Understanding evolution in systems of systems

Fendali, Mohamed Hichem; Meslati, Djamel; Borne, Isabelle

doi:10.1109/syseng.2017.8088283

Cited by 4 publications

(2 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Evolution is often considered as a major challenge in complex SoS, given the heterogeneity of constituent systems, hyper-connectivity of systems involved, the emergent behavior, and the evolutionary development processes (Fendali et. al., 2017).…”

Section: Sos Moe Evolutionmentioning

confidence: 99%

Framework for Complex SoS Emergent Behavior Evolution Using Deep Reinforcement Learning

Raman

Murugesan

2022

INCOSE International Symp

View full text Add to dashboard Cite

Advances in technology have made it easy to integrate multiple modern systems to form complex system‐of‐systems (SoS) to achieve unparalleled levels of functionality that are otherwise not achievable by the constituent systems in isolation. In fact, with the recent explosion of machine learning techniques to build autonomous systems such as drones and self‐driving cars, there is a pressing need to ensure that they collaboratively and safely operate in an SoS context. However, in general, the characteristic emergent behaviors of complex SoS – that directly impact its operational measures of success or Measures of Effectiveness (MOEs) – is very difficult, if not impossible, to manually explore, anticipate, and arbitrate just from knowledge of its underlying systems. Further, there are multiple scenarios of evolution in such complex SoS, including evolution in the emergent behavior of the SoS. The continuous, continual, and evolving nature of the SoS and constituent system environment's state and possible actions, adds further complexity. In this paper, we present a novel approach that leverages Reinforcement Learning, a machine learning approach, to inculcate adaptable intelligence in constituent systems to adapt their behaviors in tandem with the evolution of emergent behavior at the SoS level. By augmenting the reward mechanism of RL by leveraging SoS‐Constituent System MOE Relationship, that relates and ranks System MOEs vs. SoS MOEs, we inculcate an Intelligent‐Behavior Evolution Agent, with the necessary constraints to learn to maximize the SoS and system‐level MOEs, while adapting itself to the evolution in SoS. We illustrate our approach and demonstrate its feasibility and potential by applying it to a power grid SoS case example. The effectiveness and performance of the approach are quantified.

show abstract

Section: Sos Moe Evolutionmentioning

confidence: 99%

Framework for Complex SoS Emergent Behavior Evolution Using Deep Reinforcement Learning

Raman

Murugesan

2022

INCOSE International Symp

View full text Add to dashboard Cite

show abstract

“…Hence, evolution is often considered as a major challenge in system‐of‐systems, given the heterogeneity of constituent systems, hyper‐connectivity of systems involved, the emergent behavior, and the evolutionary development processes . Architecture evolution deals with changes to the static SoS architecture—for instance, changes to how the constituent systems are networked to each other.…”

Section: Decision Learning Framework Extended For Complex Sosmentioning

confidence: 99%

Decision learning framework for architecture design decisions of complex systems and system‐of‐systems

Raman

D’Souza

2019

Systems Engineering

View full text Add to dashboard Cite

Architecting complex systems and complex system‐of‐systems (SoS) have evinced keen interest recently. Architectural design decisions have a significant bearing on the operational measures of success, referred to as Measures of Effectiveness (MOEs), of the systems and SoS. Architecting complex systems and SoS involves making architecture design decisions despite uncertainty (due to knowledge gaps) on the implications associated with the decisions. The learning of whether the decision is optimal or not, and the impact on the MOEs and the emergent behavior of the SoS, often occur later, resulting in Learning Cycles. This paper proposes an integrated decision learning framework for architecture design decisions for complex systems and SoS. The proposed framework adopts a decision oriented view that factors the uncertainty associated with architectural decisions and the learning cycles and feedback loops experienced. The framework enables leverage of machine learning approaches to learn from the decision learning cycles experienced and factor it into the uncertainty assessments of the decisions. By inculcating various aspects such as knowledge gaps and learning cycles, by building models such as Learning Cycle Model and Uncertainty Model, and by incorporating deployment approaches such as codification of decision attributes and decision uncertainty assessments, the proposed framework enables progressive maturity of the architectural knowledge base and aids robustness in architecture design decisions.

show abstract