Research Foundations for System of Systems Engineering

Transcription

1 Research Foundations for System of Systems Engineering Charles B. Keating, Ph.D. National Centers for System of Systems Engineering Old Dominion University Norfolk, VA, USA Abstract System of Systems Engineering (SoSE) continues to evolve as a potentially promising response for analysis, design, and transformation of increasingly complex systems problems. The purpose of this paper is to establish critical foundations and offer a coherent framing for further research in SoSE. To achieve this purpose the paper is organized to: (1) Introduce a contemporary perspective of SoSE, focused on identifying both convergence and divergence in the literature, (2) Develop five logical levels that can help to understand divergence in SoSE and sharpen future research efforts, (3) Articulate several of the critical research challenges that SoSE must address to be viable, and (4) Identify a preliminary set of critical topical research areas for development of a more integrated research agenda for SoSE. The paper concludes with implications for close coupling of research and practice for the accelerated development of SoSE through an integrated research agenda. Keywords: System of Systems Engineering, Research, Methodology 1 Introduction System of Systems Engineering (SoSE) is certainly a term that continues to be in good currency. However, a brief look at the use of the term finds a wide array of how it is used, what it means, and the nature of associated phenomena. In the early development of phenomena, fragmentation can be useful by enriching the dialog as our understanding of phenomena gain maturity and clarity. In this sense fragmentation in SoSE has been previously pointed out and continues [1]. However, at some point, continued divergence only leads to further fragmentation and serves little purpose for advancing our understanding of phenomena in meaningful ways. It might be suggested that SoSE is at a point where some convergence is necessary if the concept is to more forward in a coherent direction. Convergence in the research of phenomena is not an easy task and requires a solid foundation as a reference point. This is certainly the case with SoSE. There has been, and continues, to be much good work done in SoSE [2],[3],[4]. However, advancement of phenomena is always indebted to the establishment of solid foundations that serve as a grounded reference point for further elaborations. In this light, this paper suggests the groundwork necessary to support the establishment of the research foundations for SoSE. There are four primary objectives for the paper. First, the current state of SoSE is explored. This exploration is engaged to establish the degree, and source of fragmentation in SoSE. Second, the foundations upon which SoSE development must rest are developed. Foundations along epistemological, ontological, methodological, application, and method domains for research are explored. Third, the critical challenges that must be faced as SoSE moves forward are articulated. These challenges are offered in the spirit of identifying critical areas of necessary convergence to make SoSE research coherent and purposeful. Fourth, with deference to the current state of SoSE research, several critical directions for an integrated research agenda are suggested. This is not to claim exclusivity of research in SoSE. On the contrary, the suggestion for research direction is intended to begin a dialog of convergence and increase the probability of accelerating our understanding the phenomena associated with SoSE. The paper is organized to address each objective and concludes with implications for further development of SoSE research. 2 Current Perspectives of SoSE The term system of systems is somewhat problematic. System of systems (SoS) as a concept has been around for some time. In fact, early on, Beer [5] identified the notion of metasystem as an integrated system composed of other systems from a cybernetic perspective. Hutchins [6] has suggested that the concept of system of systems is poorly defined, and in fact tautological, since systems themselves are considered to be comprised of subsystems and therefore a system of systems is itself just a system. Nevertheless the system of systems concept has had multiple instructive perspectives put forward. We examine several of these below. Keating, et. al have described a system of systems as A metasystem, comprised of multiple embedded and interrelated autonomous complex subsystems that can be diverse in technology, context, operation, geography, and conceptual frame.these complex subsystems must function as an integrated metasystem to produce desirable results in performance to achieve a higher-level mission subject to constraints [1, p. 4]. Maier identifies a system

2 of systems as an assemblage of components, themselves considered as systems, with the added distinction of managerial and operational independence of components [2]. SoSE is a term with much less accounting in the literature. Keating et. al have described SoSE as The design, deployment, operation, and transformation of metasystems that must function as an integrated complex system to produce desirable results [1]. However, SoSE is still in the embryonic stages of development and there are emerging areas of convergence as well as divergence. Early in development multiple perspectives are important as the struggle to understand associated phenomena and conduct SoSE applications moves forward. However, as a starting point, there are several fundamental points of convergence, as well as divergence that can enhance the dialog. 