On negotiation protocol/strategy exerting capabilities in large-scale multi-agent systems.

点击翻译按钮获取中文摘要

We proposed the negotiation method, especially the effective task assignment method, in large-scale autonomous multi-agent systems based on contract net protocol and market-based allocation. In task assignment by the market- and bid-based approaches, agents will concentrate on a few specific "good" tasks for their selections. Thus, if the systems are large, the system’s efficiency often falls as a result. However, this performance degrade highly depends to the degree of the system’s workload. We introduced the concept of phantom tasks that are used to estimate the local workloads, and proposed the method in which agents decide their allocation strategy depending on the estimated workload.