cars. The user (the car driver) has only to enter the address or co-ordinates of the destination for the system to start working. The system continuously determines where it is and calculates the best route to the destination using an internal database of roads. The system then displays a small map of the area around the car and advises the driver of all the necessary turns as they become due. Of note is the fact that although the system must calculate the whole route in detail only the next one or two instructions need to be given to the driver.

The car driver can accept the advice simply by making the manoeuvre and this allows the system (via the satellite positioning system) to detect that the advice has been followed. The system can then update the list of instructions and move on to the next one. Many systems include a voice synthesis system and can be programmed with preferences or constraints before or during a journey.

The car driver may also miss an advised turn by mistake or choose to ignore the advice by following an alternative route. In these cases, without any input from the driver, the system will calculate a revised set of instructions. Sometimes, it may be necessary to recalculate the whole route and the final destination may be approached along a different road.

United Kindom Network Rail Signalling Automation

Railway Semaphore Signals

Semaphore Signals Still In Use At Shrewsbury In 2003

A Modern Railway Signal Box

A Modern Air Traffic Control Centre Could Look Like This Signal Box

Much of the United Kingdom’s railway network is already automatically controlled from around ten Integrated Electronic Control Centres (IECCs). Within each of these centres a small number of signalmen (typically between 3 and 10) monitor several hundred miles of track. Signalling actions are determined by the Automatic Route Setting (ARS) system and sent to a triplicated microprocessor based system called the Solid State Interlocking (SSI) which controls the points and signals. The stated aim of these systems is to enable the signalman to do his job better as the more tedious and repetitive tasks have been automated. ARS operates with the publicly declared objective of achieving ‘minimum overall delay’ and the algorithms are arranged to measure and support this. Conflicts are taken in principle one at a time at intersections and there is no mechanism to optimise all train paths together. Visitors to the UK National Railway Museum at York can observe the live operation of signals by the adjacent centre.

National Railway Museum IECC

The Single European Sky and SESAR

The New Model Concept was conceived to facilitate the Eurocontrol/European Commission Single European Sky (SES) initiative. The New Model will be found to be an enabling factor. The New Model Concept then goes further and envisages a World-Wide Interoperating Network for Aviation. The New Model could be an enabling technology for the new (November 2005) jointly sponsored Single European Sky ATM Research (SESAR) programme.

The Single European Sky

The European Commission and Eurocontrol SESAR Project

Flying Trains

Several railway signalmen were being shown around an Air Traffic Control Centre. During the obligatory presentation the controller remarked that Air Traffic Control would have to be considered much more demanding than train signalling as the aircraft could not be brought to a halt in the sky. ‘Ah!’ said one of the railwaymen, ‘That may be true, but we can’t jump one train over another like you can!’