Secure Multi-Party Computation for Analytics-as-a-Service

Abstract

Secure Multi-Party Computation (MPC) offers a compelling cryptographic framework enabling multiple parties to collaboratively compute analytical results while preserving the privacy of each participant’s data. In the context of Analytics-as-a-Service (AaaS), MPC promises a secure path for clients to benefit from powerful analytics without exposing private inputs. This paper examines pre-2020 developments in MPC tailored to AaaS scenarios, including lightweight MPC applications, hybrid protocols, and real-world deployments. A notable instance is the deployment of an MPC web application in Boston for analyzing wage disparities—allowing organizations to compute aggregated statistics without revealing individual data points OpenBU. Tools like Conclave (2019) offer a hybrid approach, compiling analytical queries into a combination of local plaintext operations and MPC steps, dramatically improving scalability for big-data analytics arXiv. Foundational MPC principles—including secret sharing, homomorphic encryption, and garbled circuits—support the development of these secure analytic services DatatasWikipedia. This paper synthesizes such research, providing a focused literature review, and summarizes advantages (e.g., privacy, regulatory compliance), challenges (e.g., computational and communication overhead), methodological strategies, and deployment results. The analysis concludes with reflections on future trajectories that might enable broader adoption of MPC-based AaaS.