AMTA-Sec: An Adaptive Multi-Layer Trust Architecture For Attack-Resilient Distributed Blue Carbon MRV Systems

Abstract

Blue carbon ecosystems such as mangroves, tidal marshes, and seagrass mead-ows are among the most effective natural carbon sinks and play a significant role in climate change mitigation. Distributed IoT telemetry, edge computing, machine learning, and blockchain-enabled carbon registries are now increasingly being used in modern Monitoring, Reporting, and Verification (MRV) systems to facilitate automated environmental monitoring and carbon credit verifica-tion. In contrast to centralized MRV systems, large scale distributed MRVs systems pose significant cybersecurity threats, such as telemetry spoofing, replay attacks, malicious edge compromises, validator collusion, trust poisoning, and adversarial manipulation of machine learning based soil organic carbon (SOC) estimation models. This paper introduces Adaptive Multi-Layer Trust Architec-ture (AMTA-Sec) to solve these problems, which aims to ensure the reliability and security of distributed blue carbon MRV systems against attacks. This framework highlights dynamic trust computation, edge-level anomaly detection, blockchain-assisted integrity verification, secure machine learning model protec-tion, and distributed trust orchestration across the sensing, edge, cloud and ledger layers. The experimental results showed that the proposed architecture is scalable and cyber-resilient by achieving an accuracy of 93.6 % for malicious telemetry detection, 91.2% for model poisoning detection, 100% for edge integrity violation detection and a reduction of 37% in blockchain write overhead across various scenarios of cyberattacks.