The holonic enterprise: a model for Internet-enabled global manufacturing supply chain and workflow management

Merges the latest results obtained by the holonic manufacturing systems (HMS) consortium with the latest developed standards for platform interoperabilit y released by the Foundation for Intelligent Physical Agents (FIPA) to propose a novel e-business model: the holonic e-enterprise (HE). The HE extends both the HMS and FIPA models. On one side it extends the holonic manufacturing paradigm with one top level, the interenterprise one. On the other side it extends the multi-agent system (MAS) paradigm to the hardware (physical machine) level. Received October 2001 Revised June 2002 Accepted July 2002 This paper builds on the results obtained over the past ten years by the members of the Intelligent Systems Group (ISG ± http://isg.enme. ucalgary.ca) at the University of Calgary, under the leadership of Professor Douglas H. Norrie. Without his vision and continuous support, this work would not have been possible. Thank you, Doug, for being such an inspiring mentor to us all( as autonomous entities, yet cooperating to form apparently self-organizing hierarchies of subsystems, such as the cell/tissue/organ/ system hierarchy in biology (Christensen, 1994). Holons at several levels of resolution in the holarchy behave as autonomous wholes and yet as cooperative parts for achieving the goal of the holarchy. Within a holarchy, holons can belong to different clusters simultaneously, displaying rule-governed behavior. The rules define a system as a holon with an individuality of its own; they determine its invariant properties, its structural configuration and functional pattern. The duality autonomy-cooperation as main contradictory forces within a holarchy is balanced by the rules that define the functionality of such a system of semiautonomous holons. From a software engineering perspective, a holon, as a unit of composition retaining characteristic attributes of the whole system (holarchy), can be viewed as a class. Thus the object-oriented paradigm seemed suitable for modeling holarchies as software systems (Booch, 1994). The multi-agent systems paradigm In response to the need for modeling the complexity of interactions in large-scale distributed systems, agent technology has emerged as a paradigm for structuring, designing and building software systems that require complex interactions between autonomous distributed (software) components. While the object-oriented paradigm models systems focusing on the structural, static characteristics of their parts, which are defined through encapsulation and inheritance, the agent paradigm models systems focuses on the underlining dynamics defined by the interactions between their parts. In contrast to the passive way in which objects communicate by invoking methods in one another in a way controlled externally by the user (e.g. from a `̀ main’’ program), agents are capable of initiating communication and deciding (like a human) when and how to respond to external stimuli (e.g. manifested on them as requests from other agents). From this perspective the agent paradigm extends the object paradigm in that agents can be regarded as proactive objects (Wooldridge, 2001) that have an internal mechanism which governs their behavior, enabling them to initiate action as well as respond to the outside environment in an autonomous way. With this in mind one can define: an intelligent agent as a software entity which exhibits, in some significant measure, autonomy, intelligence, and environmental awareness, and which interacts with its environment to achieve internal goals; a MAS as a software system in which program modules (the individual agents) are given autonomy and intelligence and an underlining coordination mechanism (implementing rules for collaboration, like for holarchies) which enables collaboration between such modules (agents) to attain system objectives. A software representation of a holarchy thus appears natural as MAS, consisting of autonomous yet cooperative agents. From this perspective a MAS is regarded as a system of agents (software holons) which can cooperate to achieve a goal or objective. The MAS (software holarchy) defines the basic rules for cooperation of the agents (software holons) and thereby limits their autonomy. In this context, autonomy is defined as the capability of an entity (i.e. agent or holon) to create and control the execution of its own plans and/or strategies, while cooperation is the process whereby a set of entities develop mutually acceptable plans and execute them. The common denominator between holonics and MAS as paradigms is obviously the focus on the dynamics of the interactions. However, in a MAS there is no pre-assigned condition that the interactions should be driven by cooperative forces, while in a holonic system this is a precondition for the existence of the holarchy per se (the glue that binds the holarchy together driving it towards the common goal). It is this `̀ teamspirit’’ that characterizes a holarchy, in that all its component parts at all levels of resolution work together towards achieving the goal in an optimal manner. This `̀ togetherness’’ drives the self-organizing power that configures all the sub-holons to optimize the interactions within the holarchy to reach the common goal with maximum efficiency. On the other side, in a MAS, agents may interact based on competitive rather than cooperative rules (e.g. electronic markets or other competitive/ conflicting environments such as military scenarios; competing over resources or societal/political disputes, etc.) ± which is excluded as a possibility in a holarchy. The Internet The MAS paradigm has challenged the software world and with it the world of information technologies through its ability to enable emulation in Cyberspace of realworld societies as virtual communities of agents. The marriage between MAS and the Internet has created a parallel world of [ 539 ] Mihaela Ulieru, Robert W. Brennan and Scott S. Walker The holonic enterprise: a model for Internet-enabled global manufacturing supply chain and workflow management Integrated Manufacturing Systems 13/8 [2002] 538±550 information that `̀ lives’’ in the Web universe emulating our games in all aspects of life, be they economic, financial, business, school or health-related, or even just-for-fun in computer games. MAS enable cloning of real-life systems into autonomous software entities with a `̀ life’’ of their own in the dynamic information environment offered by today’s Cyberspace. The WWW connects by invisible links these entities through their virtual `̀ clones’’ forming `̀ societies’’ in which the virtual entities (mostly modeled as software agents) have their own `̀ life’’ interacting with an autonomy of their own. When such virtual societies are driven towards a common purpose they cluster into collaborative holarchies (Ulieru, 2002). Enterprises partially `̀ cloned’’ as agents that interact over the Internet, can cluster as well into holarchies to form global virtual organizations. Two main enterprise-related paradigms have emerged supported by this technological