Integration Framework for Complex Systems


Information is seen as one of the main resources that systems analysts try to use in an optimal way. In this chapter we show how this resource can be used in integration issues. We introduce the problem of information-based integration, propose a solution, and briefly discuss future trends in this area. Systems become increasingly complex. Their decomposition into smaller units is the usual way to overcome the problem of complexity. This has historically led to the development of atomized structures consisting of a limited number of autonomous subsystems that decide about their own information input and output requirements, that is, can be characterized by what is called an information closure. In a real-world context, autonomous subsystems consist of groups of people and/or machines tied by the flow of information both within a given subsystem and between this subsystem and its external environment (Esteve, 2002; Szczerbicki, 2003; Tharumarajah, 1998). Autonomous subsystems can still be interrelated and embedded in larger systems, as autonomy and independence are not equivalent concepts. These ideas are recently gaining very strong interest in both academia and industry, and the atomized approach to complex systems analysis is an idea whose time has certainly come (Liu & Ling, 2003; Orlowski,2002).
Complex systems (for example manufacturing) are often viewed as sets of components (agents, subsystems) supporting separate functions. Many organizations operate in this highly compartmentalized manner. It appears that the general direction of systems in the future, however, is toward linking together function-specific agents into fully integrated entity. An integrated system is a system that consists of agents/subsystems that efficiently contribute to the task, functional behaviour, and performance of a system as a whole. It is believed that such an integration can be achieved through the flow of information. “Integration,” as used in this article, should not be confused with integration at a physical level by means of computer networks or computer buses. Rather, the semantics of integration is addressed – the information that subsystems should share. While structuring the approach presented, the first consideration was to design some tools that could be easily implemented as components of an intelligent system supporting development of system configurations integrated by the flow of information. Due to the complexity and creativity associated with the early stages of such a development it is quite clear that the way a practicing analyst solves a system configuration problem cannot be easily implemented. This explains the need for new tools and approaches that solve the problem but at the same time can be supported by a computer. Elements of such an approach are presented in this article.


We propose a three-stage approach for the development and analysis of complex systems. The involved stages are systems decomposition, subsystem modelling representation, and integration at the level of information flow. Figure 1 depicts the general underlying idea behind this approach.

Figure 1. Overview of a three-stage approach to complex systems integration

Overview of a three-stage approach to complex systems integration
The essence of information-based integration problem can be formulated as follows and illustrated as in Figure 2.
Theoretical framework to provide support for systems integration as outlined includes the following fundamentals:
• the syntax for connections of autonomous subsystems,
• the mechanism for guiding the generation of such connections,
• the integration algorithm.
The aim of this article is to briefly outline these fundamentals.
The author, together with his collaborators, has been researching information-based integration issues since the early nineties, starting at The University of Iowa (Kusiak, Szczerbicki & Vujosevic, 1990, 1991), Iowa City, USA, continuing at The GMD FIRST, Berlin, Germany (Szczerbicki, 1994) and currently working on these issues at The University of Newcastle, Newcastle, Australia (Szczerbicki, 2003). The aim of this entry, which is based on previous research publications by the author, is to overview the integration problem from the perspective of the author’s experience and to place it among the work of others.
Integration problem is recently gaining very strong interest in both academia and industry. This is particularly apparent in the area of design and modelling of manufacturing systems (O’Grady, 1999). Model development and synthesis (that resembles autonomous systems development and integration) is frequently based on the general systems theory and it uses hierarchical structures and a number of model base concepts (Esteve, 2002; Rolstadas & Andersen, 2000; Wyzalek 1999). In Raczkowsky and Reithofer (1998), the future development of a hierarchical communication model for coordination of a set of agents performing several functions is addressed. A conceptual modeling approach to represent the complexities in CIM (Computer Integrated Manufacturing) systems, including such issues as information acquisition, storage, dissemination and the time and costs associated with such informational activities is proposed in Tharumarajah (1998). The problem of coordination of multiagent manufacturing systems developed to fulfill their functional requirements advocating a decentralized approach in which each agent has relative autonomy over its own actions is discussed in Pacholski (1998). In O’Grady (1999), the application of modular paradigm to integration of systems in which planning, grouping, and scheduling are the central functional areas is described. The role of the flow of information in the process of integration is discussed in Prakken (2000) and Orlowski (2002). In Kamrani and Sferro (1999), the integration of manufacturing agents (information islands and automation islands) using knowledge-based technology is proposed for the factory of the future.

