Journal of Engineering Research and Reports

Urban areas in the Global South are at the forefront of the climate crisis, contributing over 70% of global CO₂ emissions while lacking access to intelligent, transparent, and equitable carbon governance systems. Existing carbon markets, plagued by opacity, centralization, and static MRV (Monitoring, Reporting, and Verification) practices, are inadequate for dynamically managing decentralized, sectoral emissions in rapidly evolving megacities. This research proposes a novel, AI-powered carbon market intelligence framework that integrates cutting-edge technologies: Long Short-Term Memory (LSTM) networks, Graph Neural Networks (GNNs), Deep Reinforcement Learning (DRL), blockchain-enabled smart contracts, federated learning (FL), digital twins, and explainable AI (SHAP, LIME). The system is modular, privacy-preserving, and designed for real-time urban-scale decarbonization, adaptive policymaking, and citizen-level participation. Using Dhaka, Bangladesh, a climate-vulnerable megacity, as the primary use case, and Nairobi as a secondary scalability testbed, this study simulates a comprehensive pipeline: IoT sensors stream data to digital twins; AI models forecast emissions and carbon prices; smart contracts trigger transparent offset issuance; and federated models ensure localized learning without compromising data sovereignty. The system achieves high predictive accuracy (R² > 0.92), 27.6% emission reductions in waste-energy sectors, and 12.3% gains in offset ROI over static baselines. Smart contract execution remains under 4.5 seconds, with negligible energy use under Proof-of-Stake blockchain. The explainability layer enhances stakeholder trust and policy interpretability, while gamified P2P carbon trading and participatory digital twins democratize climate action. The framework aligns with global instruments, including the UNFCCC Enhanced Transparency Framework, Article 6 mechanisms, Verra and Gold Standard protocols, and ICAO’s CORSIA, positioning it for integration into national and voluntary carbon markets. Ethical safeguards address algorithmic bias, data privacy, system resilience, and governance decentralization via DAOs. A full AI sustainability audit quantifies environmental trade-offs, demonstrating that avoided emissions exceed compute footprints by orders of magnitude. This paper delivers the first end-to-end, federated-AI and blockchain-driven carbon governance system for urban infrastructures in the Global South. It enables a paradigm shift toward real-time, transparent, and just carbon markets, offering a scalable blueprint for Net Zero-aligned smart cities worldwide. The proposed architecture not only advances scientific frontiers but also lays the groundwork for high-impact funding, policy integration, and global replication.