Parallel Computation Of The Homomorphic Hash Function And Its Application

Homomorphic Hash Function (HHF) is a class of function with homogeneity,and they can be viewed as functions mapping variable length message to fixed-length Hash value or the message digest. The HHF was proposed which based on discrete logarithm problem. And it needs a lot of power calculation in the practical application, which results in many problems such as time and bandwidth, so that it has low efficiency. To improve the performance of HHF in practical applications, existing literature mainly focus on the using of technology, such as batch, or using platform to accelerate. To improve the speed of modular operation effectively, Montgomery algorithm was used to avoid modular operation. The modular power algorithm was improved based on the improved Montgomery algorithm in this thesis. The algorithm’s parallelism was analyzed based on CUDA platform. When the data’s integrity need to be verified in the cloud storage, cloud just needs multiply a few pieces of the hash value randomly, and then sends the result to the verifier. The verifier can use the homogeneity to judge whether the stored data integrated or not if we use HHF. Existing verifiable methods can achieve the purpose of verification. If we want cloud storage used more flexibly, data privacy and dynamic update should be considered. Although the existing schemes involve different aspects of the data privacy and dynamic update, they did not give attention to the three parties which the schemes involve. So designing a scheme which can resist loss attacking, tamper attacking, collusion attacking, and support data update has more realistic significance.Improving HHF’s computing performance can reduce the HHF’s computation time significantly and make its applications more widely. The data which stored in the cloud may be faced with loss attack or tamper attack which are caused by the cloud service provider, or the collusion attack which is caused by the third party verifier and the cloud service provider. To solve these problems, the security of the data is very important. In addition, if the data which stored in the cloud can realize dynamic update, the cloud storage can be used more flexibly.Kong et al. proposed an improved Montgomery algorithm, and the algorithm was used to optimize modular exponentiation algorithm. The theoretical analysis shows that, the number of multiplication which has a higher complexity has decreased by 25%. And a parallel algorithm of HHF was designed. In addition, its parallelism based on CUDA was analyzed combing with the characteristics of GPU. At the same time, a protocol was designed to achieve the public data integrity verification based on cloud storage, HHF and encryption algorithm. This protocol adopted the method of encryption to protect the privacy of the data and resist the cloud service provider’s loss attack and tamper attacks. It can also resist the collusion attack which was caused by cloud service provider and the third party verifier. What’s more, the index table was used to implement the dynamic update of the data.