Flexible and Intelligent Learning Architectures for SoS (FILA-SoS) Volume 3 – Fuzzy Genetic Optimization Model
Enterprises and System of Systems
Report Number: SERC-2015-TR-021-4
Publication Date: 2015-02-28
Project: Flexible Intelligent Learning Architectures for Systems of Systems (FILA-SoS)
Dr. Cihan Dagli
Dr. Nil Kilicay-Ergin
Dr. David Enke
Dr. Dincer Konur
Dr. Ruwen Qin
Dr. Abhijit Gosavi
Multi-faceted systems of the future will entail complex logic and reasoning with many levels of reasoning in intricate arrangement. The organization of these systems involves a web of connections and demonstrates self-driven adaptability. They are designed for autonomy and may exhibit emergent behavior that can be visualized. Our quest continues to handle complexities, design and operate these systems. The challenge in Complex Adaptive Systems design is to design an organized complexity that will allow a system to achieve its goals. This report attempts to push the boundaries of research in complexity, by identifying challenges and opportunities. Complex adaptive system-of-systems (CASoS) approach is developed to handle this huge uncertainty in socio-technical systems. Although classically (Dahmann, Rebovich, Lowry, Lane, & Baldwin, 2011) four categories of SoS are described in literature namely; Directed, Collaborated, Acknowledged and Virtual. However, there exist infinitely many SoS on the edges of these categories thus making it a continuum. Many SoS with different configurations can fill this gap. These four types of SoS vary based on their degree of managerial control over the participating systems and their structural complexity. The spectrum of SoS ranges from Directed SoS that represents complicated systems to Virtual SoS that are complex systems. Acknowledged SoS lie in between this spectrum. This particular SoS is the focal point of our research endeavor. Acknowledged SoS and Directed SoS share some similarities such as both have (Dahman & Baldwin, 2011) SoS objectives, management, funding and authority. Nevertheless, unlike Directed SoS, Acknowledged SoS systems are not subordinated to SoS. However, Acknowledged SoS systems retain their own management, funding and authority in parallel with the SoS. Collaborative SoS are similar to Acknowledged SoS systems in the fact that systems voluntarily work together to address shared or common interest. Flexible and Intelligent Learning Architectures for SoS (FILA-SoS) integrated model is developed in this research task provides a decision making aid for SoS manager based on the wave model. The model developed called the FILA-SoS does so using straightforward system definitions methodology and an efficient analysis framework that supports the exploration and understanding of the key trade-offs and requirements by a wide range system-of-system stakeholders and decision makers in a short time. FILA-SoS and the Wave Process address four of the most challenging aspects of system-of-system architecting:
1. Dealing with the uncertainty and variability of the capabilities and availability of potential component systems
2. Providing for the evolution of the system-of-system needs, resources and environment over time
3. Accounting for the differing approaches and motivations of the autonomous component system managers
4. Optimizing system-of-systems characteristics in an uncertain and dynamic environment with fixed budget and resources
Some of the highlights of FILA-SoS are listed in terms of its capabilities, value added to systems engineering, ability to perform “What-if Analysis”, modularity of integrated models, its potential applications in the real world and future additions to the current version.
FILA-SoS has a number of unique capabilities such as integrated model for modeling and simulating SoS systems with evolution for multiple waves. It also has modularity in the structure where the models can be run independently and in conjunction with each other. Besides there are a couple of different models for both architecture generation and SoS behavior and various individual system behavior negotiation models between SoS and individual systems. In terms of value added FILA-SoS aids the SoS manager in future decision making. It also helps in understanding the emergent behavior of systems in the acquisition environment and impact on SoS architecture quality. FILA-SoS serves as an artifact to study the dynamic behavior of different type of systems (non-cooperative, semi-cooperative, cooperative). It enables us to identify intra and interdependencies among SoS elements and the acquisition environment. FILA-SoS can provide a “What-if” Analysis depending on variables such as SoS funding and capability priority that can be changed as the acquisition progresses through wave cycles. It has the ability to simulate any architecture through colored petri nets. In addition, it can simulate rules of engagement & behavior settings: all systems are non-cooperative, all systems are semicooperative, and all systems are cooperative or a combination. Some of the potential applications include modeling a wide variety of complex systems models such as logistics, and cyber-physical systems. It also acts as a test-bed for decision makers to evaluate operational guidelines and principles for managing various acquisition environment scenarios. Future Capabilities that are currently in progress are extending the model to include multiple interface alternatives among systems and incorporation of risk models into environmental scenarios. 2 Integrated Model Structure for FILA-SoS Version 1.0 is described. It provides a short description of all independent models that make up the FILA-SoS integrated model and reports the workings of the model with three notional System-of-Systems namely; Toy Problem for aircraft carrier performance assessment, ISR (intelligence surveillance and reconnaissance) and SAR (search and rescue). The project reports span 17 volumes. Each report describes the various aspects of the FILA-SOS integrated model:
Volume 1: Integrated Model Structure Volume 1 is the Integrated Model Structure report for FILA-SoS Version 1.0. It provides a short description of all independent models that make up the FILA-SoS integrated model. Integrated FILA-SoS developed is tested in three notional System-of-Systems namely; Toy Problem for Aircraft Carrier Performance Assessment, ISR (intelligence surveillance and reconnaissance) and SAR (search and rescue). FILA-SoS integrated model is currently being validated with a real life data from a medium sized SoS. The results of this validation are given in volume 17.
