State Space Exploration of Spatially Organized Populations of Agents

In this paper, we aim at modeling and analyzing the behavior of a spatial population of agents through an exploration of their state space. Agents are localized on a dynamic graph and they have internal states. They interact with an environment. The evolution of the agents and of the environment is specified by a set of rules. The framework is carefully designed to enable the construction of a global state space that can be automatically build and analyzed. The formalism, called IRNs for integrated regulatory networks, may be seen as an extension of logical regulatory networks (à la Thomas) developed in systems biology with spatial information and generalized to use arbitrary data values and update functions of this values. This thus allows to model systems with multiple agents that may be located on a varying spatial structure, may store and update local information, may depend on varying global information and may communicate in their neighborhood. A model of such a system can be defined as an IRN, and then analyzed using model-checking to asses its properties. This paper sketches the modeling framework and its semantics. We show how IRN may be used for the modeling of a population of simple agents, the automatic analysis of various reach ability properties and the use of symmetries to reduce the size of the state space.