Figure 2. Illustrative example of a three-level hierarchical tree of the bottom-up integration process

Given the informational inputs and outputs of autonomous subsystems, find the overall system being designed that meets the desired functions and is integrated through the flow of information.
Illustrative example of a three-level hierarchical tree of the bottom-up integration process
Traditionally, information system analysts have been solving the integration problem in an ad hoc manner. What we propose through our research is a formal integration approach suitable for computer implementation.
Information-based integration problem should be seen as one of the challenges within the broader, more general aim to enhance systems performance through the flow of information. Engineering, operations research, information science and management science use scientific and engineering processes to design, plan, and schedule increasingly more complex systems in order to enhance their performance. One can argue that systems have grown in complexity over the years, mainly due to increased strive for resource optimization combined with a greater degree of vagueness in the system’s environment. Information is seen as one of the main resources that analysts try to use in an optimal way in complex systems. Proper design of information flow, its management, its use, its maintenance, that is, information engineering, is critical in systems’ abilities to act appropriately in an uncertain environment – to act intelligently (Baba et al., 2001; Bogdan, 2000; Gunasekaran & Sarhadi, 1997; Morabito, 1997; Prakken, 2000; Tharumarajah, 1998).


Autonomous subsystems are matched using informational input and output defined at representation stage (Figure 1). For example, in the domain of environmental engineering, information may represent geographical positioning, urban planning, sources of pollution, traffic data, and the like. In the domain of manufacturing systems, informational inputs and outputs may represent material availability, tool availability, machine availability, number of parts produced, and number of products assembled. After the matching has been accomplished, the informational input variable of a given subsystem represents the value of the informational output variable of the subsystem to which it has been connected. For example, if the input variable X of autonomous subsystem AS2 is matched with the output variable Y of AS1, then the syntax of this connection will be given as:
AS1.Y–> AS2.X (1)
Similar simple syntax can be used for all structures that can be produced during the integration process. These structures are enclosed into higher-level subsystems using ports. The informational input and output ports provide an interface to the subsystem environment. This interface is used to develop hierarchical structures. With the syntax of autonomous subsystems connections in place, the mechanism for guiding the generation of such connections is required. The mechanism needs to represent qualitative system theoretic knowledge, and so it is based on IF… THEN production rules.
Generation of connections between elements in model base of autonomous subsystems is guided by the following production rules (Kusiak, Szczerbicki & Vujosevic, 1990, 1991; Szczerbicki, 2003):
• Rule 1
IF there is only one element left THEN do not generate connections
• Rule 2
IF a single element that is left includes boundary inputs and outputs only THEN it is an overall system
• Rule 3
IF there are more than one element
THEN select a connection for an input boundary element
• Rule 4
IF there are elements other than the boundary elements
THEN do not specify any connections that involve boundary elements only
• Rule 5
IF an element is an input boundary element THEN it cannot accept an input from any other element
• Rule 6
IF an element is an output boundary element THEN it cannot provide an input to any other element
• Rule 7
IF two elements have identical output and input variables
AND there are no production rules that prevent from connecting them
THEN specify the connections for these elements
• Rule 8
IF there are no elements with identical input and output variables
AND there are elements with partially identical input and output variables
AND there are no production rules that prevent from connecting them
THEN specify the connection for these elements beginning with the closest match
• Rule 9
IF a connection for an input boundary element has been specified
THEN continue with selecting connections for elements that have not been listed in the specifications
• Rule 10
IF there are boundary elements only THEN specify connections between them
These production rules are domain independent and were structured using the underlying general systems theory. The analyst may, however, add domain-based production rules. They may follow, for example, the safety requirements, emission data, traffic data, or other constraints imposed by the analyst.
The last tool needed for simulation of the information-based integration process is an integration algorithm. The algorithm guides the simulation process across various levels of integration illustrated in Figure 2. It was developed with the assumption that in order to enter the next level of integration it is enough to generate just one connection in a given step. Elements taking part in integration that are not matched at integration level i are considered for matching at level The algorithm terminates at the level at which it will no longer be possible to match subsystems into pairs (no connections will be generated). At each integration level, production rules presented previously are fired during simulation process to generate connections between remaining integration elements. The integration algorithm is the last element of fundamentals of a framework to support systems integration process. The algorithm includes the following simple steps (Kusiak, Szczerbicki & Vujosevic, 1990, 1991; Szczerbicki, 2003):
• Define database of autonomous sub systems (bottom level of illustrative example of integration process in Figure 2) Set level = 1.
• Generate connections between elements at current^
• If no connections are generated, stop; Otherwise, match elements in database into pairs using the existing connections.
• Define informational input and output variables for subsystems generated by the matching process.
• Remove from the database all elements that have taken part in the matching process.
• Add to the database all subsystems generated by the matching process.
•. Set level = level + 1 and go to Step 2.
The presented framework for information-based integration has been applied in numerous real-life based cases in which simulation was used to arrive at integrated problem solution. Some particularly successful applications included system modelling and integration for a coal mine (Szczerbicki & Charlton, 2001), integrated agile manufacturing strategy (Szczerbicki & Williams, 2001), modelling for steel processing (Szczerbicki & Murakami, 2000) and integration of maintenance services (Szczerbicki & White, 2003). In all these cases all three stages of systems decomposition, representation and integration were present (see Figure 1), integration representing the last step in the process of arriving at a systems structure integrated by the flow of information.


