Service-Oriented Resource Allocation in UAV-Assist Internet of Things

Abstract

Internet of Things (IoT) is a promising technology for realizing massive interconnection, especially when it is enabled by Unmanned Aerial Vehicles (UAVs) in terms of flexibility and operability. However, limited cache resources and channel resources limit the implementation of high throughput and low-loss communication systems. Furthermore, in order to better user experience and system resource utilization, it is imperative for different services to acquire individualized resource configurations. This paper proposes a Service-Oriented Resource Allocation (SoRA) algorithm based on a scenario involving multiple users, a relay, and sink. Firstly, SoRA uses a fuzzy complementary matrix to evaluate different types of service. Secondly, the storage and communication resources are allocated jointly to maximize system utility. Finally, the deep Q-learning strategy is adopted to allocate the resources dynamically according to the network conditions. The simulation results show that the proposed SoRA algorithm reduces the delay by 43.49%, reduces the average packet loss by 14.28%, and decreases the average power consumption by 25.36%.

Keywords

• Unmanned aerial vehicle
• Internet of Things
• resource allocation
• deep reinforcement learning