[Paper Review] Infinite Choices of Data Aggregations with Linear Number of Keys.

Taeho Jung, Xiang‐Yang Li|arXiv (Cornell University)|Aug 28, 2013

Privacy-Preserving Technologies in Data23 references1 citations

TL;DR

This paper proposes a privacy-preserving data aggregation protocol that enables any subset of n participants to compute accurate sums and products of their inputs using only θ(n) keys, requiring just one communication round. Unlike prior methods, it operates without a trusted third party or secure channel, offering robustness against eavesdropping while significantly reducing communication complexity compared to O(n)-round alternatives.

ABSTRACT

Privacy-preserving data aggregation has long been a hot research issue. It is becoming increasingly important due to the widespread data collection for various analysis purposes. In this paper, we present a novel arithmetic protocol which computes sum and product of n individuals’ input values without disclosing them, which is in turn used to develop an efficient accurate model to aggregate the data in a privacy-preserving manner. Unlike other approaches, our model initiates from an environment without secure communication channel but is robust to the eavesdrop attacks, and it does not rely on a trusted third party either. After the keys are prepared, only 1 communication round is needed to conduct each aggregation while some approaches require O(n) rounds. Notably, we allow any subset of n participants to privately conduct accurate data aggregation with only θ(n) keys while similar works let every participant generate and hold θ(2) keys or more.

Motivation & Objective

• To address the challenge of efficient and private data aggregation in environments without a trusted third party or secure communication channels.
• To reduce the communication complexity of existing protocols, which often require O(n) rounds of interaction.
• To minimize the number of keys per participant, allowing only θ(n) keys per participant instead of θ(2) or more as in prior work.
• To enable accurate aggregation of data values—specifically sums and products—while preserving individual input privacy.

Proposed method

• The protocol uses a novel arithmetic protocol to compute the sum and product of n individuals' inputs without revealing the inputs themselves.
• It operates in a setting without a secure communication channel, making it resilient to eavesdropping attacks.
• The method relies on a key preparation phase that enables private aggregation using only θ(n) keys per participant.
• Only one communication round is required per aggregation, significantly reducing interaction overhead compared to multi-round protocols.
• The protocol ensures correctness and privacy by leveraging mathematical properties of the aggregation functions and key distribution.
• It supports any subset of n participants to perform aggregation independently, without requiring full participation or centralized coordination.

Experimental results

Research questions

• RQ1 Can data aggregation be performed accurately and privately with only a linear number of keys per participant?
• RQ2 How can privacy be preserved in a setting without a trusted third party or secure communication channel?
• RQ3 Can the number of communication rounds be reduced to a single round while maintaining accuracy and security?
• RQ4 What is the minimal key requirement per participant to support private aggregation of sums and products?
• RQ5 How does the protocol scale in terms of communication efficiency and key management compared to existing approaches?

Key findings

• The protocol achieves accurate data aggregation using only θ(n) keys per participant, significantly reducing key overhead compared to prior works that require θ(2) or more keys per participant.
• Only one communication round is needed per aggregation, reducing communication complexity from O(n) rounds to a constant round complexity.
• The protocol is robust against eavesdropping attacks and does not require a secure communication channel.
• It operates without a trusted third party, enhancing decentralization and trust minimization.
• The method supports any subset of participants to perform aggregation independently, increasing flexibility and scalability.
• The approach maintains strong privacy guarantees by ensuring individual inputs remain hidden during the aggregation process.