EMT-MAPPO: Task-Aware Multi-Agent Scheduling of Cooperative Awareness Messages and Video Streams in NR-V2X

DOI: https://doi.org/10.62517/jbdc.202601402

Author(s)

Bin Wang*

Affiliation(s)

College of Communication Engineering, Chongqing Polytechnic University of Electronic Technology, Chongqing, China *Corresponding Author

Abstract

The joint resource allocation problem for cooperative awareness messages (CAMs) and video streaming in NR-V2X Mode 2 is investigated, where safety-critical and bandwidth-intensive services compete for limited radio resources under dynamic fast-fading channel conditions. To address this challenge, we propose EMT-MAPPO, an enhanced multi-task multi-agent proximal policy optimization framework that augments each agent’s local observation with task-aware CAM queue occupancy and video-frame transmission progress. This task-aware state augmentation enables the policy to better coordinate periodic safety message delivery with frame-based video transmission. We compare EMT-MAPPO with standard MAPPO, MAA2C, BP-SPS, and random scheduling and further evaluate convergence behaviour across 4, 8, and 16 agents. Experimental results show that EMT-MAPPO achieves the highest cumulative reward and the better overall trade-off between frame success rate (FSR) and CAM success rate (CSR).

Keywords

Cooperative Awareness Messages (CAMs); New Radio Vehicle-to-Everything (NR-V2X); Multi-Agent Reinforcement Learning (MARL); Enhanced Multi-Task MAPPO (EMT-MAPPO); Resource Allocation; Video Streaming

References

[1]Kandu SS, Prakash A, Tiwari M, Tripathi S, Gupta N, Pal R. C-V2X and 5G-V2X: Challenges and Improvement Strategies. In: Proceedings of the 2025 Third International Conference on Microwave, Antenna and Communication (MAC); 2025; Bhopal, India. Piscataway, NJ: IEEE; 2025. p. 1-6. [2]Shin C, Farag E, Ryu H, et al. Vehicle-to-everything (V2X) evolution from 4G to 5G in 3GPP: Focusing on resource allocation aspects. IEEE Access. 2023; 11: 18689-18703. [3]3GPP. Study on NR Vehicle-to-Everything (V2X). 3GPP TR 38.885, Release 16. Sophia Antipolis: 3GPP; 2019. [4]Wael CBA, Dogheche EH, Armi N, Subekti A, Dayoub I. Leveraging 3GPP Features and Optimization Techniques for 5G NR-V2X Resource Allocation: A Survey. IEEE Open J Intell Transp Syst. 2025; 6: 967-994. [5]Zhou X, Hui F, Liu J, Wang W, Zhang J. Resource Reservation in C-V2X Networks for Dynamic Traffic Environments: From Vehicle Density-Driven to Deep Reinforcement Learning. IEEE Trans Veh Technol. 2025; 74(11):17429-17444. [6]Dayal A, Shah VK, Choudhury B, Marojevic V, Dietrich C, Reed JH. Adaptive Semi-Persistent Scheduling for Enhanced On-road Safety in Decentralized V2X Networks. In: Proceedings of the 2021 IFIP Networking Conference; 2021; Espoo and Helsinki, Finland. Piscataway, NJ: IEEE; 2021. p. 1-9. [7]Hyeon D, Lee C, Kim H, Cho S, Paek J, Govindan R. Probabilistic resource rescheduling for C-V2X based on delivery rate estimation. J Commun Netw. 2024; 26(2):239-251. [8]Jung S, Kim JH, Kim J. Intelligent extra resource allocation for cooperative awareness message broadcasting in cellular-V2X networks. IEEE Trans Netw Serv Manag. 2025; 22 (2):1677-1689. [9]Fu Q, Liu J, Wang J, et al. Toward throughput optimization: A congestion control scheme for CAM delivery via NR sidelink. IEEE Internet Things J. 2025; 12 (9):12206-12218. [10]S. Heo and H. Kim.Two-Dimensional Compound Configured Grants for Video-Aware NR Sidelink Mode 1 Scheduling. IEEE Trans Veh Technol. 2026. [11]Wolff VA, Xhoxhi E, Schiegg FA, et al. Uncertainty and prioritization: Empirical evaluation of a VoI-based CPM generation pipeline using real-world data. In: Proceedings of the 2025 IEEE Vehicular Networking Conference (VNC); 2025; Piscataway, NJ: IEEE; 2025. p. 1-8. [12]Yordanov N, Çavuşoğlu B. Transient, deadline-aware RAN slicing for NR-V2X using finite-capacity time-varying slice queues. Veh Commun. 2026:101064. [13]Huan Z, Sun J, Chen Z, et al. Resource Allocation in V2X Networks: A Double Deep Q-Network Approach with Graph Neural Networks. Comput Mater Contin. 2025; 84(3):5427-5443. [14]Song S, Zhang Z, Wu Q, Fan P, Fan Q. Joint optimization of age of information and energy consumption in NR-V2X system based on deep reinforcement learning. Sensors. 2024; 24(13):4338. [15]Wang S, Lei L, Maheshwari P, Bellefeuille S, Zheng K, Niyato D. Multi-Agent DRL for V2X Resource Allocation: Disentangling Challenges and Benchmarking Solutions. arXiv preprint arXiv:2603.06607. 2026. [16]Wang C, Guan Y, Peng S, et al. Data prioritization aware resource allocation in internet of vehicles using multi-agent deep reinforcement learning. Neural Netw. 2025; 190: 107671. [17]Gao A, Zhu Z, Zhang J, et al. Matching combined heterogeneous multi-agent reinforcement learning for resource allocation in NOMA-V2X networks. IEEE Trans Veh Technol. 2024; 73(10):15109-15124. [18]Hegde A, Song R, Festag A. Radio resource allocation in 5G-NR V2X: A multi-agent actor-critic based approach. IEEE Access. 2023; 11: 87225-87244. [19]M. M. Saad, M. A. Tariq, M. Ajmal, D. Kim and G. Srivastava. Federated Multiagent Reinforcement Learning for Resource Allocation in NR-V2X Mode 2. IEEE Internet Things J. 2025; 12(13):23402-23417. [20]Morshed M, Zaman M. Curriculum-assisted multi-agent reinforcement learning for scalable V2X resource allocation. Phys Commun. 2026; 76. [21]I KP, Venkatesh TG. Integrated Sensing, Communication, and Computing for NR-V2X: A Cross-Layer Resource Allocation Framework Using Multi-Agent Reinforcement Learning. arXiv preprint arXiv:2605.24972. 2026. [22]Fair Resource Allocation in C-V2X Networks: A Reinforcement Learning Approach for Perception Message Dissemination. In: Proceedings of the IEEE International Conference on Cloud and Big Data Computing (CBDCom); 2025 Oct; Hakodate, Japan. Piscataway, NJ: IEEE; 2025. [23]Basset L, Bouriga A, Dhaou R. Learning to Talk: Emergent CAM/CPM Scheduling via MAPPO in Constrained V2X Networks. In: Proceedings of the 2025 International Conference on Modeling, Analysis and Simulation of Wireless and Mobile Systems (MSWiM); 2025; Piscataway, NJ: IEEE; 2025. p. 767-770. [24]Khan I, Haladappa MS. Risk-Aware Multi-Agent Advantage Actor-Critic Based Resource Allocation for C-V2X Communication in Cellular Networks. Proc Eng Technol Innov. 2025; 29:47-60. [25]Gu B, Chen W, et al. Multiagent reinforcement learning-based semi-persistent scheduling scheme in C-V2X mode 4. IEEE Trans Veh Technol. 2022; 71(11):12044-12056.