An Emergency Alert Dissemination Scheme for V2X Communication using Solace PubSub+
- 주제(키워드) Quality of Service , V2X communication , Emergency Alert Dissemination , Solace PubSub+ , ML , DSRC , C-V2X , Intelligent Transportation Systems , Decentralized Architectures , Low-latency Networking , Vehicular Safety Applications
- 주제(DDC) 621.39
- 발행기관 아주대학교 일반대학원
- 지도교수 Young-Bae Ko
- 발행년도 2026
- 학위수여년월 2026. 2
- 학위명 석사
- 학과 및 전공 일반대학원 컴퓨터공학과
- 실제URI http://www.dcollection.net/handler/ajou/000000035595
- 본문언어 영어
- 저작권 아주대학교 논문은 저작권에 의해 보호받습니다.
초록/요약
Vehicular safety systems require ultra-reliable, sub-tactile latency and contextually dis- criminative dissemination of emergency alerts, especially within unpredictable, densely pop- ulated traffic environments. Although modern V2X technologies, notably DSRC and C- V2X, have made significant progress in physical-layer communication, current dissemination schemes are still hindered by static prioritization techniques, broadcast redundancy, and a fragile reliance on cloud-centric infrastructures. These limitations hinder timely hazard com- munication and weaken resilience during connectivity disruptions. This thesis introduces QoS-AED. This decentralized, AI-orchestrated emergency alert dissemination architecture utilizes real-time machine learning inference, adaptive TTL management, hierarchical topic encoding, and broker-embedded suppression in- telligence. The system operationalizes Solace PubSub+ as an edge-resident MQTTv5 broker, enabling deterministic TTL enforcement, semantically stratified topic routing, and cryptographic hash-based duplicate elimination entirely independent of remote cloud ser- vices. A finely tuned XGBoost classifier predicts alert severity using real-world contextual attributes, including weather variables, visibility gradients, and road conditions. Severity outputs are further synthesized with RSU proximity and vehicular dynamics into an APS, which governs TTL modulation and situationally adaptive dissemination reach. Leveraging 20,000 samples from the US Accidents March 2023 dataset, QoS-AED attains 80.1% severity classification accuracy, 99.53% suppression efficiency, and a 40.2 ms mean dissemination latency, decisively outperforming PrioMQTT and J2H across latency, throughput, and reliability metrics. Protocol-layer evaluation via a SUMO+NS-3 co- simulation environment further corroborates the scheme’s robustness, revealing deterministic responsiveness under both DSRC and C-V2X conditions. The results substantiate that AI-enabled semantic awareness, combined with edge-brokered dissemination intelligence, can be seamlessly embedded within ETSI-compliant V2X commu- nication stacks. Altogether, QoS-AED establishes a scalable, fault-tolerant, and Internet- independent emergency alerting paradigm, offering a reliable, future-proof foundation for next-generation intelligent transportation infrastructure. Keywords: Quality of Service, V2X communication, Emergency Alert Dissemination, Solace PubSub+, ML, DSRC, C-V2X, Intelligent Transportation Systems, Decentralized Ar- chitectures, Low-latency Networking, Vehicular Safety Applications.
more목차
I. Introduction 1
1.1 ITS Evolution and V2X Communication 2
1.2 Role of Edge Brokers, Machine Learning, and Event-Driven Architectures 3
1.3 Research Gap 4
1.4 Motivation 4
1.5 Problem Statement 5
1.6 Major Contributions 7
II. Background and Related Work 9
2.1 Broker-Based Prioritization and MQTT Extensions 9
2.2 Centralized, Cloud-Centric and SDN-Assisted Architectures 10
2.3 Machine Learning and Opportunistic Vehicular Routing Approaches 11
2.4 Gap Analysis and Rationale for the Proposed Approach 11
III. System Architecture 13
3.1 QoS-AED scheme Overview 13
3.1.1 OBU Layer: ML-Driven Decision Module and Interface Role 15
3.1.2 RSU Layer: Edge Dissemination and Broker-Side Enforcement 15
3.1.3 Cloud Layer: Aggregation and Feedback Interface 16
3.2 Communication and Simulation Environment 16
3.3 Data Pipeline and Severity-APS-TTL Engine 17
3.4 Solace Broker: Routing, TTL Enforcement, Suppression 21
3.5 CAM/DENM Compatibility Mapping 23
3.6 Topic Hierarchy and Wildcards 24
3.7 Design Rationale 26
3.8 Summary 26
IV. Experimental Evaluation 27
4.1 Experimental Environment and Dataset 27
4.2 Implementation and Simulation Setup 28
4.3 Results Analysis 28
4.3.1 Machine Learning Model Performance 28
4.3.2 OBU-RSU End-to-End Implementation 29
4.3.3 Suppression Effectiveness and Latency Reduction 30
4.3.4 Latency Evaluation under APS-Driven TTL Dissemination 32
4.3.5 DSRC and C-V2X Baseline Comparison 36
4.3.6 Comparative Benchmarking 36
4.4 Discussion 38
V. Conclusion and Future Work 40
5.1 Conclusion 40
5.2 Future Work and Integration Outlook 41
5.2.1 5G-V2X and Multi-Access Expansion 41
5.2.2 Edge Orchestrator Deployment 41
5.2.3 Hardware RSU/OBU Validation 41
References 42
