The Evolution of Legacy Acknowledged System of Systems to Directed - Research Paper Example

? System of Systems Engineering Introduction Systems of systems refer to acollection of technical and large-scale applications that are decentralized, exclusively distinguishable and autonomous yet mutually connected to form a complex network of systems (Jamshidi, p.17). Systems collaborate to form a complex and integrated network of systems which achieves new operational capability but at the same time, shows some emerging negative symptoms that must be controlled. This dissertation is aimed at studying the dynamics of emergent behavior demonstrated by SoS with help of a research on practical scenario of a large-scale system. The paper highlights thesis of prior researches conducted in this regard and a corresponding modeled system simulation that caters problems faced in previous studies and enables systems engineers to assess emergent attributes of SoS. In closing notes, we draw conclusions regarding possibilities of projecting impact of emergent behavior through application of modeling and simulation techniques. Figure 1: Dynamic and complex US$54 trillion SoS (IBM Value Analysis of OECD data) Problem description SoS are formed in new ways and through natural development, with no easily understandable symptoms and indicators. Therefore, it has a tendency to depict unpredictable behavior and effects which must be identified beforehand in order to develop an approach to control them. The basic concept of emergent behavior entails the attribution of combined performance of all constituents to be the overall behavior of SoS (Maier, p. 268). Researchers have put extensive amount of efforts and time on understanding the complexities of emerging characteristics of SoS. Emergent behavior refers to a state that arises as a product of collaborative interactions of components of SoS. This behavior is highly complicated and cannot possibly be controlled through inadequate conventional systems engineering tactics such as regular hierarchical structures and centralized management. Present engineering practices do not constitute methodologies and analytical techniques to accommodate these obstacles: it is an intricate challenge for engineers to be able to speculate emergent behavior for an underdeveloped SoS. However, managing emergent behavior of SoS is also necessary since it is capable of generating exclusive benefits. On the contrary, this behavior opens new horizons for engineers to experiment with in order to support the needs of users. This behavior contributes to the deciding factors regarding the methodologies and approach that users will adopt in future to conduct the business. Moreover, they enable engineers to identify those weaker areas of system which are source of hindrance for users to meet their needs and where further nurturing resources must be deployed. It enables the system to combat threats demonstrated by interoperation and attain traits such as flexibility, scalability and cost leadership. Emergence is mostly unavoidable; therefore it is pivotal that engineers learn how to coexist with it, effectively using this phenomenon to establish an invulnerable and highly elastic SoS that can easily respond to changing environment. If emergent behavior is managed effectively, systems forming SoS interact with each other, creating a synergetic effect that enables the entire network to fulfill its essential purposes that cannot be possibly attained in isolation as a standalone system. Therefore, it is only fair to state that the evolutionary process of SoS is a direct product of effective management of its emergent behavior. Figure 2: Simple example of emergent behavior of SoS (Thwink.org) Emergent behavior has largely been researched into and many experiments have revealed different definitions and characteristics to it. Dyson and George claimed that it is a behavior shown by SoS which is not easy to foresee and analyse through evaluation of its components; instead, it must be seen as a single system on the whole. They state “emergent behavior, by definition, is what’s left after everything else has been explained!” (Dyson p.57.) At another occasion, it has been defined as a collection of straightforward systems and their actions, adding up to generate results that have multiplying complexities (Rollings and Adams, p. 43). According to Parunak and VanderBok, a distributed approach leads to creation of a network of systems that displays behavior far more complicated than the characteristics attributable to systems that constitute to the network. This behavior emerges as a byproduct of interaction between components of the network and hence is called emergent behavior. They suggest that the consequences that follow this behavior are contradictory to initial projections and planning developed by designers of the system (Parunak and VanderBok, p. 2). Lastly, Fisher, one of the most popular researchers in field of systems engineering, terms this behavior as a united outcome from regular interaction and contact among elements of