Home // ICONS 2021, The Sixteenth International Conference on Systems // View article

Reinforcement Learning for Emergent Behavior Evolution in Complex System-of-Systems

Authors:
Anitha Murugesan
Ramakrishnan Raman

Keywords: Systems of Systems; Emergent Behavior; Measures of Effectiveness; Reinforcement Learning; Complexity.

Abstract:
The ease of inter-connectivity among modern systems is permeating numerous System-Of-Systems (SoS), wherein multiple, independent systems interact and collaborate to achieve unparalleled levels of functionality that are otherwise unachievable by the constituent systems in isolation. This has resulted in exponential increase in complexity associated with modern systems and SoS. Complex SoS are characterized by emergent behavior which is very difficult, if not impossible, to anticipate just from knowledge of constituent systems. The emergent behavior manifests at the boundary of the SoS and impacts the Measures of Effectiveness (MOEs) of the SoS. In the context of SoS, each constituent system has its own MOEs, while the SoS has its own MOEs. Constituent systems collaborate and interact with each other, towards achieving the desired functionality and behavior at SoS level. Recently, there is an explosion in the adoption of Machine Learning techniques and models in various systems, and these techniques are increasingly being used to control many physical systems, such as cars and drones. Reinforcement Learning is a type of machine learning approach that allows agents to optimally learn strategies through interactions with its environment. This paper presents a novel approach towards using reinforcement learning models and techniques for evolving MOEs of the constituent systems and SoS towards addressing emergent behavior. The proposed approach, through SoS-Constituent System MOE Relationship, enables constituent systems to learn and adapt their behaviors in tandem with the evolution of emergent behavior at SoS level.

Pages: 5 to 10

Copyright: Copyright (c) IARIA, 2021

Publication date: April 18, 2021

Published in: conference

ISSN: 2308-4243

ISBN: 978-1-61208-838-9

Location: Porto, Portugal

Dates: from April 18, 2021 to April 22, 2021