Data Clean Room Privacy Budget Partitioning Against Averaging Attacks

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Differential privacy in data clean rooms is vulnerable to averaging attacks, necessitating the need for effective privacy budgeting to restrict the number of allowed computations and protect sensitive data within acceptable bounds.

Innovation Solution

Implementing a method that partitions database tables based on data addition, assigning distinct privacy allowances to each partition, and comparing specified privacy currency against remaining allowances on a partition-by-partition basis to allow or disallow queries, ensuring that the total information revealed remains within acceptable limits.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If differential privacy is implemented without partitioning, then data utility is improved through broader access, but privacy protection deteriorates due to vulnerability to averaging attacks

Engineering Contradiction:
Improvedata utilityVSAvoidprivacy protection
Core Design Contradiction:
Loss of informationVSReliability

Solution Approach 1:

The patent divides the database into multiple partitions based on data addition time, with each partition having its own independent privacy budget. This segmentation prevents averaging attacks by isolating queries to specific time periods while still allowing cumulative data utility across partitions.

Inventive Principle:
Principle #1Segmentation

2Device complexity

If a single privacy budget is applied to the entire database, then implementation complexity is reduced, but adaptability deteriorates when new data is added

Engineering Contradiction:
Improveprivacy budget management complexityVSAvoidadaptability to new data
Core Design Contradiction:
Device complexityVSAdaptability or versatility

Solution Approach 1:

The patent implements a dynamic privacy budget allocation system where the database is automatically partitioned when new data is added. Each partition receives an appropriate privacy budget based on its specific characteristics, allowing the system to adapt to changing data requirements while maintaining manageable complexity through automated partition management.

Inventive Principle:
Principle #15Dynamics

3Reliability

If partitioning is implemented for privacy budgeting, then privacy protection is improved through attack prevention, but device complexity increases due to partition management

Engineering Contradiction:
Improveprivacy protectionVSAvoidpartition management complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements an automated partition management system that self-adjusts when new data is added to the database. The system automatically creates new partitions, allocates appropriate privacy budgets, and updates query routing without requiring manual intervention, thereby maintaining strong privacy protection while minimizing the operational complexity burden.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP4672063A1Privacy budget allocation in a data clean room
Publication Date: 2025.12.31 GOOGLE LLC
  • EP4672063A1 patent drawingFigure 1A~1B
  • EP4672063A1 patent drawingFigure 2
  • EP4672063A1 patent drawingFigure 3~4

AI summary

Methods and systems for restricting queries to a database according to a privacy budget. The technology includes assigning a first privacy allowance to first data in a database table, the first privacy allowance being an amount of a privacy currency, and assigning a second privacy allowance to second data in the database table; receiving a query from a database user, the query including a specified amount of the privacy currency; and allowing processing of the query only when, for each one of the first data and second data that must be accessed to service the query, the specified amount of privacy currency is equal to or less than a remaining privacy allowance for the data.