systems which can be an automated piece of equipment or even an organic or biological being (Fisher, p. 13). Research plan As the size of system grows, so does the level of complexity and unpredictability faced in due course. The important question that arises at this point of time is whether the systems engineers are capable of anticipating such behavior of SoS and are equipped with tools and techniques that shall facilitate them in controlling and managing effects caused by it. Modern approach is capable of controlling and predicting emergent behavior, thereby utilizing its effects to the benefit of SoS (Hsu and Butterfield, p. 59). When adopting this methodology, the speculations made shall be based on probabilities and best estimates but do not present a pure and accurate result. We shall be using agent-based modeling to study the emergent behavior of SoS. Agent-based modeling involves flexible systems for which no lesser complex models than the system itself is capable of anticipating future behavior of SoS accurately and comprehensively (Bankes, p. 7199). This approach seems most appropriate since the overall configuration and pattern of behavior is guided by the individual interactions of member systems of network. This modeling approach shall comprise a series of various analyses, breakdown structuring procedures and scrutiny through simulations. In chronological sequence, the following steps shall be conducted to achieve goals of this research: Development of hypothesis Initial planning, selection and development of SoS which provides operational infrastructure desirable for this research Recognition and classification of components of system and their interdependence Replication of operational infrastructure in a modeling language through application of simulation (Huynh and Osmundson, p. 25) Design of SoS through identification of significant parameters and factor levels Conversion of prototype into a language and mode that is easily executable Comprehensive experimentation on model to develop understanding Performance of test runs through simulation to study behavior exhibited by SoS Analysis of results and establishment of conclusions based on initial thesis Modeling techniques involve various analytical expert tactics based on mathematical, scientific or automated methodologies (Brenner and Werker, p. 231). To complete this experiment, we adopt an approach whereby a prototype agent is created as an algorithm based on such rules, with cushions for minor deviation from planned design. The behavior of components is dependent on inputs delivered by SoS interaction. Similarly, if the component is a human, then we model him or her in the same manner using algorithms and sets of rules and probabilities. Likewise designing methods shall be adopted and applied for system functions. This methodology can be extended to cover various types of behavior of agents, varying from industry to industry and the size and architecture of the network. Hit and trial method can be adopted, trying and analyzing all types of systems as components of the network and all types of emergent behavior that can possibly be stimulated therein. For example, the extent of success of internet-based online e-commerce systems such as eBay and Napster can be assessed by measuring the number of visitors attracted to use their services and the number of transactions resulting from system variables. We have used various data collection methods to facilitate our research. Various scholarly resources have been referred to develop arguments for our thesis and for understanding work already done in this field. Further, we have studied various practical experimentations done across the globe that aim at studying in detail the tools for speculating emergent behavior of SoS. Though, we have no preconceived notions for this research, however, it mandates the mentioning of our planned and expected outcomes. Our thesis currently rests on the idea emergent behavior of SoS poses one of the most serious threats to any system and systems engineers and must be handled tactfully. This can be successfully done with the help of agency modeling tools and simulation technology, which is capable of projecting forecasts and estimates of how a certain SoS might react in future. Proactive approach can be developed to be well-prepared for these consequences beforehand. Research summary and results The study area represented a large-scale engineered power grid situated in North America consisting of backbone and local transmission lines running along 680,000 miles and 2.5 million miles in length respectively. It has typical electric power components including generators, transformers and substations. Figure 3: Representation of a model of an abstracted power grid system of systems We have conducted few limited tasks in this experiment due to lack of resources and other limitations. A prototype of power grid has been modeled as presented in Figure 3, showing deregulated operations where trading agents are independently working to earn maximum profits. Figure 3 represents a