Distributed constraint optimization, resource allocation and scheduling in large scale agent networks

Abstract

A request may need to be propagated several times before a team of agents capable of handling it is gathered. After a team of agents is formed, the members of the team need to jointly schedule their activities taking into account interdependencies, as well as their other scheduled activities. This is in itself a hard problem which our framework has tackled using a combination of dynamic team reorganization and distributed constraint optimization methods. Every agent team that is assigned a request must find within a finite period of time if there is a proper way to serve it. In that case it will search for the optimal assignment of the appropriate resources to the request. Otherwise, the team must decisively reach to the conclusion that either the request can not be ultimately served or the specific agent team has inadequate It has to be noticed that in our effort we do not deal with communication decisions for optimizing information sharing/exchange as done in (Goldman and Zilberstein, 2003), or for proactively exchanging information (Zhang et al., 2004) This is orthogonal to our research which may further increase the efficiency of the proposed method. However, we point that this can not be done in any way such that agents share a global view of the environment state (