Coalition formation among unmanned aerial vehicles for uncertain task allocation

Abstract

Unmanned Aerial Vehicles (UAVs), mobile stations and self-operation sensor nodes consist a potential next generation wireless network. The key problem is task allocation among agents to accomplish tasks. In this paper, the self-organizing UAVs networks model is considered which a number of data packets from sensor nodes are collected by several UAVs to send to the control center. The data collection problem is considered as the task allocation problem. Most literatures about task allocation problem consider the tasks are static which the information of the task is completely known. However, due to the dynamic environment of task, the status, and locations of tasks varying dynamically, the task is uncertain because the information of the task is incompletely known by UAVs. In this paper, the uncertain task allocation problem among UAVs is modeled as Bayesian coalition formation. Each formed coalition consists of tasks and UAVs who move among various tasks in the coalition using polling scheme. The Bayesian coalition game based on possible environments is proposed to model the interaction between uncertain tasks and UAVs in coalition formation. According to the achieved utility and cost in the features of the throughput and delay, members in the coalition make distributed decisions to leave or join in the coalition. A belief update mechanism of uncertain tasks is used to calculate the probability that the task is in the range of potential coalition in each environment. Simulation results demonstrate that the Nash stable coalition structure can be obtained depending on the proposed Bayesian coalition formation algorithm. The Nash stable coalition structures with incomplete information and complete information converge to the same coalitional structure.