Curve fitting based efficient parameter selection for robust provable data possession

摘要

Cyberspace faces a series of threats that can spread rapidly across distributed systems. Many such transmissible cyber threats aim to damage users' data. In recent years, the popularity of cloud computing has driven a lot of users to save their data in the cloud. The centralization of users' data in the cloud has created new opportunities and incentives for transmissible cyber threats targeting users' data. In this context, in addition to cloud vendor's security mechanisms, it is important to allow users to efficiently verify the integrity of their data saved in the cloud. The seminal sampling based PDP (Provable Data Possession) scheme can attain effective probabilistic verification with high computational efficiency for the integrity of users' data saved in the cloud by use of a set of randomly sampled data blocks. By integrating with the forward error correcting (FEC) technique, a recently-proposed robust PDP scheme can protect against arbitrarily small amounts of data corruption and has therefore been widely adopted in practice. It is a core task to determine the number of sample blocks in this scheme because this parameter plays a fundamental role in balancing the security of the scheme and the computational cost. A smaller value can comprise the security while a larger one can incur extra high computational cost. Existing work mainly leverages the Monte Carlo methods to estimate this parameter. However, these methods suffer from heavy computational cost. In this paper, we propose a method to determine this parameter based on the curve fitting technique. Specifically, we formally analyze the parameter selection process of the robust PDP scheme, and leverage the curve fitting technique to improve the efficiency of parameter selection while ensuring the optimality of the number of samples for the robustness of the scheme. Extensive experimental results demonstrate the effectiveness and efficiency of our approach. Specifically, it can be 25 times faster than the existing solution for 1, 000, 000 times simulated attacks. (C) 2018 Elsevier Inc. All rights reserved.