Privacy-Preserving Multi-Party Analytics via Secure Sum Protocol
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional solutions for privacy-preserving multi-party analytics on horizontally partitioned data struggle to scale for generic analytical tasks, especially when more than two parties are involved, due to increased computational complexity and lack of applicability to big data scenarios.
Innovation Solution
A secure protocol using a non-trusted mediator and collusion-resistant data anonymization, where each party generates and shares public and private keys to encrypt and shuffle data segments, allowing for secure summation and optimization of multi-party data through a secure sum protocol, enabling privacy-preserving analytics across multiple parties.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional privacy-preserving protocols are used for multi-party analytics, then data privacy is protected, but computational complexity increases significantly when more than two parties are involved
Solution Approach 1:
The patent divides the data into horizontal partitions where each party holds a subset of data rows with complete attributes. This segmentation allows each party to perform local computations independently, reducing the computational burden on any single party while maintaining overall system privacy. The secure sum protocol further segments the aggregation process into controlled steps with the mediator.
Solution Approach 2:
The patent introduces a trusted mediator that facilitates secure computations between multiple parties. The mediator receives encrypted data from parties, performs secure summation operations, and distributes results back to parties. This intermediary approach simplifies the complexity of multi-party interactions by centralizing the coordination function, making the system scalable beyond two parties.
2Adaptability or versatility
If conventional protocols are applied to generic analytical tasks, then specific tasks can be solved, but scalability to big data analytics is difficult
Solution Approach 1:
The patent designs a universal framework that can handle various analytical tasks (sums, averages, min, max, etc.) through a single secure sum protocol architecture. The protocol is not task-specific but rather provides a general-purpose mechanism that can be applied to different analytical queries by adjusting the aggregation function, thereby achieving both versatility and scalability.
Solution Approach 2:
The patent combines multiple parties' data partitions into a unified analytical result through the secure sum protocol. By merging the computational results from all parties in a privacy-preserving manner, the system achieves big data analytics capabilities while maintaining the privacy benefits of distributed processing. The mediator consolidates results from all parties to produce comprehensive analytics.
3Adaptability or versatility
If data is shared across multiple parties for collaborative analytics, then broader understanding is achieved, but data privacy and security concerns increase
Solution Approach 1:
The trusted mediator acts as a security intermediary that handles all data aggregation operations. Parties share data encrypted with the mediator's public key, and the mediator performs secure summation without accessing plaintext data from multiple parties simultaneously. This intermediary structure enables collaborative analytics while preventing direct exposure of sensitive data between parties.
Solution Approach 2:
The patent uses encrypted copies of data that can be processed without revealing the original data content. Each party provides encrypted data copies to the mediator, which performs computations on these encrypted copies. This copying approach allows collaborative analysis while maintaining data privacy, as the mediator works with ciphertext rather than plaintext data.
Data Source
AI summary
Examples disclosed herein relate to: computing, by a computing device at a party among a plurality of parties, a sum of local data owned by the party. The local data is horizontally partitioned into a plurality of data segments, with each data segment representing a non-overlapping subset of data entries owned by a particular party; computing a local gradient based on the horizontally partitioned local data; initializing each data segment; anonymizing aggregated local gradients received from the mediator, wherein the aggregated local gradients comprise gradients computed based on a plurality of data entries owned by the plurality of parties; receiving, from a mediator, a global gradient based on the aggregated local gradients; learning a global analytic model based on the global gradient; and performing privacy-preserving multi-party analytics on the horizontally partitioned local data based on the learned global analytic model.