System of Systems (SoS
H.Kopetz September 2012
• • • • • •
Introduction Definition of an SoS Problem Solving Emergence SoS Research Topics Conclusion
• The global availability of communication services makes it possible to connect independently developed systems to provide new synergistic services and more efficient economic processes. • In the Internet-of-Things (IoT) embedded systems form the leaves of the (enhanced) Internet. We call this common approach System-of-Systems (SoS) approach
. • Since the core issue in all these systems is the same—the integration of nearly autonomous systems to solve stated problems—a common approach to the design of all these systems should be established.
: • Internet of Things • Network centric systems • Widely Distributed Embedded Systems • Sensor Networks • Cloud Computing • Web Services • Service Oriented Architecture (SOA) • Enterprise information systems • E commerce .g.. denoted by different names. e.Scope of Systems of Systems
The integration of autonomous system is occurring in different domains.
Definition (1): System of Systems (SoS)
An SoS is an integration of a finite number of constituent systems (CS) which are independent and operable and which are networked together for a period of time to achieve a certain higher goal. (from State of the Art Report T-Area-SoS. May 2012) Maier’s criteria (1998) on properties of an SoS: • Operational independence of component systems • Managerial independence of component systems • Geographic distribution • Emergent behavior • Evolutionary development process
.Constituent Systems of an SoS
An SoS can consist of three types of constituent systems: • Physical Systems. like physical part of a machine • Cyber Systems. like the control system • Humans Since in general the human behavior cannot be completely modeled—it is impossible to model the conceptual landscape of a human— the behavior of an SoS with human involvement is unpredictable.
An omniscient external observer. who has a global view of all interactions. • Interactions in the Physical Space (PhS). energy or human intervention (PhSInteractions). is normally not available. seeing only the interactions with its immediate partners. Any constituent system of an SoS has only a local view of the interactions.
. realized by the exchange of messages (CS-interactions). realized by the flow of material.Interaction within an SoS
There are two types of interactions possible among the PSSs that form an SoS • Interactions in Cyber space (CS).
Control Evolution Testing Implement..Fundamental Paradigm Shift in SoS Design
Characteristic Scope of System Requirements and Spec. Technology Faults (Phys. Test phases Given and fixed Exceptional Controlled Process model New-SoS Not known Changing Distributed Uncoordinated Continuous Unknown Normal Accidental ???
. Design) Emergence System development Old-Classic Fixed (known) Fixed Central Version contr.
.Definition (2): System of Systems (SoS)
A System of Systems (SoS) is a problem solving system (PSS) where some actions that are needed to arrive at a solution to a problem are delegated from the system that interfaces with the problem owner to other autonomous problem solving systems (PSS). conform to different architectural styles. and can operate on any scale from fully automatic to control by humans. of heterogeneous technology. • An autonomous PSS may reject a request from the problem owner (e.g. • The autonomous PSSs (or constituent systems) that participate in an SoS can be under different management control.. geographically distributed. due to insufficient incentives).
. Finding a future concrete state that satisfi cing 2.Problem Solving Systems (PSS)
Problem: an undesirable state. Problem solution: a (future) state that is satisficing Problem Solving consists of two steps: 1. Transforming a system from the current state to this identified future state within a given time limit.
Design Issues: • How to specify goal states (interface specification)? • How to ensure that the provided solution is satisficing? • How to find or design a process that transforms the current state to the goal state? • How to design an incentive structure such that an autonomous system will contribute to finding a solution? • How long shall we wait?
An PSS rests on the insight that it is much easier to specify a goal state—a problem solution— than to specify a process that leads from the current state to the goal state.
Alternatives: • Hiking with skis • Riding a snow-cat • Flying a helicopter The decision which alternative (way) to follow and which resources (means) are required is delegated to a CS.Example: Avalanche Alarm
When an avalanche alarm has been received by a rescue station. Only the goal—bring the rescue team to the site within a short duration—is specified at the SoS level.
. a rescue team must be brought to the site within a short duration.
