The intrinsic complexity of self-organising MASs (multi-agent systems) makes it difficult to predict global system evolutions at early stages of the design process. Simulating high-level models to analyse properties of a MAS design can anticipate detection of incorrect / wrong design choices, and allow tuning of system parameters.

In this paper, we take abnormal-behaviour detection as a case study, and devise an artifact-based MAS architecture inspired by principles of the human immune systems. We use stochastic π-calculus to specify and run quantitative large-scale simulations, which allow us to verify the basic applicability of our IDS (intrusion detection system) and possibly obtain a preliminary set of its main working parameters.