Chen, Shengxin, Agarwal, Vaibhav ORCID: https://orcid.org/0000-0001-5362-2936, Lele, Atharva and Tewari, Hitesh (2025) CarbonChainRL - A Blockchain Oracle-Based Reinforcement Learning System for Dynamic Carbon Trading Markets. In: 2025 IEEE International Conference on Distributed Ledger Technologies (ICDLT). IEEE, pp. 1-8

Full text not available from this repository.

Abstract

As global initiatives for carbon neutrality accelerate, existing Emissions Trading Systems (ETS) are increasingly challenged by volatile pricing, limited transparency, and slow policy responsiveness. This paper introduces CarbonChainRL, a novel blockchain-based framework augmented with a Deep Reinforcement Learning (DRL) agent to enable adaptive and decentralized market regulation. Leveraging a Deep Q-Network (DQN) integrated with Ethereum smart contracts, CarbonChainRL facilitates dynamic supply-demand adjustment policies that indirectly influence carbon pricing. The model operates within an 11dimensional state space, capturing real-time market indicators such as supply-demand imbalances, tax signals, and Market Stability Reserve (MSR) balances. A carefully designed reward function incentivizes price stability, supply-demand equilibrium, and prompt regulatory intervention. The framework employs a hierarchical oracle mechanism that fuses RL-driven predictions with historical market data, utilizing confidence scoring to inform on-chain policy triggers. Simulation results demonstrate rapid convergence of the model and highlight the effectiveness of the MSR module-powered by dynamic TNAC-based calculations-in stabilizing market conditions. Additionally, the system supports multi-sector registration, quota allocation, and real-time auctioning. Preliminary findings indicate that CarbonChainRL enhances market responsiveness and efficiency, outperforming conventional ETS architectures.

Item Type:	Book Section
Status:	Published
DOI:	10.1109/icdlt66400.2025.11466602
Subjects:	T Technology > T Technology (General)
School/Department:	London Campus
URI:	https://ray.yorksj.ac.uk/id/eprint/15768

University Staff: Request a correction | RaY Editors: Update this record