Modularity and Openness in Modeling Multi-Agent Systems

TL;DR

This paper enhances the modularity and openness of formal models for multi-agent systems by refining interference functions in MIS, enabling more flexible and precise modeling of complex multi-agent interactions.

Contribution

It introduces modifications to MIS structure that significantly improve modularity and openness, and formalizes multi-agency and openness concepts within MIS representations.

Findings

Improved MIS allows flexible modeling of multi-agent systems.

Demonstrated on trains, tunnel, controller, and cryptographers examples.

Formal definitions of multi-agency and openness in MIS.

Abstract

We revisit the formalism of modular interpreted systems (MIS) which encourages modular and open modeling of synchronous multi-agent systems. The original formulation of MIS did not live entirely up to its promise. In this paper, we propose how to improve modularity and openness of MIS by changing the structure of interference functions. These relatively small changes allow for surprisingly high flexibility when modeling actual multi-agent systems. We demonstrate this on two well-known examples, namely the trains, tunnel and controller, and the dining cryptographers. Perhaps more importantly, we propose how the notions of multi-agency and openness, crucial for multi-agent systems, can be precisely defined based on their MIS representations.