PCBS: a privacy-preserving carbon trading framework for AI-empowered IoT with blockchain and secure multi-party computation

Abstract

Carbon trading in AI-empowered IoT systems requires strict protection of sensitive information during demand declaration and transaction matching. Blockchain frameworks improve transparency and traceability but still expose participants to privacy risks across the entire workflow. This paper proposes PCBS, a privacy-preserving carbon trading framework tailored for AIoT environments where trading demands are authentically generated by predictive models. PCBS integrates blockchain with Pedersen commitments for confidential storage and a secure multi-party sorting protocol for encrypted demand aggregation. The design conceals sensitive transaction data, supports anonymous interaction among IoT nodes, and retains regulatory supervision without disclosing plaintext information. Theoretical analysis verifies that PCBS guarantees anonymity, confidentiality, and regulability throughout the trading process and ensures cryptographic accountability against malicious deviations. A prototype built on Hyperledger Fabric and MP-SPDZ demonstrates its practicality on AIoT platforms. Experimental evaluation shows that PCBS reduces runtime and communication overhead while maintaining stable throughput and low latency under varying user scales. These findings confirm that PCBS achieves a balance between privacy protection and system efficiency in IoT-driven carbon emissions trading.