JR/T 0196-2020 in English

Interpretation of the core content of the standard

Analysis of key technical requirements

1. Role system of participants

The standard clearly defines 6 types of participants: task initiator, scheduler, algorithm provider, data provider, calculator and result user. The calculator must be deployed across management domains, and a single entity cannot concurrently serve as multiple computing nodes.

2. Security model classification

The standard requires that the statistical security parameter l ≥ 30 (2^-30 indistinguishability) and the computational security parameter k ≥ 112 (cracking complexity 2¹¹²) should reach financial-level security strength.

Typical application implementation suggestions

1. Biometric recognition

Process facial feature data through secret sharing technology to achieve:

• Registration stage: original image→calculation factor conversion
• Verification stage: terminal directly uploads calculation factors
• Avoid plaintext transmission and storage of facial data

2. Joint risk control modeling

Financial institutions adopt homomorphic encryption + gradient descent solutions:

1. Local data of each institution remains encrypted
2. Model parameters are exchanged through the MPC protocol
3. Final model accuracy requirement ≥ 32 effective bits

Standard evolution background

This standard fills the gap in the application of MPC technology in the financial field and complements JR/T 0171 "Technical Specifications for the Protection of Personal Financial Information". The key technical indicators refer to the ISO/IEC 29101 privacy architecture framework and are combined with the actual needs of my country's financial industry:

• 2018: The first batch of banks pilot MPC joint anti-fraud
• 2019: Implementation of the national standard for cloud computing (GB/T 32400)
• 2020: This standard is released to regulate the application of MPC in the financial industry