Privacy-preserving Autonomous Aggregate Data Analytics in Untrusted Cloud
Primary Investigator:
Bharat Bhargava
Ganapathy Mani, Denis Ulybyshev, Bharat Bhargava, Jason Kobes, Puneet Goyal
Abstract
Intelligent Autonomous Systems (IAS) are highly reflexive and very cognizant about their limitations and capabilities, interactions with neighboring entities, as well as the interactions with its operational environment. IAS should be able to conduct data analytics and update policies based on those analytics. These tasks should be performed autonomously i.e. with limited or no human intervention. In this paper, we introduce advanced aggregate analytics over untrusted cloud and autonomous policy updates as a result of those analytics. We will be using Active Bundle (AB), a distributed self-protecting entity, wrapped with policy enforcement engine as our implementation service. We propose an algorithm that can enable individual ABs to grant or limit permissions to their AB peers and provide them with access to anonymized data to conduct analytics autonomously. When these processes take place, ABs do not need to rely on policy enforcement engine every time, which increases scalability. This workflow also creates an AB environment that is decentralized, privacy-preserving, and autonomous.