Dynamic Agent Allocation for Large-Scale Multi-Agent Applications
Myeong-Wuk Jang, Gul Agha · 2005
Although distributed computing is necessary to execute largescale multi-agent applications, the distribution of agents is challenging especially when the communication pattern among agents is continuously changing. This paper proposes two dynamic agent allocation mechanisms for large-scale multi-agent applications. The aim of one mechanism is to minimize agent communication cost, while that of the other mechanism is to prevent overloaded computer nodes from negatively affecting overall performance. In this paper, we synthesize these two mechanisms in a multi-agent framework called Adaptive Actor Architecture (AAA). In AAA, each agent platform monitors the workload of its computer node and the communication pattern of agents executing on it. An agent platform periodically reallocates agents according to their communication localities. When an agent platform is overloaded, the agent platform migrates a set of agents, which have more intra-group communication than inter-group or inter-node communication, to a relatively underloaded agent platform. These agent allocation mechanisms are developed as fully distributed algorithms, and they may move the selected agents as a group. In order to evaluate these mechanisms, preliminary experimental results with large-scale micro UAV (Unmanned Aerial Vehicle) simulations are described.