Inaccessibility in multi-agent systems

In standard situations, agents in a multi-agent community need to communicate each other to accomplish cooperation and coordination of joint activities. Normally it is expected that each agent can contact any other using a reliable communication infrastructure. Often one has to think about situations when a communication subsystem becomes disintegrated and some agents become isolated from the rest of the community. The paper provides basic analysis of such inaccessibility situations, and reviews possible technologies to remedy the resulting difficulties. The concepts of a so-called doubler agent and stand-in agent for maintaining agents accessibility is introduced. A brief discussion on suitability of the acquaintance models for maintaining accessibility on a symbolic level is also touched.