Securing Snapshot Pruning in IOTA Tangle 2.0: A Cooperative Deep Reinforcement Learning Approach for Edge-Cloud Smart Meter Networks

Decentralized security architectures powered by distributed ledger technology (DLT) have become prominent solutions for elevated security in Web 3.0. The IOTA Tangle is an open-source DLT, available for the Internet of Things (IoTs), as transactions can be carried out freely. IOTA 2.0 is the recent version of the IOTA Tangle, yet its local snapshot mechanism has certain drawbacks — security violations, increased costs and a lack of centralized coordination. Specifically, the situation-unaware decision for pruning increases the cost for energy service providers in the smart meter setting. Data analytics becomes infeasible due to the localized pruning policies. To address these issues, we develop a deep-Q actor-critique network-based cooperative pruning algorithm for IOTA 2.0 that preserves cryptographic properties for stored data in every edge server, namely, first and second preimage resistance, collision resistance and integrity. Our informal security analysis against active adversaries reveals the satisfaction of the security properties. Extensive simulation using the Monte Carlo method unveils that the proposed method is efficient regarding 30% improvement in storage reduction and 35% improvement in energy reduction, committing to the lower bound in solution. Our study quantifies the cost benefits of the proposed algorithms for five random cities with varying population densities for the years 2025 to 2029. We implement and test the algorithms in a real-world setting to benefit diverse stakeholders. Our solutions can be adopted by energy companies for cost-benefit and the IOTA Tangle foundation for future up-gradation of its snapshot mechanism.