AA. Libros Electrónicos · 2019

Towards a Distributed Systems Model based on Multi-Agent Systems for Reproducing Self-properties

In this chapter, we introduce a preliminary model of a distributed system (some algorithms and principles are modelled) which has been inspired by multi-agent systems. Agents can communicate in a local fashion and establish a cooperation process in order to compute some desired functions as a traditional distributed system, producing an output that is transparent to an end user. Each agent can sense new information, and acts following certain information in a decentralized way. As a result, the output of the implemented functions achieves the same result as a traditional distributed system by using local behaviours. Finally, we show that it is possible to define crashes in the proposed model in a local, natural and easy way. The purpose of doing this is to use this simulated model to introduce behaviours to agents that allow the system to recover from failures by itself, in turn, achieving self-* properties.