Decentralized Reconfiguration for Self-Organizing Resource-Flow Systems Based on Local Knowledge

The introduction of self-organization into a system promises, among other things, to reduce the system's complexity and to increase the system's robustness against failures and its adaptability to changes in its environment. An example for systems that profit from self-organization are resource-flow systems, e.g., production lines. Such systems are characterized by a number of independent agents that process resources by applying capabilities according to a given task. This paper introduces a decentralized reconfiguration mechanism that restructures a part of a resource-flow system in case of a failure. In order to do so, agents coordinate based on local knowledge and combine themselves into groups which are called coalitions. Each coalition then tries to restore the system's functionality, returning a previously consistent and correct system to a new consistent state, thus re-enabling correct processing of resources. As only local coalitions are formed, the parts of the system not involved in the reconfiguration process stay functional, meaning that the overall system does not come to a standstill.