SCAR - Scatter, Conceal and Recover
This thesis describes a secure and reliable method for storing data in a distributed hash table (DHT) that leverages the inherent properties of the DHT to provide a secure storage substrate. The framework presented is referred to as "Scatter, Conceal, and Recover" (SCAR). The standard method of securing data in a DHT is to encrypt the data using symmetrical encryption before storing it in the network. SCAR provides this level of security, but also prevents any known cryptoanalysis from being performed. It does this by breaking the data into smaller blocks and scattering these blocks throughout the DHT. Hence, SCAR prevents any unauthorized user from obtaining the entire encrypted data block. SCAR uses hash chains to determine the storage locations for the data blocks within the DHT. To ensure storage availability, SCAR uses an erasure coding scheme to provide full data recovery given only partial block recovery. This thesis presents the details of SCAR. First, the framework, related protocols, and mechanisms are described. Second, a prototype implementation is presented showing the feasibility of SCAR. Third, analytical models are discussed that characterize SCAR's behavior, the models are then validated using experimental results. Lastly, the models are analyzed to further understand the tradeoff between data security and data availability. The exploration of this tradeoff leads to the conclusion that SCAR can effectively balance this tradeoff when the nodes of the network are "sufficiently" available.
Advisor:Daniel Mossé; Taieb Znati; Ahmed Amer
School:University of Pittsburgh
School Location:USA - Pennsylvania
Source Type:Master's Thesis
Date of Publication:06/14/2007