Information-based system integration is gradually becoming one of the main challenges of our new millennium of information age. The most prominent future trend in this area is focused on using Web-based technologies for tackling problems of integration. Within this trend we have a rapid increase of research efforts towards developing metamodelling architectures and interoperability of Web-enabled information flows (Terrase et al., 2003), semistructured data integration (Liu & Ling, 2003), schema integration (Castana et al., 2003), and Web-based aggregation architectures (Bussler, 2003). The explosive popularity of the Web makes it an ideal integration tool as it opens the possibility to integrate geographically distributed systems. Also, in the future the Web may become a universal platform to synthesize a number of integration approaches (for example like the one presented in this article) into one universal Web-based integration interface.


Information flow integration is one of the major activities of the design process of an integrated system. The outcome of the integration process is the overall system integrated through the flow of information.
In this entry, integration problem is formulated as follows: Given the informational inputs and outputs of autonomous subsystems, find the overall system being designed that meets the desired functions and is integrated through the flow of information.
Autonomous subsystems are integrated using an algorithm into an overall system that has a hierarchical structure. General production rules supporting generation of connections for subsystems relate to the underlying systems theory. They are structured independently of the system’s domain and cannot be modified by a system analyst. Production rules ensure that only feasible variants of the designed system are explored.


Autonomous (Sub)System: A system that decides about its own information input and output requirements.
Complex System: From mathematical perspective a system which is described by differential or difference equations; from informational perspective a system for which information is the main resource and functioning in information rich environment.
General System Theory: Collection of tools, approaches, hypotheses and models that can be used for scientific discovery.
Information Based Integration: Process in which given informational inputs and outputs of autonomous (sub)systems, analyst develops the overall system being designed that meets the desired functions and is interconnected through the flow of information.
Information closure: Boundary defined by information and its sources necessary for autonomous (sub)systems functioning.
Information Engineering: Proper design of information flow, its management, its use, and its maintenance.
Information system: A collection of organised procedures collecting, storing, processing and retrieving data. It provides information to support the organisation.
Intelligence: Ability of a given system to act appropriately (i.e. to increase the probability of the achievement of given aims) in an uncertain and changing environment.
System: A collection of components (subsystems) together with their interrelations.

Next post:

Previous post: