Multi-Agent Deep Reinforcement Learning for Cooperative Connected\n Vehicles

2020/01/07 by Dohyun Kwon, Kwon, Dohyun, Joongheon Kim +1 · 1 citation
Engineering · #Advanced MIMO Systems Optimization #FOS: Computer and information sciences #FOS: Electrical engineering #Machine Learning (cs.LG) #Millimeter-Wave Propagation and Modeling #Signal Processing (eess.SP) #Wireless Body Area Networks #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.2001.02337

openalex publication_date 2020/01/07 · openalex created_date 2020/07/16 · openalex updated_date 2026/07/28

Abstract

Millimeter-wave (mmWave) base station can offer abundant high capacity\nchannel resources toward connected vehicles so that quality-of-service (QoS) of\nthem in terms of downlink throughput can be highly improved. The mmWave base\nstation can operate among existing base stations (e.g., macro-cell base\nstation) on non-overlapped channels among them and the vehicles can make\ndecision what base station to associate, and what channel to utilize on\nheterogeneous networks. Furthermore, because of the non-omni property of mmWave\ncommunication, the vehicles decide how to align the beam direction toward\nmmWave base station to associate with it. However, such joint problem requires\nhigh computational cost, which is NP-hard and has combinatorial features. In\nthis paper, we solve the problem in 3-tier heterogeneous vehicular network\n(HetVNet) with multi-agent deep reinforcement learning (DRL) in a way that\nmaximizes expected total reward (i.e., downlink throughput) of vehicles. The\nmulti-agent deep deterministic policy gradient (MADDPG) approach is introduced\nto achieve optimal policy in continuous action domain.\n

Cited by

• Hierarchical Federated Graph Attention Networks for Scalable and Resilient UAV Collision Avoidance

Related

• Optimal Beam Association for High Mobility mmWave Vehicular Networks: Lightweight Parallel Reinforcement Learning Approach
• Joint Relay Selection and Beam Management Based on Deep Reinforcement Learning for Millimeter Wave Vehicular Communication
• Deep-Reinforcement-Learning-Based Distributed Vehicle Position Controls for Coverage Expansion in mmWave V2X
• Joint Power Allocation and Beamformer for mmW-NOMA Downlink Systems by Deep Reinforcement Learning
• Multi-agent deep reinforcement learning (MADRL) meets multi-user MIMO systems