Multi-Server ORAM Using XOR PIR and Secret Sharing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-server Oblivious Random Access Machines (ORAMs) lack security against malicious adversaries and incur high communication or computation overhead, limiting their practicality in real-world applications.
Innovation Solution
A novel multi-server ORAM system utilizing an XOR-based Private Information Retrieval (PIR) protocol for data retrieval and a linear secret sharing scheme with homomorphic computation for eviction, ensuring security and efficiency against malicious adversaries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If existing multi-server ORAM constructions are used, then client bandwidth blowup is reduced to O(1), but computation overhead becomes expensive due to homomorphic encryptions
Solution Approach 1:
The system segments the homomorphic encryption operations between client and servers. The client performs only lightweight operations while servers handle the expensive homomorphic computations using their stored secret shares, dividing the computational burden and reducing client overhead.
Solution Approach 2:
The servers act as intermediaries that perform the computationally intensive homomorphic encryption operations on behalf of the client. By using the servers as mediators with pre-stored secret shares, the client avoids direct computation of expensive homomorphic operations.
2Productivity
If existing multi-server ORAM constructions are used, then efficiency is improved, but security is limited to semi-honest adversaries only
Solution Approach 1:
The system applies preliminary anti-action by having servers store secret shares and perform authenticated computations before any potential malicious attack. The authentication mechanisms and verification protocols are built into the computation flow in advance, preventing malicious adversaries from exploiting vulnerabilities.
Solution Approach 2:
The system implements feedback through authentication protocols where servers verify client requests and clients verify server responses. This bidirectional feedback mechanism ensures that even if some servers behave maliciously, the system can detect and respond to such behavior, maintaining security guarantees.
3Reliability
If O(log N) client-server communication blowup is used, then security is maintained, but communication overhead increases
Solution Approach 1:
The system transitions from a single-server model to a multi-server architecture, adding the dimension of distribution. By spreading the storage and computation across multiple servers, the system achieves both reduced communication overhead for the client and maintained security through the collective security properties of the distributed system.
Data Source
AI summary
An oblivious distributed file system is provided using an oblivious random access machine (ORAM), including an ORAM balanced tree structure, where each node in the tree is configured to store data blocks, the structure including at least two shares. The system also includes at least two ORAM servers, each of the servers configured to communicate with a client ORAM device, and programmed to facilitate storage of a different subset of the shares of the tree structure using a distributed file system and to implement an access procedure of a tree-based ORAM using the tree structure, including a retrieval phase and an eviction phase. In the retrieval phase, the servers utilize an authenticated Private Information Retrieval (PIR) protocol to retrieve data blocks as requested from the client ORAM device. In the eviction phase, the servers utilize a linear secret sharing scheme.