Dynamic Controller Assignment in Software Defined Internet of Vehicles Through Multi-Agent Deep Reinforcement Learning

Autor(es)

Yuan, Tingting ^[1] ; Neto, Wilson da Rocha ^[2] ; Rothenberg, Christian Esteve ^[2] ; Obraczka, Katia ^[3] ; Barakat, Chadi ^[1] ; Turletti, Thierry ^[1]

Afiliação do(s) autor(es)

^[1] Inria, Diana Project Team, F-06902 Sophia Antipolis - France

^[2] Univ Estadual Campinas, Fac Elect & Comp Engn, BR-13086902 Campinas - Brazil

^[3] Univ Calif Santa Cruz, Dept Comp Engn, Santa Cruz, CA 95064 - USA

• Tipo de documento: Artigo Científico
• Fonte: IEEE TRANSACTIONS ON NETWORK AND SERVICE MANAGEMENT; v. 18, n. 1, p. 585-596, MAR 2021.
• Citações Web of Science: 6

Resumo

In this article, we introduce a novel dynamic controller assignment algorithm targeting connected vehicle services and applications, also known as Internet of Vehicles (IoV). The proposed approach considers a hierarchically distributed control plane, decoupled from the data plane, and uses vehicle location and control traffic load to perform controller assignment dynamically. We model the dynamic controller assignment problem as a multi-agent Markov game and solve it with cooperative multi-agent deep reinforcement learning. Simulation results using real-world vehicle mobility traces show that the proposed approach outperforms existing ones by reducing control delay as well as packet loss. (AU)

• Processo FAPESP: 17/50361-0 - Distributed inteligent vehicular environment: enabling ITS through programmable networks
• Beneficiário: Christian Rodolfo Esteve Rothenberg
• Modalidade de apoio: Auxílio à Pesquisa - Regular