Joint Optimization of Task Assignment and Virtual-Function Deployment for Cluster-Collaborative UAV Networks

Unmanned Aerial Vehicle (UAV) networks are emerging as a promising solution for enhancing the 5th Generation (5G) communication infrastructures. UAVs with high mobility and flexible deployment, can provide ubiquitous connectivity and computation. From the implementation perspective, to flexibly manage the computational resources in UAVs, the emerging Network Function Virtualization (NFV) can be utilized. Additionally, cluster structure is an efficient networking architecture for task implementation in UAV. However, the division of clusters and deployment of Virtual Function (VF) both affect the performance of UAV networks. This paper proposes a comprehensive framework for the joint optimization of task assignment and VF deployment in cluster-collaborative UAV networks. For this framework, this paper proposes a mathematical model, allowing UAV cluster partitioning and VF deployment, aimed at maximize task revenue while minimizing task delay and energy consumption. A Deep Reinforcement Learning (DRL)-based joint task assignment and VF deployment (DJTSVD) algorithm is proposed to solve this NP-hard problem. In the DRL framework, the state space, action space, and reward function of the agent are defined. To evaluate the effectiveness of the proposed solution, a series of experiments with different parameter settings are conducted. The results demonstrate that the DRL-based algorithm effectively captures the complex relationship between task assignment and VF deployment in UAV networks, improving task revenue while reducing delay and energy consumption.