simple prototype of a highly complex power grid system. It consists mainly of five power generating nodes, three consumer nodes, three trading agents and transmission lines acting as a connection between power generating nodes and consumer nodes. The function of power generating nodes is to produce certain power of electricity that can be offered at specific prices, varying for each node. Correspondingly, consumer nodes are products that demand a certain power of electricity for their regular consumption at a certain price, varying for each node. On the other hand, transmission lines carry electricity loads according to their distinct capacities and probabilities of power or transmission failure. The trading agents seek to act as a bridging gap between the two parties, seeking to earn maximum agency commission by purchasing service at the least cost and selling it at the highest price. If any of the generating nodes sells off units of electricity, it manages demand and limited supply by increasing its selling prices for future customers. On the contrary, if it fails, the selling prices are reduced to attract demand. Consumer nodes also behave in a predictable manner, reducing their acceptable purchasing prices if they succeed in negotiating on their purchasing powers and vice versa. These behaviors pertain to the individual components (systems) of the entire power grid (system of systems). Finally, the emergent behavior of the power grid as a whole is evaluated by conducting comparative analysis between the prices and transmission system failure rates for any given model in the presence and in the absence of a trading agent. One of the emerging behaviors that occur involves an irregular behavior depicted by one of the trading agents who attempts at creating an artificial scarcity of units of electricity at one of the customer nodes so as to gain from the radical boosts in demand through increased sales or otherwise raising prices. Trading agent(s) obstructs the transmission n lines to consumer node 2 for a short period, creating a widening gap between supply and demand. As a result, shortage of electricity occurs at node 2 and consumers are now willing to pay higher than before for same number of units as electricity is now dearer to them. On the other hand, a SoS without trading agents demonstrates a very different picture, with the cost for a regulated power grid much lower than the average price set by trading agents in former scenario. The actual goal of the power grid or the SoS as a whole is to supply uninterrupted electricity service to consumer nodes at reliable and affordable prices. Because the individual goals of components of SoS are very distinct, it causes disturbances in the overall performance of the grid, affecting the ultimate customer. Further problem sweeps in when there are more than one service provider of electricity for same league of customers. As a result, same power grids are being utilized by different vendors who own different systems within the SoS. If a single system owned by some specific vendor falters, the other systems will suffer as well. Thus, the SoS, initially configured under one set of operating principles, now needs to adapt multiple sets of principles. This leads to emergence of new behaviors such as increase in frequency of power failures, technical faults in grids and increasing costs of providing services to customers. In order to successfully analyse these emerging activities, we adopted an approach to evaluate the overall dynamic behavior of SoS itself. We utilized modeling techniques and simulation to replicate important aspects of the SoS, its components and the transmission and interaction amongst them. Several methodologies can be adopted for modeling the complex networks of SoS (Oliver et al., p. 168). We developed an engineering approach that was mainly targeted at drawing concrete perceptions about interactions and interfaces. We analysed the system, developed some modifications through the usage of simulation and modeling and compared the exiting power grid module to the prototype prepared by us. Few changes in the infrastructure were capable of generating highly positive results. However, since we were not able to diagnose all symptoms of emergent behavior of SoS, we only aimed at rectifying ones we could easily identify (Osmundson and Huynh, p.12). Every component has a certain load bearing capacity and when carrying it altogether, the collective system exhibits entirely different pattern of loading, which is determined by the operating policy of this system. This policy prescribes short term instructions for guidance such as power dispatch rules etc which ought to help the management in fulfilling needs of customers in terms of low costs. In turn, this policy is prepared and motivated through aggregate demand of customers for power services at different substations (Chassin et al., p.23). Short-term customer demand is based on routine consumption and influenced by regular cyclical fluctuations while long-term demand is a function of density of population in any