development: the Web-centric enterprise and the virtual enterprise. Unlike existing point solutions that focus on a single-department or activity product, such as data management or product-designand-manufacturing, the Web-centric model (Hornberger, 2001) addresses product and process life-cycle management across the extended enterprise regarded as a global organization. At the core of the Web-centric enterprise model is the Internet-enabled software infrastructure acting as a worldwide open DSE. Such an integrated framework enables sharing of information, services and applications among suppliers, employees, partners and customers via: Deployment of automated, intelligent software services (e.g. Internet-enabled negotiations, financial transactions, advertising and bidding; order placement/ delivery, etc.). Complex interactions between such services (e.g. compliance policies; argumentation and persuasion via complex conversation protocols, etc.). Dynamic discovery and composition of services to create new compound value added services (e.g. dynamic virtual clustering of synergetic partnerships of collaborative organizations aiming to achieve a common goal). A virtual organization or company is one whose members are geographically apart, usually working by electronic linking via computers while appearing to others to be a single, unified organization with a real physical location. Within a virtual organization, work cannot be completed without support of an information technology infrastructure in linking the parts. The virtual enterprise (VE) paradigm differs from theWeb-centric paradigm in that a VE is a distinct organizational form, not just a property of any organization. Thus, Web-centric organizations that can use communications extensively, but not in a way critical in fulfilling the goal of the organization (e.g. a multinational corporation with dispersed parts being on the same satellite network whose use, however, is not critical for completing the production process) are not VE. In today’s global economy in which enterprises put together their competitive advantage to leverage a higher purpose otherwise impossible to achieve, the VE is an appropriate model for strategic partnerships. Such a strategic partnership model calls for new perspectives on competition in the global open Internetenabled economy. The networked economy mandates the shift from industrial age, `̀ brick-and-mortar’’ strategic thinking to an emphasis on new alliances and a rethinking of traditional partnerships. Alliances and partnerships can be formed in ways that increase value for all players. The concept of co-opetition (Brandenburger and Nalebuff, 1996) builds on the duality inherent in all relationships with respect to win-win and win-lose interactions. The success of most businesses is dependent on the success of others, yet they must compete to capture value created in the market and protect their own interests. The main issues to be addressed when developing a business strategy based on co-opetition are: Who are the players in the network and how can they collaborate to maximize value? Which relationships are complementary in nature ± which companies can add value to what they provide? Which players are competitors, and are there mutually beneficial ways to create value? What can they do to sustain their competitive advantage over time? Holonic collaboration In this section, we discuss how the holonic paradigm supports collaborations of autonomous entities. This results in the basic holonic notion of autonomous and cooperative building blocks (i.e. holons) that are used to lay the foundation of the HE discussed in the next section. [ 540] Mihaela Ulieru, Robert W. Brennan and Scott S. Walker The holonic enterprise: a model for Internet-enabled global manufacturing supply chain and workflow management Integrated Manufacturing Systems 13/8 [2002] 538±550 The mediator architecture A system decomposition and analysis based on holonic principles naturally suggests a distributed software implementation, with autonomously executing cooperative entities as building blocks. As illustrated in Figure 1, the stable intermediate forms/holons of the system can be implemented at the lower levels by objects, at the medium level by agents, and at the higher levels by groups of agents, with these mapping decisions being application-specific. (Of course, these are only the software portions of the holons.) Also, holons should have an interface which is simple and cohesive, just like in objectoriented systems (Eliens, 2000) or any effective organizational structure, and which is itself a holon. In the previous section we emphasized on the cooperative forces that drive the holons towards achieving the common purpose of the holarchy. How does one build agents and groups of agents which fulfill the holonic philosophy? (The third option, mapping holons into objects, is an almost trivial task that needs no discussion here.) The basic condition for holonic systems is that a holon is simultaneously a `̀ whole’’ and a `̀ part’’ of some other whole/holon. This means that holons may contain other lower level holons, and may themselves be contained in other higher level holons, resulting in a recursive architecture. The agents to be used to implement this holonic system will be considered independently executing processes on some computer/machine/ device. In this case, if a one-to-one mapping of holon to agent is performed, it is much more difficult to implement an agent practically (than it is to conceptualize a holon) which is itself a component of a higher level agent and which also contains lower level agents. Here, the concept of a mediator agent comes into play. The mediator will fulfill two main functions. First, it acts as the interface between the agents in the holon and between the agents outside the holon (i.e. acts as a type of facilitator); conceptually, it can be thought of as the agent that represents the holon. Second, it may broker and/or supervise the interactions between the subholons of that holon; this also allows the system architect to implement (and later update) a variety of forms of interaction easily and effectively, thereby fulfilling the need for flexibility and reconfigurability. Such a mediator as described can actually be considered a static mediator, and will exist primarily at the boundary of a homogeneous holon (such as an ordering holon in a supply chain example), as illustrated in Figure 2. In manufacturing holarchies the mediator encapsulates the mechanism that clusters the holons into collaborative groups (Maturana and Norrie, 1996). This type of mediator is a `̀ dynamic mediator’’. In the case of interactions between heterogeneous holons, such as an order holon and various resource scheduling holons, ease of system design may be supported by employing a dynamic mediator agent to broker and/or supervise the Figure 1 Multi-granular decomposition of holons in agents and objects [ 541 ] Mihaela Ulieru, Robert W. Brennan and Scott S. Walker The holonic enterprise: a model for Internet-enabled global manufacturing supply chain and workflow management Integrated Manufacturing Systems 13/8 [2002] 538±550