Coordination of concurrent one-to-many negotiations in multi-agent systems
Khalid Mansour · Swinburne figshare (Swinburne University of Technology) · 2014
Automated negotiation is a key interaction mechanism in distributed and multi-agent systems. By adapting negotiation as an interaction method, autonomous agents can interact, compete and cooperate with one another effectively. When the scale of negotiation expands to involve more than two agents, the negotiation process becomes more complex. This thesis addresses the problem of managing concurrent and dependent instances of automated negotiations in the one-to-many multi-agent systems where agents are self-interested, unwilling to disclose their private information (e.g., utility structure, deadline, etc.) and the available information about the opponents is mainly the current history of their offers. In particular, this thesis considers the one-tomany negotiation where a buyer agent negotiates with multiple seller agents over one or more objects concurrently. Dependent instances of negotiation occur when different negotiations are related and the situation of one instance affects the situation of one or more other instances.