Distributed decision support system for airport ground handling management using WSN and MAS☆
Introduction
The world economic crisis has forced airlines to modify the way in which they are operating: improving productivity and reducing costs. Modern airports can be extremely complex organizations, which often involve governmental organizations, private companies, airlines, aircraft operations and airport operators (Ashford et al., 1998). New ideas like common use are emphasizing cost reduction to improve productivity. The concept of “Common Use Model” indicates that airport operators can gain centralized control over facilities and technologies, increase passenger-processing options and acquire shared use efficiencies as they move from exclusive use toward common use. Airport common usable space is defined as the space in which any airline may operate and as space that is not specifically dedicated to any single airline. In this model, all airline usable airport space is available to be used by any airline. The goal of the full “Common Use Model” is to minimize the amount of time any given airline resource is not in use, as well as to maximize the full use of the airport (Belliotti, 2008). Airports benefit from increased utilization of existing resources. In a full common use airport, airlines are assigned with no preferences given to any individual airline, similar to the air traffic control process. To manage resources properly, computer software and systems are put in place to perform complex calculations, monitor usage and provide status reporting.
In the “Common Use Model”, the ground handling (GH) companies are the key point because of their central role. The GH companies have to provide service to different types of incoming flights, airlines, passenger, cargo services, etc., while working in a dynamic and safety restricted shared environment. These companies need to reduce costs and offer accurate results to the service contract with the airlines to avoid been charged for delays. In most of cases, GH companies are directly hired by airlines in a rigid contract to assist with supplies. The way to reduce costs is to optimize the resources attending the incoming flights. The optimization is focused on the reduction of the number of resources and the time to assist a flight to increase productivity.
New control systems are steadily increasing their complexity due to pressing requirements for productivity and flexibility. This complexity becomes especially poignant when dealing with typically big plants, airports or centers for logistics; where the usual control is too rigid and hard to manage, maintain and update (Thorne et al., 2007). This control is based on continual planning algorithms (DesJardins et al., 1999), which presents three specific information deficiencies in distributed shop floor operations (Brenner and Nebel, 2009). Three requirements that are especially relevant to provide decision support in high volatile industrial environments are:
- •
The usage of the resources: it is not easy to know when a resource is available or busy. The information feedback of the state of the resources is complex with a high dependency on the surrounding environment.
- •
The expected duration of the processes: considered as a single task each process has different lengths depending on the external and internal conditions. The process cannot be completely defined as a mathematical static function or by a template as the real situation is in constant change.
- •
The coordination of the involved elements in a task or process: it is not easy to define the rules of negotiation in a collaborative environment and to reflect the strategic policies in every single operational decision.
For a modern system however, finding an adequate response to a broad range of possible new situations must be business-as-usual and this should apply even while the system is being modified/updated (Garcia et al., 2003). Thus, distributed control becomes standard practice as it allows a division of the problem in smaller and more affordable ones. In this case, the control system relies on a negotiation process to take decisions that affect more than one of its parts (Durfee, 1996). The need for an improvement in the connection between the real environment and the information systems is essential (Sandholm, 1999); Enterprise Resource Planning (ERPs), expert systems or other new information/management systems have given place to new architectures, which need constant feedback of information on the real environment.
This research also proposes the use of a WSN for acquiring tracking information on resources using RFID tags attached to mobile resources that are used at ground handling operations (Sung et al., 2007). This new visibility framework based on RFID is a powerful tool to help distributed continual planning because of the continuous information feedback about the resources it provides. Following these ideas, the new control system (Fig. 1) needs to integrate heterogeneous information coming from the manufacturing/logistics environment through elements such as sensors (RFID or Barcode) or actuators (PLC, Robots or simple actuators) as well as from more complex information systems as ERP, WMS or even expert systems.
Once the information has been captured from high volatile environments, it is necessary to take short-term and even online decisions whenever deviations or disturbances from the plan arise (Schönberger and Kopfer, 2007). These assignment tasks are closely related to the processes of review, control, planning, updating of information and 24-h-data availability. The current scheduling/planning operations, usually centralized at a control department, are based on software tools with a great lack of field communication. To adapt these tools, new ideas and concepts coming from those used at the development of Intelligent Manufacturing Systems (IMS) can be applied in distributed planning and decision making (Abramovici and Filos, 2009). The current trend is to manage the information using new concepts as alignment, agile manufacturing, or interoperability. Most of these are being implemented through new artificial intelligent techniques such as Bayesian networks (Jensen, 1996), multiagent systems (Bussman et al., 2004), Neural networks (Yu et al., 2009) or Fuzzy systems (Lee et al., 2008), which define expert DSS capable of adapting to disturbances.
The key contribution of the research is therefore the robust combination of MAS, DSS and WSN by taking advantages of all their individual features. The proposed combined framework is being implemented at the Ciudad Real Central Airport in Spain. This is a small airport and its demand is based on low cost companies and cargo operations. The airport wants to improve the productivity of ground handling resources by reducing delays, overloads and overstaff. This will reduce costs, following the path set by low cost companies. In the case at hand, software and hardware had to be developed from scratch and particularized to this airport. Thus, the first step was to develop a pre-validation environment that was presented as a demo show at the airport. This made possible the testing of the proposed MAS and the combination of ICTs before the deployment of the complete system.
