Network Layer-Oriented Task Allocation for Multiagent Systems in Undependable Multiplex Networks

In a multiplex network, agents are connected by multiple types of links, and the network can be split into more than one network layer which is composed of the same type of links and involved agents. Traditional task allocation methods of multiagent systems only consider the situations of agents themselves, but neglect the effects of network layers in multiplex networks. To solve such a problem, this paper takes network layers into account and presents a novel network layer-oriented task allocation model for multiplex agent networks, with such a model, first the network layers that can satisfy the objectives of task allocation will be allocated, then the final agents will be selected from the allocated network layers. Moreover, this paper deals with the situation of undependable networks, where the resource access of tasks may be undependable, and implements a task allocation based on negotiation reputation. It shows that the network layer-oriented task allocation model leads to an improvement in the success rate and execution time of tasks in multiplex networks when compared to traditional agent-oriented task allocation methods, moreover, such a model has good scalability for the size of tasks and robustness for dynamic undependability.