2.1 Convergence in SoS/SoSE Perspectives There are several points of convergence in developing perspectives of SoS/SoSE. Multiple authors have elaborated on the nature and meaning of SoS/SoSE [1],[2],[3],[4],[6]. This has led to a constructive exploration of the terms. Although there is not a universally accepted definition of the terms, it appears that the following points of convergence are representative of the current themes emerging for SoS/SoSE: A SoS involves the integration of multiple, potentially previously independent, systems into a higher level system (metasystem). The function of a SoS generates capabilities beyond what any of the constituent systems is independently capable of producing. Integration into a SoS evokes some degree of constraint for previously independent systems. A SoS brings together systems in order to perform a higher level mission/purpose of which each member system plays an integral role. A SoS is a complex system and as such exhibits dynamic and emergent behavior, is difficult to grasp, and problematic to engineer. This set of convergent themes provides an important starting point for better understanding of the phenomena associated with engineering complex systems of systems. Having established this rudimentary convergence, we shift our attention to apparent divergences in perspectives of SoS/SoSE. 2.2 Divergence in SoS/SoSE Perspectives The literature engenders both explicit and tacit divergence in perspectives of the nature, organization, and applications of the phenomena associated with SoS/SoSE. To further the dialog and development of SoSE, several themes of divergence are beginning to emerge. These themes may be argumentative. However, they are intended to suggest areas for further dialog and exploration as SoSE moves forwards. The objective is not to suggest that the divergences be definitively reconciled for universal acceptance, or that one particular perspective must be adopted over all others. On the contrary, they are provided to sharpen the areas that are in need of further research and development as the phenomena associated with SoSE continue to be the subject of research. Divergences include: Beyond some high level agreement, there is not considerable broad acceptance of what constitutes a System of Systems [6]. Although SoSE is a term that has been used, it has not received broad acceptance in definition, underlying perspectives related to philosophy, methodology, or standards [1]. Research in SoS/SoSE has been fragmented, with a range of efforts and applications, with little research development targeted to underpinnings of knowledge [7]. Distinctiveness of SoSE in relationship to associated domains has not been clearly established [8]. As this set of divergent areas is examined, there are two glaring points of discovery. First, SoSE is in early stages of development, and therefore, should not be expected to exhibit the stature of maturity found in other domains such as systems engineering. Second, the nature of divergence lies beyond the technical questions of definitions, techniques, and approaches. Instead, the divergence in SoSE is at much more fundamental levels (philosophical, methodological, axiological, axiomatic). These fundamental issues are not capable of being addressed simply by additional applications, new definitions, or new techniques. To address the level of these fundamental divergences in SoS/SoSE requires exploration of the underlying sources of divergence. This difficulty is exacerbated by the immaturity in the field. In effect, the divergences identified above can exist at tacit levels that might render their resolution intractable in the current dialogs of SoSE. This has created a conundrum for SoSE. If we cannot identify the source of divergence in perspectives, how can constructive dialog be engaged, and more importantly, how can the field move forward through

3 resolution of divergence that has not yet been rigorously identified? The framework in the following section was developed to better develop the source for divergence in SoS/SoSE. The framework suggests that divergence stems from attempting to mix levels, including systems philosophy, axiomatic, methodological, application, and methods domains. The result of mixing levels can produce intractable divergence, since the source of the divergence remains concealed. In the following section the five logical levels of SoSE are developed. These levels offer insight with respect to understanding the source of divergence in the development of SoSE and associated phenomena. 3 Understanding Divergence in SoSE An examination of the divergent areas for SoS/SoSE compels an argument for organizing the research necessary to foster a higher level of convergence. If we conclude that SoSE is in its infancy, then it will be valuable to develop integrated research directions relevant to the most pressing issues to be addressed. Closer scrutiny suggests that the divergence in SoSE can be partitioned into five critical areas: System Philosophic, Axiomatic, Methodological, Application, and Method domains (Figure 1). System Philosophic System Axiomatic System Methodological System Application System Method Epistemological and ontological underpinnings Accepted knowledge (principles, theory, concepts, laws) that explain SoSE & associated phenomena Guiding frameworks that are used to guide inquiry and gain knowledge concerning complex systems Specific classes of problems that are appropriate for the application of SoSE to address Techniques, processes, or tools used to facilitate aspects of SoSE application Figure 1. Logic s of SoSE. In the following discussion, each of the different logic levels is briefly developed. At the philosophic level the very nature and purpose of SoS/SoSE is the subject for examination. Alignment at this level serves to inform a consistent worldview, or what Checkland [9] refers to as weltanschauung, for a system in focus. In turn, this worldview is based in the philosophical underpinnings used to inform the perspective of SoSE (epistemological and ontological) and drives purposeful decision, action, and interpretation from an internally consistent reference point. Following the work of Flood and Carson [10], Figure 2 provides a brief summary of endpoints of the philosophical spectrum for epistemology (concerned