Section snippets
WSN TO improve the visibility in high volatile environments
The first important stage towards making sound decisions is to gather the appropriate information. Even more so whenever the decision support system is to efficiently operate in real time conditions. Following the three requirements presented in the previous section; this DSS uses a combination of the technologies RFID and Zigbee to track the resources at an airport, a factory or a logistics center. The identified resources can be vehicles, products, pallets, machinery and employees. The idea
ZigID: the proposed WSN in three levels
Using the proposed technologies, this section presents a WSN architecture divided in three levels of elements (Fig. 3). The division is based on an adaptation of the adequate hardware devices to each application. The first level, or mobile-device level, is the end node level. The features of these end nodes are low consumptions, integrated sensors, short cover distances and identification process. The objective of the second level is the communication and routing of the information provided by
Intelligent product driven DSS: MAS+EPC
Once the information from the environment is correctly updated, sound decisions might be taken. In this section, the possibility of using MAS to appropriately use the data is studied, prior to the development of the DSS for the management of handling operations at airports.
As a new information system paradigm, manufacturing agent/holon based systems were developed to provide a high degree of specialization/autonomy in distributed DSS (Martinez Lastra and Colombo, 2006). Among the main MAS
Test case: the Ciudad Real Airport
The goal is to develop a WSN to improve the visibility in the terminal to help with resource management at the new Ciudad Real Central Airport (Fig. 9). This airport entered in operation at the beginning of 2008 at Ciudad Real (Spain). The airport focuses in the low cost market; which means that it needs to reduce costs in ground handling operations, in airline management and in other airport operations. The design of the airport is based in a hub with four aircraft parking positions facing the
Conclusions
The result of this work is the robust combination of a WSN based on RFID as identification technology and Zigbee as communication technology as well as its real implementation in an airport. The existing developed standards abstract the implementation of the low layers of the WSN, focusing the work in the deployment of software, the definition of the data model and the specification of final user requirements. The aim of the WSN is to cover big distances in the environment while performing the
References (36)
- et al.
A fuzzy AHP and BSC approach for evaluating performance of IT department in the manufacturing industry in Taiwan
Expert Syst. Appl.
(2008) Agent-based distributed manufacturing control: a state-of-the-art survey
Eng. Appl. Artif. Intell.
(2009)- et al.
Engineering framework for agent-based manufacturing control
Eng. Appl. Artif. Intell.
(2006) - et al.
Enterprise integration and interoperability in manufacturing systems: trends and issues
Comput. Ind.
(2008) - et al.
Reference architecture for holonic manufacturing systems: PROSA
Comput. Ind.
(1998) - et al.
Identifying source (s) of out-of-control signals in multivariate manufacturing processes using selective neural network ensemble
Eng. Appl. Artif. Intell.
(2009) Design and simulation of demand information sharing in a supply chain
Simulation Modelling Pract. Theory
(2007)- et al.
Industrial integration of ICT: opportunities for international research cooperation under the IMS scheme
J. Intell. Manuf.
(2009) - et al.
Radio-triggered wake-ups with addressing capabilities for extremely low power sensor network applications
Int. J. Wireless Inf. Networks
(2009) - et al.
Airport Operations
(1998)
Continual planning and acting in dynamic multiagent environments
Auton. Agents Multi-Agent Syst.
Multiagent System for Manufacturing Control. A Design Methodology. Springer Series on Agent Technology
Introduction to planning in multiagent systems
Multiagent Grid Syst.
A survey of research in distributed, continual planning
AI Mag.
Planning in distributed artificial intelligence
Foundations of Distributed Artificial Intelligence
RFID application development with the Accada middleware platform
IEEE Syst. J.
Cited by (22)
A turnaround control system to automatically detect and monitor the time stamps of ground service actions in airports: A deep learning and computer vision based approach
2022, Engineering Applications of Artificial IntelligenceCitation Excerpt :Therefore, the time wasted in the aircraft turnarounds can be minimized and the profits of the airport and airline companies can be maximized by reducing the delay-associated and human-oriented costs. Furthermore, these systems can be employed in more complex systems including similar intelligent service-based components such as real-time allocation of teams of ground handling operators to aircraft turnarounds (Saha et al., 2021), resource allocation for the airport ground handling management (Ansola et al., 2012) and flight manager assistants (Wilkins et al., 2008). A deep learning and computer vision based system that automatically detects, tracks and monitors the airport ground service actions during the turnarounds is proposed in this article.
Evaluation of the airport ferry vehicle scheduling based on network maximum flow model
2021, Omega (United Kingdom)Citation Excerpt :In recent years, with the rapid development of the global civil aviation industry, research on the management of airport ground handling process has received increasing attention [1,2].
Simulation model of security control system functioning: A case study of the Wroclaw Airport terminal
2017, Journal of Air Transport ManagementCitation Excerpt :The implementation of the method made it possible to increase the efficiency of the process. Ansola et al. (2012) also described dynamic resource management. They presented a system operating on the basis of a three-level architecture composed of an end node (RFID), a router (RFID and ZigBee) and a coordinator (ZigBee and Database).
Physical agents and RFID for the implementation of the shop floor control on a distribution centre
2015, RIAI - Revista Iberoamericana de Automatica e Informatica IndustrialDistributed automatic control system supported by RFID applied to a logistic centre test bench
2012, IFAC Proceedings Volumes (IFAC-PapersOnline)A decentralized framework for multi-agent robotic systems
2018, Sensors (Switzerland)
- ☆
The material in this paper was partially presented at the 2010 IFAC Conference (IMS 2010), July 01–07, 2010, Lisbon, Portugal.