Research On The Key Technologies Of Multi-agent Computing For Dynamic Topology Networks

Nowadays, multi-agent computing plays a critical role more and more in the research of distributed artificial intelligence. In multi-agent systems, multi-agents can cooperate to perform the assigned task. In the research of multi-agent computing, there are four technologies that should be solved well: agent cooperation and coordination, agent communication, agent migration, agent trust and security.In now multi-agent systems, the underlying networks are always dynamic and the network topologies are always changed in the operation, which can be called network topology dyanmicity. Dynamic network topology modifications are essential for many reasons, e.g. to maintain a connection due to node mobility. Dynamic topology networks are characterized by the dynamicity of the links of different network components. The links could be physical connections or logical links of the different components. These networks are organized together to accomplish certain tasks.To make multi-agent computing adapt for the dynamic topology networks, this dissertation explores the key technologies of multi-agents in dynamic topology network, such as cooperation & coordination, communication, migration and fault tolerance, trust, etc. This dissertation mainly includes the following five parts.1. The agent cooperation and negotiation model for the variation of underlying network topologyAiming at the dynamics of underlying network topology variation and the agent distribution, a novel adaptive multi-agents coordination model is explored in this part. In this part, a series of algorithms for multi-agent task and resource negotiation (based on game theory and matrix) are provided. The provided algorithms consider the factors of network topology and agent distribution, and can implement effective task allocation and resource negotiation for current network topology. Therefore, the adaptation of agent coordination for dynamic underlying network topology can be achieved, which is also proved by the case studies and performance analyses in the dissertation.2. The construction of Agent communication architecture based on graph theoryBlackboard architecture is one of the popular agent communication mechanisms, where agents don't interact with each other directly but through blackboard. The blackboard architecture includes central fashion and distributed one. In blackboard communication architecture, the location of central blackboard (or distributed sub-blackboards) and communication topology among sub-blackboards are two important issues that can influence the agent communication performance very much. However, there are few works about such issues; and in the existing agent systems, the central blackboard (or distributed sub-blackboards) is (or are) usually randomly located in the underlying network. To solve such problem, this paper presents a model for constructing agent blackboard communication architecture based on graph theory. The model computes the location of central blackboard or sub-blackboards based on median location method,...