particular territory. Any geographic shifts or changes in population patterns, discovery of sustainable alternative energy sources or otherwise economic changes in local industry or market are factors that affect the long-term demand curve. If any of the components is carrying load more than its allowed capacity, then it shall trip or eventually lose functionality. This affects the whole power grid by limiting the loading capacity of the entire network. Other adverse events include transformer failure due to overburdening with power load and re-dispatching of operator enforcing a limitation on power flow in transmission lines to reach its maximum allowed rating. As a result, power flow gets redistributed amongst the network components, causing few nodes to carry load more than designated according to their capacity. This creates a spiral multiplier effect, also known as ‘cascade effect’: breakdown of one node leads to overburdening of second node which also fails and so on, eventually bringing down the whole system to a collapse. If this situation occurs causing the load at substations to be limited or zeroed, then the SoS is said to have faced a ‘blackout’. A power grid’s function is to simply generate and supple electrical power to all household and business units. To ensure that power outage doesn’t occur, this SoS must continuously balance the sharing of load amongst the grid components and regulate power generation such that the output equates the availability and capability to carry load. This is controlled through various measures such as tuning the generation output and modifying the configuration of SoS through turning switches on or off as they case may be. Every component is interdependent on the activities carried out by other member constituents of SoS and therefore close coordination is essential. In case of mismanagement, failure of any single element has multiple fold repercussions, bringing the entire system and transmission lines running across a vast geographical area to a halt. Utility companies, running power grids, around the world, have established voluntary control areas which act as a source of reduction in overall supervision costs by tailoring power generation in each area to match the prevailing demand in different territories. These units have better visibility and short span of control which enables them to monitor closely and influence power generation according to needs. Cooperation and coordination plays a vital role since transmission of electricity spread across large geographical areas, requires smart operations and due diligence. Hypothesis support From the results mentioned above in detail, it is safe to consider this experimentation to be a success in favor of our initially established thesis. Our thesis was composed of two agendas. First was the complexity and significance of problem chosen. Emergent behavior is suggested to be one of the most drastic effects that SoS face duet to interaction of its elements. It is highly unpredictable and difficult to control. As a result, the whole system is susceptible to be on verge of breakdown or ultimate shutdown. The scenario elucidated above shows clearly that due to interaction between trading agents and customer nodes, the system as a whole exhibited a behavior which caused severe consequences for the SoS and pushed it into adverse circumstances. The second part related to the measures that can help overcome this problem. A detailed discussion and testing through agent modeling techniques showed that systems engineers can easily mitigate the unpredictability and complexity of emergent behavior through prototyping and simulation technologies. Moreover, this methodology shall enable researchers and engineers to benefit from the dynamics of emergent behavior of any SoS, deriving synergy, cost effectiveness and customer satisfaction as ultimate achievements. Therefore it is only fair to presume that this research favors our hypothesis and strengthens our initial stance, backed up with evidences and supporting studies. Conclusions The significance of SoS in today’s dynamic environment is of value for industries and business entities. However, to address their concerns regarding the demerits and obstacles faced in this regard, systems engineers must adopt techniques and methods that enable them to analyse these obstacles including anticipation of emergent behavior of SoS as a whole and achieving desirable results that are not departing from initially planned outcomes. Agent modeling techniques, simulation runs and other analytical tools are useful for modeling SoS in executable form and developing a comprehensive understanding of nature of interaction between components of SoS and learning from results of experiments. We can comfortably conclude from given evidences and results that emergent behavior of SoS needs to be controlled and managed effectively in order to mitigate the potential risks that it carries while capitalizing on the benefits that it has to offer for the system as a whole. These benefits can be derived with the help of modeling