A Coalition-Based Communication Framework for Task-Driven Flying Ad-Hoc Networks (1812.00896v2)

Abstract: In this paper, we develop a task-driven networking framework for Flying Ad-hoc Networks (FANETs), where a coalition-based model is outlined. Firstly, we present a brief survey to show the state-of-the-art studies on the intra-communication of unmanned aerial vehicle (UAV) swarms. The features and deficiencies of existing models are analyzed. To capture the task-driven requirement of the flying multi-agent system, a coalition-based framework is proposed. We discuss the composition, networking mode and the classification of data transmission. After that, the application scenario of UAV coalitions is given, where large-scale, distributed and highly dynamic characteristics greatly increase the difficulty of resource optimization for UAVs. To tackle the problem, we design an intelligence-based optimization architecture, which mainly includes the game model, machine learning and real-time decision. Under the guidance of game theories and machine learning, UAVs can make comprehensive decisions by combining the previous training results with their sensing, information interaction, and game strategies. Finally, a preliminary case and promising open issues of UAV coalitions are studied.