Deep Learning Aided Routing for Space-Air-Ground Integrated Networks Relying on Real Satellite, Flight, and Shipping Data

Article Computer Science, Information Systems

Deep-Learning-Aided Packet Routing in Aeronautical Ad Hoc Networks Relying on Real Flight Data: From Single-Objective to Near-Pareto Multiobjective Optimization

Dong Liu et al.

Summary: This article proposes using deep learning to assist data packet routing in aeronautical ad hoc networks (AANETs), with the aim of minimizing delay and maximizing path capacity and lifetime. A deep neural network is trained using historical flight data to map local geographic information to optimal next hop. Simulation results show that this DL-aided routing outperforms existing position-based routing protocols and approaches the optimal results obtained using global link information.

IEEE INTERNET OF THINGS JOURNAL (2022)

添加到收藏夹

Article Computer Science, Information Systems

Hybrid Satellite-Terrestrial Communication Networks for the Maritime Internet of Things: Key Technologies, Opportunities, and Challenges

Te Wei et al.

Summary: With the increasing marine activities, there is a growing demand for high-speed and reliable maritime communications. Existing communication technologies may lead to performance loss when used on the ocean, requiring customization for the maritime environment. Future maritime communications are envisioned to be environment-aware, service-driven, and integrated satellite-air-ground networks.

IEEE INTERNET OF THINGS JOURNAL (2021)

添加到收藏夹

Article Engineering, Electrical & Electronic

Deep Reinforcement Learning Aided Packet-Routing for Aeronautical Ad-Hoc Networks Formed by Passenger Planes

Dong Liu et al.

Summary: This paper applies deep reinforcement learning to routing in aeronautical ad-hoc networks (AANETs) with the goal of minimizing end-to-end delay. By utilizing a deep Q-network (DQN) and a deep value network (DVN), the relationship between optimal routing decisions and system dynamics is captured. Simulation results demonstrate that both DQN-routing and DVN-routing outperform the benchmark protocol in terms of lower end-to-end delay, with DVN-routing performing similarly to optimal routing based on perfect global information.

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY (2021)

添加到收藏夹

Article Computer Science, Hardware & Architecture