Toward Trustworthy Coordination in Intelligent Edge Agent Networking: A Trust-Driven Transaction Propagation Mechanism

Abstract

Intelligent edge networks increasingly use blockchain to support trustworthy collaboration among distributed autonomous edge agents. However, dynamic and heterogeneous edge environments expose intelligent edge agents to spam transaction attacks, where adversaries exploit repeated per-hop verification to exhaust limited computational resources and disrupt both AI workloads and blockchain operations. To address this problem, we propose a two-stage trust-driven transaction propagation mechanism combining BLS multi-signatures with reputation evaluation. In the first stage, transactions propagate within a small scope to accumulate accountable endorsements and filter suspicious transactions early. After sufficient endorsements are collected, the second stage enables rapid large-scope propagation through endorsement authentication, reputation screening, and interaction-aware probabilistic verification, reducing redundant verification while preserving continuous inspection. This design can also be extended to blockchain-enabled cross-domain coordination in various distributed computing environments, such as JointCloud computing and computing power networks. Extensive simulations under spam injection, collusion attacks, and on-off attacks demonstrate that the mechanism maintains robust performance under adversarial attacks, limits invalid transaction propagation, and reduces verification computational resource consumption by at least 56% under spam injection compared with the original blockchain mechanism.