modeling of complex systems and environments july 17th, 2007 agenda introduction of the laisc the...
TRANSCRIPT
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Applying Negotiation Analysis & Pretopological Concepts to study Cockpit’s
Perspective to Collaborative Decision Making
Complex Dynamics Simulation for Air Traffic Management
Matthias Groppe - Marc Bui
Laboratoire d’Informatique et des Systèmes Complexes Ecole Pratique des Hautes Etudes, Paris-Sorbonne
July 17th, 2007
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction to the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
■ Introduction to the LaISC
■ The Science of Complex Systems
■ Decision Making in Complex Systems – The Collaborative Decision Making Project
■ Agent Modeling
■ Conclusion
AGENDA
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction to the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
The LaISC
The Laboratoire d’Informatique et des Systèmes Complexes of l’Ecole Pratique des Etudes, Paris- La Sorbonne
The area of research is concerned with the theory and modeling of adaptive complex systems, especially networks without centralized control
Some subjects of research of the laboratory
Air Traffic ControlDistributed Decision Making Complexity
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction to the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Research Objectives of the LaISC
Conceive methods and theoretical means to study adaptive complex systems without having centralized control
To study emergent complex behavior
To study the results of complexity and complex systems
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction to the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Air Transport Management Systems: Scientific Approach
Based on lateral approach to the definition of
Availability & Complexity
- Availability of resources
- Number of Interactivities
Simulation via Multi-Agent Systems
Agents as entities that can make decisions based on data available
Modeled as rule-based expert systems in a hierarchical organization to facilitate interactivities like negotiation or information sharing.
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
The Science of Complex Systems
One research thread is, how self-organization emerge through the understanding of de-centralized control mechanism.
E.g. behavior of the flight of a group of birds
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
The Science of Complex Systems (2)
Another important research thread is an understanding of networks and how their topology impacts the properties of the system.
Network of small worldsExamples
Internet pagesSocial networksAir routes Collaborative Decision Making
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
General Definition of a Complex System
A complex system is a network composed of mutually interacting elements, where the global behavior of the system can not be deduced from the sum of its components and their properties.
Santa Fe Institute http://www.santafe.edu
New England Complex Systems Institute http://necsi.org
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Properties of Complex Systems
Emergence Phenomenon
Appearance of new property in the system
Phase Transition
Brutal change in the state of the system
Transition Threshold
Critical value or key parameter
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
•Airlines‘ Schedule
•Planning
Information
•Flight Progress
Information
•Predictions
messages
•Status messages
•Operational
Planning
Information
Airport
Decision
Making
Database
Collaboration
Source: Eurocontrol (2004)
Example of Emergence in ATM Decision Making
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Decision Making in ATM Systems: A real complex thing
The Air Traffic in Europe
It‘s control is executed by Eurocontrol (36 countries)
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Air Traffic Management: some numbers
30 000 flights per day
Average distance of 750 km or 1h 30 minutes
15.000 controllers (55.000 personnel in total)
600 sectors
Cost of ATM in Europe: 7 Billion Euros/ Year
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
ATM at Airports: The Capacity Crunch
E.g. Frankfurt International Airport:
1980: 220.000 flights & 18mil. passengers/ year
2000: 560.000 flights & 50mil. passengers/ year
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Fighting the Crunch: Introduction of Collaborative Decision Making (CDM)
Information sharing & coordination among all actors involved
Increasing of Common Situational Awareness
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
The Collaborative Decision Making Concept (simplified)
Working together at operational level of Aircraft Operators, Ground Handling Agents, Airport, ATC & CFMU
From ad-hoc culture to global collaboration in planning & air traffic management
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Procedure of Collaborative Decision Making
Information Management (Airport, Security, Ground Handlers, ATC,
Central Flow Management Unit, Airlines)
A-CDM PlatformA-CDM Platform
Message Manager
Airline
ATC
DMDBDecision Database
Airport
Community
Ground Handlin
g
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
ATM at Turnaround: The Pilot’s Perspective
Information Sharing & Coordination during Turnaround of an Airbus 330
40,5
12,0
13,4
13,7
19,1
5,5
0 10 20 30 40 50 60
minutes after aircraft at position
Deboarding
Catering
Cleaning
BoardingLoading
Fueling
Turnaround
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Pilot’s Perspective during Flight
Critical for Common Situational Awareness & Information Sharing: From Off-Block to On-Block
AO
BT
From Off-Block to On-Block!!
