Informatica

(1) A five-layer decoupled smart logistics data trading platform architecture is constructed. Different from existing single-layer blockchain trading frameworks, this work hierarchically divides logistics data workflows into perception, generation, transmission, trading platform and application layers, and introduces a blockchain oracle module to realize consistent synchronization between on-chain transaction records and off-chain IPFS raw cipher storage, solving the cross-layer data inconsistency defect of prior monolithic architectures.

(2) A quantitative model parameter quality evaluation algorithm based on cosine similarity, together with a dual-weight dynamic incentive mechanism, is designed. Most existing federated data trading studies lack standardized quantitative quality metrics and static reward allocation rules; our method extracts gradient feature vectors from local FL models, calculates similarity against benchmark parameters to grade data quality, and distributes transaction revenue weighted by both model quality and service participation time to motivate sustained user contribution.

(3) A complete lattice-based cryptographic suite including homomorphic proxy re-encryption and aggregate lattice signature is deployed for secure data circulation. Traditional pairing-based privacy schemes are vulnerable to quantum cracking, while standalone lattice encryption works cannot support dynamic cross-buyer ciphertext conversion; our lattice primitive supports homomorphic aggregation and one-time re-encryption transfer, realizing anti-quantum, flexible authorization of federated model trading ciphertexts. Theoretical security proofs and multi-dimensional simulation experiments verify that the proposed scheme achieves post-quantum privacy protection, accurate data quality quantification and high-efficiency dynamic logistics data transactions.