Volume 2: Meta-Architecture Generation Multi-Level Model Volume 2 describes Meta-Architecture Generation Multi-Level Model. The multi-level metaarchitecture generation model considers constructing an SoS architecture such that each capability is provided by at least one system in the SoS and the systems in the SoS are able to communicate with each other. Secondly, it has multiple objectives for generating a set of SoS architectures namely; maximum total performance, minimum total costs and minimum deadline. Finally, the model establishes initial contracts with systems to improve performances.
Volume 3: Fuzzy-Genetic Optimization Model Volume 3 illustrates the second meta-architecture generation model known as the Fuzzy-Genetic optimization model. This model is based on evolutionary multi-objective optimization for SoS architecting using genetic algorithms and four key performance attributes (KPA) as the objective functions. It also has a type-1 fuzzy assessor for dynamic assessment of domain inputs and that forms the fitness function for the genetic algorithm. It returns the best architecture (metaarchitecture) consisting of systems and their interfaces. It is a generalized method with application to multiple domains such as Gulf War Intelligence/Surveillance/Reconnaissance Case, Aircraft Carrier Performance Assessment Case and Alaskan Maritime Search and Rescue Case.
Volume 4: Architecture Assessment Model Volume 4 describes an Architecture Assessment Mode that can capture the non-linearity in key performance attribute (KPA) tradeoffs, is able to accommodate any number of attributes for a selected SoS capability, and incorporate multiple stakeholder’s understanding of KPA’s. Assessment is based on a given meta-architecture alternative. This is done using type-1 fuzzy sets and fuzzy inference engine. The model provides numerical values for meta-architecture quality.
Volume 5: Cooperative System Negotiation Model Volume 5 specifically describes the Cooperative System Negotiation Model. The systems following this model behave cooperatively while negotiating with the SoS manager. The model 3 of cooperative behavior is based on agent preferences and the negotiation length. Each system agent has two inherent behaviors of cooperativeness: Purposive (normal behavior) and Contingent (behavior driven by unforeseen circumstances). The approach models the tradeoff between the two behaviors for the systems. A fuzzy weighted average approach is used to arrive at the final proposed value.
Volume 6: Non-Cooperative System Negotiation Model Volume 6 goes on to describe the Non-Cooperative System Negotiation Model in which systems behave in their self-interest while negotiating with the SoS coordinator. A mathematical model of individual system’s participation capability and self-interest negotiation behavior is created. This methodology is an optimization-based generator of alternatives for strategically negotiating multiple items with multiple criteria. Besides, a conflict evaluation function that estimates prospective outcome for identified alternative is proposed.
Volume 7: Semi-Cooperative System Negotiation Model Volume 7 describes the third and last system negotiation model, which illustrates the SemiCooperative System Negotiation Model. It exhibits the capability of being flexible or opportunistic: i.e., extremely cooperative or uncooperative based on different parameter values settings. A Markov-chain based model designed for handling uncertainty in negotiation modeling in an SoS. A model based on Markov chains is used for estimating the outputs. The work assigned by the SoS to the system is assumed to be a ``project’’ that takes a random amount of time and a random amount of resources (funding) to complete.