and simulation techniques that can be deployed by systems engineers to be in a better position to anticipate the behavior and respond proactively, steering the behavior in favor of the interests of the component systems and SoS as a whole. Since our research was aimed at one aspect only, therefore we cannot reliably conclude that this is an exhaustive research which covers all areas regarding this field. Many other factors come into play when SoS works in a natural environment with no controlled features and isolated facilities. Many surrounding factors affect the system and its behavior which couldn’t be easily studied in this dissertation due to lack of resources and impracticability. Future work recommendations This research was carried out in a controlled and isolated environment so that only those factors can be focused on which are basis of our thesis. As a result, we were unable to report on the entire picture of any SoS and its working in a natural setting where its components interact with number of other elements outside the SoS which were not present in this study. Therefore, this research need not be considered as a complete dissertation and further work must be done, if practicable, to understand the dynamics of SoS in the free world. The emergent behavior as demonstrated in current scenario may not be the same in different circumstances and thus with better access to databases and management of a real-life SoS and technological resources such as simulation software and modeling facilities, a more tangible conclusion can be drawn to support this thesis effectively. Final summary SoS is a complex network of systems that individually have different characteristics but collectively display entirely distinct behavior. This behavior emerges due to synergetic effect of interaction between these systems and components of SoS. This behavior is so strong that it is capable of ripping the entire system apart through cascade effects. High coordination among stakeholders and components and distributed control are required by system engineers to manage it. Researching and harnessing on this issue is vital for SoS to be successful in achieving its primary purposes. Agent modeling devices and simulation settings are prime tools that can be used to mitigate consequences connected with emergent behavior by speculating it beforehand and developing strategies to undermine its effects. If controlled effectively through these engineering tactics, the SoS has high potential to achieve levels of flexibility, cost leadership and scalability that cannot be possibly attained via conventional systems. This paper shall be a source of great contribution towards SoSE Body of Knowledge since it addresses the most crucial obstacle faced by SoS and due to its highly radical and dynamic nature, still to date the researchers and engineers have not been able to fully attain grip and certainty over projecting behavioral patterns and interactive process. Since it is a replication technique and small scale research, only limited assurance can be assumed on results achieved. Nonetheless, further research is required in this field to be in a better position to draw conclusions. Specific research results Many scholarly resources have been studied while working on this dissertation and most of them have been a source of rational arguments that contribute to my thesis. Systems of systems practically exist, current or planned, in all major sectors and industries including aviation, transportation, power generation, healthcare, utilities management, manufacturing concerns, construction of infrastructures or equipment and enterprise systems etc. (DG INFSO G3, n.p.). Due to its complexities, extensive research has been carried out in the past to address the general controls and systems involved in SoS. Emerging behaviour of SoS is followed by enhanced competences and functional capabilities through synergetic effects of several systems working together with unity of direction and command (Parisini, p. 249). The most interesting fact revealed by previous researches precludes that the behaviour exhibited by any SoS is entirely distinct and mutually exclusive and its dynamics cannot be readily projected simply by perceiving the behaviour of smaller member systems (Bar-Yam et al., n.p. ). Emergent behaviors result from the direct interactions at interfaces amongst systems and their indirect interactions with operators and software agents. As Klir suggested, we can assume that each of these interface components attempt to achieve an objective, which is regulated through rules established by interaction of SoS with its constituent systems (Klir, p.134). Fisher suggested that emergent behavior of SoS must be managed effectively to avoid the plagues of interoperability issues, cost ineffectiveness, unsatisfactory performance and deterioration in quality of services. Unfortunately, most of techniques deployed by engineers to exploit emergent behavior have proved out to be inefficient and useless (Fisher, p. 14). Every SoS as well as its components have certain