Small World Model for Agent Searching

Small world models have shown us that, under certain conditions, big networks can be searched, and the size of the network barely affects the length of the shortest paths between any pair of nodes. However, knowing that the distance between two nodes in the network is small does not mean that this shortest path can only be found using local information. For large networks, scalable systems or dynamic environments, a method that ensures that the shortest path can be located using only local information is needed. Hopefully, this method exists, as Kleinberg has proved for small world networks with a special structure. The solution proposed in this paper applies Kleinberg´s contribution, by creating a network of AMSs and DFs, forming a navigable small world network, but keeping the external view as a unitary service provided by one entity. Agents register themselves in the system, but they are reallocated into their corresponding node of the network without propagating this information to the rest of the nodes. When an agent asks for another agent or service, an active search is made. The structure of the network guarantees that the node that contains the information is located following a bounded path length and it is treatable due to the small world structure.