Volume 8: Incentive based Negotiation Model for System of Systems Volume 8 explains the SoS negotiation model also called the Incentive Based Negotiation Model for System of Systems. This model is based on two key assumptions that are to design a contract to convince the individual systems to join the SoS development and motivate individual systems to do their tasks well. Game theory and incentive based contracts are used in the negotiation model that will maximize the welfare for parties involved in the negotiation. SoS utility function takes into account local objectives for the individual systems as well as global SoS objective whereas the incentive contract design persuades uncooperative systems to join the SoS development.
Volume 9: Model for Building Executable Architecture Volume 9 illustrates the process of building Executable Architectures for SoS. The operations of the SoS is a dynamic process with participating system interacting with each other and exchange various kinds of resources, which can be abstract information or physical objects. This is done through a hybrid structure of OPM (Object process methodology) and CPN (Colored petri nets) modeling languages. The OPM model is intuitive and easy to understand. However, it does not support simulation, which is required for accessing the behavior related performance. This is achieved by mapping OPM to CPN, which is an executable simulation language. The proposed method can model the interactions between components of a system or subsystems in SoS. In addition, it can capture the dynamic aspect of the SoS and simulate the behavior of the SoS. Finally, it can access various behavior related performance of the SoS and access different 4 constitutions or configurations of the SoS which cannot be incorporated into the metaarchitecture generation models of Volume 2 & 3.
Volume 10: Integrated Model Software Architecture and Demonstration FILA-SoS Version 1.0 Volume 10 elucidates the Integrated Model Software Architecture and Demonstration based on the models described above. Volume 11 and thereon the reports are aimed at the upcoming newer version 2.0 of FILA-SoS.
Volume 11: Integrated Model Structure FILA-SoS Version 2.0 Volume 11 provides Integrated Model Structure for FILA-SoS Version 2.0 that could be implemented in a new software environment.
Volume 12: Complex Adaptive System-of-System Architecture Evolution Strategy Model for FILA-SoS Version 2.0 Volume 12 provides a model to answer the first research question “What is the impact of different constituent system perspectives regarding participating in the SoS on the overall mission effectiveness of the SoS?” It is named the Complex Adaptive System-of-System Architecture Evolution Strategy Model and is incorporated in FILA-SoS Version 2.0. This volume describes a computational intelligence based strategy involving meta-architecture generation through evolutionary algorithms, meta-architecture assessment through type-2 fuzzy nets and finally its implementation through an adaptive negotiation strategy.
Volume 13: On the Flexibility of Systems in System of Systems Architecting: A new MetaArchitecture Generation Model for FILA-SoS Version 2.0 Volume 13 is termed the Flexibility of Systems in System of Systems Architecting: A new MetaArchitecture Generation Model for FILA-SoS Version 2.0. The research question is answered through an alternative technique to meta-architecture generation besides the one described in Volume 2.
Volume 14: Assessing the Impact on SoS Architecture Different Level of Cooperativeness: A new Model for FILA-SoS Version 2.0 Volume 14 proposes a new method for Assessing the Impact on SoS Architecture Different Level of Cooperativeness. Second research question is answered through a model that allows different levels of cooperativeness of individual systems.
Volume 15: Incentivizing Systems to Participate in SoS and Assess the Impacts of Incentives: A new Model for FILA-SoS Version 2.0 Volume 15 is an extension of previous systems negotiation models based on incentivizing and is aptly called Incentivizing Systems to Participate in SoS and Assess the Impacts of Incentives: A new Model for FILA-SoS Version 2.0. It also provides an approach to answer the third research question “How should decision-makers incentivize systems to participate in SoS, and better understand the impact of these incentives during SoS development and effectiveness?”. This model is based on the fact that providing incentives only depending on the outcome may not be enough to attract the attention of the constituent systems to participate in SoS mission. Therefore, this model extends the approach as described in Volume 8 while considering the 5 uncertainty in the acquisition environment. The incentive contract is designed based on the objectives of the SoS and the individual systems. Individual system’s objective is to secure highest incentives with minimal effort while the SoS manager’s goal is to convince individual systems to join the SoS development while maximizing its own utility.
Volume 16: Integrated Model Software Architecture for FILA-SoS Version 2.0 Volume 16 gives an overview of the integrated model architecture in version 2.0 of the software. It includes all old and new models previously mentioned. Volume 17: FILA-SoS Version 1.0 Validation with Real Data Volume 17 describes the validation of the FILA-SoS Version 1.0 with a real life data provided by MITRE Corporation by from a moderately sized SoS.