SEB-IOTA: a trust-driven and hierarchically verified data broadcasting framework for smart grids with HMM-anchored reputation dynamics
Qin-Man Li, Xi-Xiang Zhang, Wei-Ming Liao, Di-Gui Zhou, Zhe-Zhe Liang, Ning QinPurpose
To address secure, low-latency broadcasting of high-value data in smart grids, this paper aims to propose secure and efficient broadcast framework based on IOTA (SEB-IOTA). It tackles three key bottlenecks: single points of failure in centralized architectures, poor real-time performance of conventional blockchain consensus and lack of dynamic trust assessment. The goal is to enable efficient, privacy-preserving and attack-resilient data broadcast for grid stability.
Design/methodology/approach
SEB-IOTA integrates four modules: Hidden markov model (HMM)-based dynamic reputation assessment to detect malicious nodes; entity anonymization using SM2/SM4 and ring signatures; Stackelberg game-theoretic bandwidth allocation; and hierarchical IOTA verification with Deep Q Networks-based tip selection (adaptive DQN-based tip selection algorithm (AD-TSA)) and reputation-based proof of work (PoW) difficulty adjustment. Simulations use SUMO and OMNeT++ under three traffic arrival models.
Findings
SEB-IOTA reduces data confirmation latency by 82.1%, improves verification efficiency by 75.3%, and achieves over 99% defense success against false data injection attacks. It maintains low broadcast delay even under bandwidth scarcity and high-value data load (75%), outperforming delegated proof of stake (DPOS), vehicle-based secure blockchain consensus and Hybrid-PoS. Security analysis confirms robustness against replay, eavesdropping and man-in-the-middle attacks.
Originality/value
This paper presents SEB-IOTA, the first framework to integrate HMM-anchored reputation dynamics with IOTA’s directed acyclic graph-based ledger for smart grid broadcasting. Its originality lies in four innovations: HMM-based dynamic trust assessment with behavioral deviation correction, entity anonymization via SM2/SM4 and ring signatures, Stackelberg game-theoretic bandwidth allocation and AD-TSA deep reinforcement learning tip selection with reputation-based PoW difficulty adjustment. The framework uniquely achieves millisecond-level latency, privacy preservation, and over 99% false data injection attack defense, filling the gap for trust-aware, real-time high-value data broadcast in smart grids.