ANALYSIS OF MULTI-PARTY COMPUTATION PERFORMANCE ON ONLINE STORES BASED ON HOMOMORPHIC ENCRYPTION

• Main Author: Falah Sutawindaya, Mohamad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/66572
• Institution: Institut Teknologi Bandung

The problem of data breaches is increasingly widespread, online service providers often miss out, resulting in data service users being exposed. Online services can also be referred to as a cloud system called semi-trusted-cloud-source, homomorphic encryption is considered as one of the answers to provide solutions for exposed data as well as practicality in carrying out operations on the data. Various architectures of cloud systems can be implemented homomorphically, one of which is Multi-Party Computation (MPC) combined with homomorphic encryption that can provide internal data security for semi-trusted source services (various architectural models have their advantages and disadvantages). The CKKS algorithm was chosen because of its operating features on available data types and a better language that allows more scenarios to be applied to the MPC system. Seeing how big the impact of homomorphic implementation on hardware is also taken into account because it is related to costs in providing service providers, this study draws several conclusions about how much the additional cost is, which briefly requires one additional piece of hardware as a service worker for the needs of one cycle homomorphic computation. This paper also mentions how big the impact on storage requirements as one of the MPC-based homomorphic encryption modules. In the end, the implementation test can still provide an MPC system that runs well which involves computational homomorphism for its use with the scenario of storing transaction and balance data encrypted in the case study of "online store".