The objective of AMADEOS is to bring time awareness and evolution into the design of System-of- Systems (SoS), to establish a sound conceptual model, a generic architectural framework and a design methodology, supported by some prototype tools, for the modeling, development and evolution of time-sensitive SoSes with possible emergent behaviors. Special emphasis is placed on evolution, emergence, dependability (e.g. safety, availability) and security, considering embedded devices and the cloud as the execution platform.
The concept of evolution will be addressed from two complementary perspectives, considering both long-term evolution and short-term unexpected changes (e.g., failures) in the constituent systems. The project starts with a study of fielded industrial SoSs, where the handling of time and the evolution aspects will be in the center of the analysis, in the domains of disaster management, transport, and smart grid applications.
The following development of the conceptual model, the architectural framework, the design methodology and some extensions to UML-based tools will form the core of the project work. In place of the traditional guarantees that were the target for more closed and static systems, the architectural framework will be based on the concept of guaranteed best adaptation under the given constraints, sometimes just monitoring how the environment evolves, and influencing how the SoS takes mitigating actions.
The viability of the framework will be validated on a case study of a CPS, a small smart grid application, where guaranteed responsiveness, evolution, dependability and security are essential requirements. The research is based on the in depth experience of some of the key researchers of the consortium in the fields of architecture design, real-time systems, dependability, security and the development of large systems-of systems in such diverse domains as disaster management, the transport sector, and energy distribution.
Scientific and technical challenges
AMADEOS does not intend to address SoSs as a whole, but it rather focuses on the following open scientific and technical challenges:
- Guaranteed Responsiveness. In the prevalent timeless SoSs it is difficult to guarantee strict deadlines that are required in many CPS applications at the SoS level, e.g., the smart grid. Since each constituent system uses its own unsynchronized clock, one has to be extremely careful about assumptions on the execution speeds and message delivery delays of the CSs in a timeless SoS. Therefore, the distributed protocols underlying cooperation and interaction in a SoS are known to be more complex, handling device failures can be extremely challenging and the temporal specification of the RUMIs is imprecise or even missing.
- Dynamicity and Evolvability. SoSs must cope with short-term discontinuities as well as with long-term adaptation. We refer to the former as SoS dynamicity, and to the latter as SoS evolvability. SoS dynamicity includes the adequate reconfiguration of SoS in specific situations, for example in the golden hour of an emergency management situation or after the occurrence of a fault. SoS evolvability includes necessary modifications that are required to keep a system services relevant in the face of the ever-evolving society (e.g., new legal requirements, business cases, etc.).
- Emergence. The interactions among the autonomous CSs can lead to unexpected emergent properties at the SoS level. For instance, does the composition of CSs that are individually safe result in a safe SoS? Understanding and handling of emergent properties at the SoS level is one of the most difficult endeavours in the domain of SoS.
The goal of AMADEOS is to establish a sound conceptual basis and a generic architectural framework that address the challenges of guaranteed responsiveness, evolvability, dynamicity and emergence in Systems-of-Systems (SoS). AMADEOS will provide basic SoS concepts, measurable SoS quality attributes, and propose a SoS design methodology, supported by extended mainstream UML-based tools, for the representation, modelling, development, evolution and validation of Systems-of-Systems.More precisely, the objectives of the project are as follows:
- Objective 1: Introducing explicit, global, synchronized time into SoS models & SoSE. With such an approach, each constituent system, interface and element of a SoS has access to a precise, synchronized global time base. This global time enables new solutions that are expected to result in guaranteed responsiveness, in reduced cognitive complexity, higher dependability and a simplification of the certification of safety-critical services. Where guaranteed responsiveness cannot be achieved, algorithms will be used to provide asynchronous and fault tolerant synchronisation.
- Objective 2: Capturing and formalizing SoS evolvability and dynamicity. Evolution of a SoS is necessary for the adaptation to environmental changes such as new business cases, legal requirements, compliance, changing safety regulations, evolving environmental protection rules, etc. On the other hand, unforeseen events can occur in a SoS, such as a disaster or a failure of a constituent system. In this case, widely applicable control protocols to tackle dependability, security and performability of SoS will be provided, allowing these systems to adapt to the dynamic environment. In fact, mitigating the effects of dynamicity and evolution will become key to the success of many SoSs.
- Objective 3: Managing emerging properties in SoS. Emerging properties and emerging behaviour in a SoS may be either desired (i.e. positive) or undesired (i.e. negative, dangerous, critical). Understanding the mechanisms of emergence will help in the composition of constituent systems, especially in predicting the effects of composition on dependability, safety, security, availability. In close relation to objective 1, we will investigate the role of time in the appearance of emerging properties and the required actions to predict or mitigate their effects on the SoS.