Trust-Aware Environmental State Consensus for Smart Agriculture with TEE-Enabled Sensing and Byzantine-Resilient...

This paper proposes a trust-aware environmental state consensus framework for smart agriculture that integrates TEE-enabled sensing, Byzantine-resilient aggregation, and lightweight blockchain-based state coordination under resource-constrained IoT environments. Unlike conventional IoT systems that treat blockchain as a transactional ledger for directly storing sensor outputs, the proposed framework utilizes blockchain as a state commitment layer that records only validated environmental state transitions. In the proposed architecture, distributed sensor readings are modeled as noisy and potentially adversarial observations of an underlying physical state rather than directly trusted measurements. To establish a reliable trust boundary between physical sensing and distributed coordination, TEE-enabled sensing devices provide authenticated and integrity-protected data outputs before blockchain processing. The TEE component is adopted as a deployed trusted execution anchor rather than a newly designed hardware security mechanism, and its role is to protect sensing-side execution and provide trustworthy inputs for subsequent coordination. Since trusted execution alone cannot guarantee the correctness of sensor observations, a Byzantine-resilient aggregation mechanism is introduced to estimate consistent environmental states under faulty or adversarial sensing conditions. The validated states are then committed through a lightweight permissioned blockchain to provide tamper-evident state finality using a K-confirmation-based commitment mechanism. The proposed framework is implemented and evaluated on a real greenhouse IoT platform with distributed sensing nodes and edge computing devices. Experimental results demonstrate that the proposed approach improves environmental state consistency under varying adversarial conditions while maintaining stable blockchain coordination and resource-aware execution performance.