stakeholders who are interested in behaviors and activities displayed by them. They have their own set of unique objectives and expectations regarding how the system should actually behave, which must be managed along with other things in order to keep the system intact and ensure its smooth operations. This is turn is possible if overall behavior can be controlled in steered in direction that not only meets stakeholders’ expectations but also is beneficial for system as a whole (Department of Defense, p. 12). Systems engineers face a major obstacle of speculating and evaluating emergent behavior exhibited by SoS, especially if it leads to adverse consequences (Osmundson and Huynh, p. 1). Taking the current example and its significance in real world, power grid has a tendency to face rare non-periodic outages, as seen by historical examples of collapse of the Canadian power system in the Quebec province in 1989 and nationwide power disruption in the eastern U.S. in 1993 affecting millions of people. This can similarly happen in power grid located in North America, rising from simple failure of a single component of the whole SoS. These disruptions are very harmful for SoS, especially those that are a product of ‘self-organized criticality’. A self-organized criticality system refers to SoS whereby the overall average system state has been reached in the presence of perturbations, adding up to major outages (Carreras et al., p. 26). References Bankes, Steven C. Agent-based modeling: A revolution? Proceedings of the National Academy of Sciences of the United States of America 99.Suppl 3 (2002): 7199-7200. Print. Bar-Yam Y. et al. The characteristics and emerging behaviors of system-of-systems. Complex physical, biological and social systems project report, New England Complex Systems Institute, January 2004. Web. Extracted from: http://necsi.org/education/oneweek/winter05/NECSISoS.pdf. Brenner, T. and Werker, C., A. “Taxonomy of Inference in Simulation Models”, Computational Economics, 30, 2007: 227-244. Print. Carreras, B. A., D. E. Newman, I. Dobson and A. B. Poole, “Initial Evidence for Self-Organized Criticality in Electric Power System Blackouts,” Proceedings of Hawaii International Conference on System Sciences, January 4-7, 2000, Maui, Hawaii. 2000 IEEE. Print. Chassin, D. P., N. Lu, J. M. Malard, S. Katipamula, C. Posse, J. V. Mallow and A. Gangopadhyaya, Modeling Power Systems as Complex Adaptive Systems, U.S. Department of Energy Contract DE-AC06-76RL01830 Report, December, 2004. Print. Department of Defense. Systems Engineering Guide for Systems of Systems, Version 1.0, Office of the Deputy Under Secretary of Defense for Acquisition and Technology, Systems and Software Engineering, Washington DC: ODUSD(A&T)SSE, 2008: p. 1-135. Print. DG INFSO G3. Report of a workshop on systems of systems (working draft), Brussels, 21 September 2009. Web. Extracted from: ftp://ftp.cordis.europa.eu/pub/fp7/ict/docs/esd/workshop-report-v1-0_en.pdf. Dyson, George. Darwin among the machines: The evolution of global intelligence. Basic Books 1997. Print. Federal Energy Regulatory Commission (FERC) Order 888, April 24, 1996. Print. Fisher, David. “An emergent perspective on interoperation in systems of systems.” 2006 p. 1-51. Print. Hsu, John C., and Marion Butterfield. “Emergent Behavior of Systems-of-Systems.” Proceedings of INCOSE , 2009. Print. Huynh, Thomas V., and John S. Osmundson, “A Systems Engineering Methodology for Analyzing Systems of Systems Using the Systems Modeling Language (SysML)”, Proceedings of the 2nd Annual System of Systems Engineering Conference, Ft. Belvoir, VA, sponsored by the National Defense Industrial Association (NDIA) and OUSD AT&L, 25-26 July, 2006. Print. Jamshidi M. (Ed.). System of Systems Engineering: Innovations for the 21st Century. New York: John Wiley & Sons, 2009. Print. Klir, George J., Facets of Systems Science, Plenum Press, New York 1991. Print. Maier, M., “Architecting principles of systems-of-systems.” Systems Engineering (1998). 1(4): p. 267 - 284. Print. Oliver, David W., Timothy P. Kelleher and James G. Keegan, Jr., Engineering Complex Systems with Models and Objects, McGraw-Hill, New York, 2007. Print. Osmundson, John, and Thomas Huynh, “Systems-of-Systems (SOS) Systems Engineering”, Proceedings of the System of Systems Engineering Conference, Johnstown, PA, sponsored by the National Defense Industrial Association (NDIA) and OUSD AT&L, June 13-14, 2005. Print. Osmundson, John S., Thomas V. Huynh, and Gary O. Langford. “Emergent behavior in systems-of-systems.” Proceedings of the 2008 INCOSE International Symposium (2008): p. 1-12. Print. Parisini T.. “Control systems technology: Towards a systems-of-systems perspective.” IEEE Trans. on Control Systems Technology, vol. 18, no. 2, p. 249, March 2010. Print. Parunak, H. Van Dyke, and Raymond S. VanderBok. “Managing emergent behavior in distributed control systems.” Ann Arbor 1001 (1997): 48106, p. 1-8. Print. Rollings, Andrew, and Ernest Adams. Andrew Rollings and Ernest Adams on game design. New Riders Pub, 2003. Print. Read More