Simulation-Based Complex Adaptive Systems

Abstract

Complex Adaptive Systems (CAS) are systems that display two primary characteristics: emergent behavior and adaptive behavior. Emergent behavior manifests in a system comprising of large number of components, often considered as agents, engaged in multi-level interactions. Adaptive behavior manifests at the agent-environment boundary when the agent situates itself in an environment. Modeling a CAS has been a challenge due to limitations in bringing these two aspects together in a single formal specification in a computational environment. Lack of simulation environment for a CAS model adds further problem in validation and verification of a CAS model. Computationally, the emergent behavior can be understood using today’s latest technology of Feature Engineering, Deep Learning and data analytics using Big Data. This would facilitate the identification of various holistic behaviors and their classification that would aid designing various observers for detecting the emergent behavior in a computational environment. This aspect is largely bottom-up. Once various observers are computationally available, they can be integrated in the agent behavior repertoire so that the emergent behavior, that is now detectable and perceivable at the agent-environment boundary, can be used and acted-upon by the agent. This situates the agent in the environment and manifests as adaptable behavior. This activity is top-down as there is a conscious design process (done by a human) that is employed for such behavior refinement. This chapter will discuss the state-of-the-art in computational and simulation support needed and provides foundation to manifest accurate emergent behavior in a computational environment as a means to perform CAS engineering.