with how knowledge of a system is gained and how that knowledge is communicated externally) and ontology (the nature of reality from which system knowledge is derived). Positivism Knowledge is absolute, objective, and can be transmitted as tangible elements Realism Reality is external to the individual and objective Epistemological Spectrum Ontological Spectrum Antipositivism Knowledge is soft, subjective, and a function of the individual Nominalism Reality is an attribution of the individual and subjective Figure 2. Philosophic Spectrum Divergence at the philosophic level can result in conflict with respect to the fundamental nature and purpose of SoSE. For example, from a Positivist/Realist worldview, there would be little tolerance of SoSE as an approach not based in the production of tangible products in an objective manner. On the contrary, from an Antipositivist/Nominalist worldview, SoSE without inclusion of subjective knowledge, as a function of individuals, would be inconsistent. The point is not whether or not the ends of the spectrum are correct or incorrect. Instead, it is important to understand where, at the philosophic level, we are placing our worldview of SoSE. A disparity in worldview placement invites divergence at the philosophic level and potential conflict concerning the basic purposes and nature of SoSE. At the system axiomatic level the concern becomes what is accepted as source knowledge. This would include the laws, principles, theory, and concepts that inform the achievement of SoSE. It is fair to say at this point that there has not been a rigorous development of this base knowledge. However, there have been significant explications of systems principles, concepts, and laws that might be considered foundational to SoSE [11],[12]. Additionally, Maier [2] has provided architecting principles for systems of systems. However, the axiomatic base for SoSE is not significantly developed to provide the foundations that are necessary to provide sufficient grounding. Axiomatic inquiry involves the search for a consistent set of principles, concepts, and laws that are applicable to all domains looking at systems of systems. Without a level of axiomatic acceptance for core knowledge, SoSE is destined to divergence at this level.

4 The methodological level is concerned with the guiding frameworks that inform engagement of system of systems engineering problems. Following Checkland [9], methodology is viewed as a general approach, without being overly prescriptive. In effect, answering questions of How SoSE is done is the purview of the methodological level. There are approaches emerging to accomplish SoSE [8] as well as traditionally accepted approaches to systems engineering [13]. However, it might be expected that as SoSE continues to mature, the available methodologies to support SoSE will continue to grow. It is important to know that a SoSE methodology, either implicitly or explicitly, draws upon grounding from both the philosophic and axiomatic levels. The degree of divergence in methodologies for SoSE, present and future, will be rooted in philosophic and axiomatic based inconsistencies. Application level divergence is derived from differences inherent in the many assorted attempts to deploy SoSE. Each application of SoSE becomes unique, based upon the selection of methodological approach, context for deployment, and the specific system of systems that is the focus of design, analysis, or transformation. Again, applications of SoSE have embedded, explicitly or tacitly, grounding in philosophic, axiomatic, and methodological distinctions. Finally, SoSE must draw upon the method level to identify the specific techniques, tools, or processes that are appropriate to a specific application. The available methods to support SoSE are endless. Although they might not have been specifically constructed for SoSE applications, they can certainly be deemed appropriate. It is important to note that the determination of appropriateness is a function of the other levels, the context, and the system of systems problem. Only through the integrated development and accounting of each of these levels can SoSE achieve the accelerated deployment gains sought to advance the field. Effective research agendas must include development for all of the levels identified above. However, it is important to note that each level is interrelated and cannot be viewed as neither independent nor mutually exclusive of the others. 4 Critical Research Challenges in SoSE There are several challenging areas of concern for future research and development in SoSE. The logical levels provide an initial starting point for organizing research. However, to push the boundaries of SoSE research, there are two important distinctions that must be made. These distinctions are developed below. 4.1 Relationship Among Research s The research paradigm that can advance SoSE in an integrated and coherent manner will require engagement in each of the five levels we have identified. However, advances sought for singular levels are not likely to produce the accelerated progression sought for SoSE. It is a misnomer to consider that research can proceed without consideration of the close coupling between the different levels that have been articulated. Figure 3 depicts the relationship among and between the different levels. supports promotes Application Method challenges evolves Philosophic supports Methodological challenges Axiomatic Figure 3. Relationships of Logical s for SoSE As Figure 3 suggests, the levels of SoSE research do not exist independent of one another. Instead, each has a relationship with the others, consistent with a systemic perspective. The nature of these interrelationships has several implications for SoSE research which are expounded below. 