Validating Emergent Behavior in Complex Systems

Abstract

Undesired or unexpected properties are frequent, as large-scale complex systems with nonlinear interactions are being designed and implemented to answer real-life scenarios. Identifying these behaviors as they happen as well as determining whether these behaviors are beneficial for the system is crucial to highlight potential faults or undesired side effects early in the development of a system, thus promising significant cost reductions. Beyond the inherent challenges in identifying these behaviors, the problem of validating the observed emergent behavior remains challenging, as this behavior is, by definition, not expected or envisaged by system designers. This chapter presents an overview of existing work for the automated detection of emergent behavior and discusses some potential solutions to the challenge of validating emergent behavior. Building on the idea of comparing an identified emergent behavior with previously seen behaviors, we propose a two-step process for validating emergent behavior. Our initial experiments using a Flock of Birds model show the promise of this approach but also highlight future avenues of research.