A Blockchain-Based Covert Communication Scheme Resilient to Internal Attacks

Blockchains offer promising platforms for covert communication schemes because of their decentralized, anonymous, and tamper-resistant nature. However, existing blockchainbased covert communication schemes focus on concealing communication from external attackers. In this paper, we propose a blockchain-based covert communication scheme designed to defend against internal attackers who have access to some covert transactions and aim to identify others. First, we design a generative-model-based transaction parameter generation method to synthesize parameters for covert transactions, preventing internal attackers from identifying covert transactions through their distinctive features. To effectively blend covert transactions into real anonymous transactions, we combine transaction clustering and identity tagging techniques to identify transactions with known identities and exclude them from the generative model's training set. Second, to prevent internal attackers from linking covert transactions through transaction labels, we utilize dynamic addresses to label covert transactions. We design a method that enables communicating parties to synchronously generate dynamic addresses without requiring suspicious off-chain interactions. Additionally, we design an on-chain key exchange mechanism to covertly update the shared key used for generating dynamic addresses. Finally, we leverage subliminal channels within transaction signatures to embed and extract communication messages. Our security analysis and experimental results demonstrate that, compared with existing schemes, our scheme achieves greater covertness against internal attackers while maintaining state-of-the-art communication efficiency.