4.2 Implications for SoSE Research There are several inherent challenges the relationships among logical levels pose for development of SoSE research. What is critical at this early stage of discussion is that: (1) there are different logical levels of SoSE that must be considered for coherence in research agendas, (2) each level is necessary, but not sufficient, to advance SoSE in an integrated fashion, (3) there is a relationship between the different levels, such that they are neither mutually exclusive nor independent of one another, and (4) to continue building effectiveness in SoSE through research, each of the levels and their interaction must be part of the research agenda. This is not to suggest that there will not be emergent debates and divergences as SoSE moves forward. Struggle is expected, should be encouraged, and provides a stimulus for constructive SoSE development. However, it is important to recognize the locus of the debate. For SoSE research to be dominated by a singular focus on only one logical level is certainly an invitation to stagnate what could otherwise be a more robust approach to research. Therefore, SoSE is in need of identification of a coherent research agenda that targets each of the levels as well as their interactions. However, SoSE research must

5 not be conceived as totally independent of prior work in systems, systems engineering, and other related fields. Although SoSE may draw on prior work, it is important to question critical distinctions that are necessary for SoSE to diverge in unique directions. It is inappropriate to look to SoSE as a simple extrapolation of prior work in related fields along the five logic levels that have been proposed. The ultimate successful evolution of SoSE will not be achieved by focus on a singular logic level. Instead, it is necessary to move SoSE forward in meaningful ways across all five levels. Only then will a rigorously grounded body of knowledge for application be generated. 5 Research Directions in SoSE Research in SoSE must be undertaken across all five levels that have been identified. As a start toward a research agenda, an initial set of framing questions and critical topic areas are identified below. 5.1 Framing Questions for SoSE Research What is a SoS? The characteristics, properties, and dimensions of a SoS are a continuing source of divergence and not rigorously articulated. Although there have been multiple suggestions of what characteristics need to be present to constitute a SoS, there is not wide acceptance, beyond high level generalizations, as to what the specific qualifications are for classification as a SoS. This is further exacerbated by the absence of detailed specific properties and dimensions that would be necessary to partition a SoS in detailed and meaningful ways. What are the attributes of a problem domain that suggests it is suitable for application of SoSE? This requires that distinctions in problem domains, classifications, and typologies be developed for appropriateness of SoSE application. All complex systems problems are not necessarily appropriate for SoSE. The determination of suitability for SoSE is an important question to be answered. How can a SoS be engineered? The approach(es) necessary to design, analyze, and transform a SoS as a holistic system are currently limited. This is not to suggest that there is not effectiveness in dealing with systems of systems. However, the maturity of the approaches and acceptance is still in the infancy for systems of systems. Although several authors have offered development of approaches [4],[6],[8], there is not sufficiently accepted convergence on how to effectively address a SoS problem. What are the philosophical, theoretical, axiomatic and methodological underpinnings for SoS/SoSE? These underpinnings have not been rigorously developed. This represents a tacit divergence between theory and application. The thrust in SoS/SoSE has been in the application domain. Although this is necessary, it will not be sufficient to support the more rapid and grounded SoSE development sought through rigorous research. What is the relationship/distinction of SoSE with respect to systems engineering? There is not clarity in the understanding of how SoSE is distinguished in relationship to systems engineering. This distinction/relationship is critical to the future directions for SoSE research and development. It is presumptuous to believe that what has been effective in developing individual complex system solutions will be directly applicable or capable of extrapolation to achieve the same level of success in SoSE. What are the systemic principles and relationship to other systems-based fields that inform/govern SoSE? This question has not been critically addressed. In effect, there must be an assumption that SoSE will develop specific principles, concepts, and laws that exist beyond existing related domains of inquiry. These questions offer a starting point for development of the high level guide for research in SoSE. 5.2 Related Topical Areas for SoSE Research SoSE is in an infancy of development. However, there are several critical research areas that are in need of further development. Although the listing is not complete, it offers an initial starting point. It must be kept in mind that each of these areas is not intended to be undertaken by excluding any of the five logic levels introduced earlier. Instead, it is appropriate to consider how each of the topic areas might be considered across all of the logic levels. System of Systems Context context exist as the circumstances, factors, conditions, or patterns that enable or constrain complex system solutions. Bounding Systems of Systems the rigorous bounding of complex systems along different dimensions (geographical, physical, time, conceptual). System of Systems Concept the characteristics, properties, and classification of a system as a system of systems. Analysis of Systems of Systems the development of rigorous approaches to engage in holistic analysis for systems of systems. System of Systems Transformation the nature, framing, and approach to transforming systems of systems from a holistic perspective.