AIB
T
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Proposal: Adding the Pilot’s Perspective to CDM
AIM
Increasing Common Situational Awareness & Information Sharing/Cooperation by Studying Cockpit’s Perspective to CDM via:
Application of a methodological approach for problems encountered with Cockpit’s perspective to Collaborative Decision Making & Information Sharing/Cooperation
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Analogy for Information Sharing ‘en-route’: The Moving Cogwheels
Each tooth represents an essential information
Deadlock of one wheel means inactivity in proactive sharing of essential information
Moving of all wheels necessary for ATM operation
Each single missing essential information may stall turning of the wheels or reduce wheel speed
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Complexity of Collaborative Decision Making: The Subsystem ‘en-route’
There are a number of problems:
1. The amount of essential information needs to be determined, because Collaborative Decision Making includes routine essential and non-routine essential information
2. The difference in the way, how decisions are made by the pilots & controllers in similar situations has to be understood in order to increase cooperation.
3. A local optimal decision making leads not necessarily in a global optimum.
4. Other uncertainty factors (e.g. weather, technical problems….)
Problem
Which is the efficiency limit for a local optimization?
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Aspects of Complexity
(1) Structural Complexity
qualitative characteristics and topology of ATM
systems
(2) Complexity through System Dynamic
temporal effects of its behavior & the characteristics of the system
Interaction between (1) and (2)
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Applying Negotiation Analysis
(1) Structural Complexity
Individuals’ perspective towards decision making and selection criteria are determined via questionnaires & semi-structured interviews
NA combines individual & interactive perspective towards decision making
NA adds third perspective: joint decision making
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Applying Pretopological Concepts
Finding groups of interdependent elements (e.g. information-action)
Highlight homogenous groups/ elements (minimal or elementary closed subset)
Highlight groups containing the homogenous groups/elements (non-minimal elementary closed subsets or function MinimalClosedSubsets)
Structural Analysis: aims to find inclusion relation between elements
Aim: Aligning of all actors in a common performance framework
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Modeling of Agents
Agents, such as aircraft, airlines, and air traffic control are defined, which represent the behavior of the actor.Further there are several event classes defined to describe the behavior of the agentsEach event class has three options:
Action is accepted (agent is e.g. providing information/ instruction), rejected, or alternative is proposedEach actor has a different type of assessing the options (e.g. satisfying or maximizing behavior)
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Modeling of Agents (2)
Denote A = {ai; i = 1,…,N }, where a is one of N agents The ATM system is expressed as: s(t) = f(s(t), c(t), e(t), t), where s(t) is the state, and control c(t) includes different variables (discrete, binary, mixed), e(t) denotes the environmental variables (wind, speeds etc.) at a the time t. We further write for the sum of actions of A:Ζ =(a, u, T, p), where u is a sequence of actions that an M-member subset of agents A takes over at a given time: u AM
(t1,t2) = {uam (t1,t2): m = 1,
…..,M}; T = T1x…xTn as the set of all possible agent types; p: T→[0,1], where p(i,t) is the probability that agent i is of type t.
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
Conclusion
Our approach to model Air Transport Management Systems allows:
To simulate decision making situations in variance of availability & complexity.
To identify the emergence phenomenon of collaboration.
To include a behavioral & cognitive perspective to decision making in complex systems.
Modeling of Complex Systems and Environments July 17th, 2007
Agenda
Introduction of the LaISC
The science of Complex Systems
Decision Making in Complex Systems – The CDM Project
Agent Modeling
Conclusion
THANK YOU