.Problem Solving: End. and Means
Problem solving is about achieving a stated goal—(the end state)— by selecting a strategy—(the way)— using resources—(the means). Ways.
non-linearity. • In many situations it is difficult to establish the bounds of an SoS—therefore it is not possible to know about all indirect and hidden interactions. • Even if the totality of interactions in an SoS were known. or behavior) called emergent that cannot be attributed to any single PSS.Emergent Qualities
• The CS and PhS interactions within the SoS can cause new qualities (structure. • Some emergent qualities are caused by positive or negative feedback. the involved CS and the PhS.
. and time lags caused by the interaction. it would be difficult to predict the appearing emergent qualities without a detailed understanding of the concrete application context. concerning both. properties.
) • The role of time • Enabling conditions at the micro-level that lead to emergence • Upward causation (stochastic?) • Downward causation (systemic?) • Supervenience of Macro qualities on Micro level • Wholism
. structure. patterns etc.Issues in the Study of Emergence
• Distinction between Base (Micro) level and Macro level • Interactions of components in physical and/or cyber space • Rules governing the novel phenomena at the Macro level (properties. behavior.
• • • • • • •
System Evolution Interface Specification Governance Trust and Dependability Autonomy Problem Specification Emergence
.. Aspect oriented system perspectives. The uncoordinated evolution of the CSs that form an SoS puts many of the established systemsengineering principle in questions: e. Topics: Finding stable services and relied upon interfaces among SoS. etc.g. system design is never completed. Validation of an SoS.System Evolution
Description: A CS is tightly integrated with its environment and must continually adapt and evolve in order to remain relevant in an ever changing world. Adaptation. Concurrency. the concept of correctness changes with time. there is no authoritative static specification.
Topics: Interface Placement. Interface Specification at syntactic and semantic level in the domains of value and real time behavior. Heterogeneity. computer supported design.Interface Specification
Description: The placement. Semantic Interface modeling and simulation. Co-design. Data semantics. tool support. Disappearing Human-Computer Interfaces (HCI). specification and design of the interfaces between the CSs and the problem owner (user) is at the very core of SoS technology.
. Management of cognitive complexity.
Topics: Definition of spheres of control. the relied upon interfaces between the CSs must be established and controlled by some authority. ownership control. governance rules. However. e.Governance
Description: Since each CS is in its own independent governance domain there is no central control over the overall SoS behavior. Establishment of incentives and a reward structure to bring about the cooperation of CSs..
. administrative control.g.
Fault Tolerance. Topics: Security. Certification.
. Hardware based security. Robustness. Non-repudiability. safety. Validation. anomaly (Intrusion) detection.Trust and Dependability
Description: Establishing trust that a selected CS will not abuse the received information and ensuring that the SoS services are robust in the face of occurring failures are prime challenges of SoS design and operation.
energy harvesting.. RFID.
. Context awareness.
Topics: Self-Awareness. power electronics).g. Energy efficiency. sensors and actuators (e. Resource control within an CS. Self-organization. NFC.Autonomy
Description: Every CS that contributes to an SOS is autonomous with its own governance. biomedical sensors. smart sensors.
Topics: Requirement engineering. Abstract versus concrete problem specification.Problem Specification
Description: Problems are normally formulated at an abstract level.
. Service and solution discovery. The concrete problem specification is guided by a search in the solution space and must be supported by an SoS. cloud services. Reuse of existing services.
an ill-structure and not well understood topic that needs fundamental research. Topics: This is.Emergence
Description: The interaction of independent CSs can result in new properties at the system level that are not present at the level of any individual CS. Bio-mathematical modeling and simulation.
. by system wide simulation.g. These new properties are called emergent properties.. at present. e.
Need for New Design Rules for SoS
There is an urgent need to study and formulate a set design rules for SoSes in future research actions.
.: • How to express and specify abstract and concrete problems and interface properties? • How to specify and guarantee temporal properties? • How to modularize a design and handle continuous evolution? • How to achieve runtime integrity by self-observation and react to normal failures (hardware. humans)? • How to achieve security by design and establish trust among the subsystems? • How to resolve data inconsistency. both at the syntactic and semantic level? Practical experience with the design rules should be acquired in prototype environments. e. design.g.
Conclusion: New Concepts Needed
If the set of components that form a System-of-Systems interact intensively the observed emergent phenomena can sometimes only be described understandably by introducing radically new concepts in operation and design. This conceptual novelty can range from the invention of a new term to the introduction of an entirely new theory.