6 Modeling and Representation of Systems of Systems the distinction in modeling approach, and the role of representation, for systems of systems. Intervention in Systems of Systems the design and deployment of initiatives to purposefully modify a system of systems or supporting worldviews. Facilitation of System of Systems Engineering the execution of methodologies and environments necessary to engage in SoSE. System of Systems Engineering Methodology the development of guiding frameworks to support SoSE execution. Although this set is not complete, it identifies several useful directions to push research in systems of systems forward in a purposeful manner. 6 Conclusion This paper has established the critical foundations for research in SoSE. Although there has been excellent work done in systems of systems, there is much yet to be accomplished. To assist in the further development of integrated research in SoSE, a framework of five logical levels (philosophic, axiomatic, methodological, application, and method) was developed. The framework provides a level of organization to ensure more robust consideration of SoSE research undertaken. In conclusion, three major implications for evolving research in SoSE are offered: (1) divergence in SoSE research must be considered with respect to the logical level(s) it is intended to address, (2) coherent research in SoSE must include all levels (philosophic, axiomatic, methodological, application, and method) if it is to maintain balanced evolution of the field, and (3) without an integration framework for SoSE research, the development of the field will be slowed, and most probably progress in a fragmented fashion, failing to fully realize the potential of this emerging domain to address engineering of complex systems of systems. [3] A. Sage and C. Cuppan, "On the Systems Engineering and Management of Systems of Systems and Federations of Systems," Information, Knowledge, Systems Management, Vol. 2, No. 4, pp , [4] P. Carlock and R. Fenton, "System of Systems (SoS) Enterprise Systems for Information-Intensive Organizations," Systems Engineering, Vol. 4, No. 4, pp , [5] S. Beer, The Heart of Enterprise, Wiley. Suffolk [6] D. Hitchins, Advanced Systems Thinking, Engineering, and Management, Artech House. Norwood [7] C. Keating, A. Sousa-Poza, J. Mun, Toward a Methodology for System of Systems Engineering, Proceedings of the American Society of Engineering Management, pp. 1-8, [8] C. Keating and A. Sousa-Poza, System of Systems Engineering Methodology, National Centers for System of Systems Engineering Technical Paper, Unpublished Manuscript, [9] P. Checkland, Systems Thinking, Systems Practice, Wiley. Chichester, [10] R. Flood and E. Carson, Dealing with Complexity, 2 nd ed., Plenum. New York, [11] L. Skyttner, Introduction to General Systems Theory, Plenum. New York [12] B. Clemson. Cybernetics: A New Management Tool, Wiley. Chichester, [13] B. Blanchard. System Engineering Management, 3 rd ed., Wiley. New York, References [1] C. Keating, R. Rogers, R. Unal, D. Dryer, A. Sousa- Poza, A., Safford, R., Peterson, W. and Rabadi, G., "System of Systems Engineering," Engineering Management Journal, Vol. 15, No. 3, pp , Sep [2] M. Maeir, Architecting Principles for Systems-of- Systems, Systems Engineering, Vol 1, No. 4, pp , 1998.