中文摘要：

云存储服务中,批量审计是高效验证云端数据完整性的关键技术.批量审计容易遭受“失效文件”攻击,并且查询失效文件代价高、速度慢,严重影响着批量审计方案的可用性和效率.针对该问题,提出一种支持失效文件快速查询的批量审计方法,该方法通过建立批量审计过程的关联性,改变了二分查询树中右孩子节点的计算方式,减少了整个查找过程的批量审计次数;并在批量审计过程中执行幂指测试,通过一次审计就可完成含有单个失效文件的子树查找过程,有效缩短了子树的查找长度;采用混合型查询方法,根据历史查询信息设置幂指测试的深度,降低了“失效文件聚集处”的查询开销.安全分析和性能表明,该方法能够快速完成失效文件定位,有效抵抗“失效文件”攻击,保证了批量审计方案的可用性和效率.在少量文件失效的情景下,相较于简单二分查找方法,文中方法耗费的批量审计次数减少了30%.

英文摘要：

Batch auditing is the key technology to efficiently verify data integrity in the cloud storage service. However, the batch audit schemes are vulnerable to ＂invalid files＂ attacks. Searching invalid files brings heavy cost and the search speed is slow, which seriously affect the availability and efficiency of hatch audit. Especially, the system still has to run many batch auditing processes to search the invalid files when there are only a few bad files in them. It is the common phenomenon in commercial cloud storage service, which generally does not appear large area of damaged files since the provider may try their best to avoid the worst case. So we focus mainly on the batch auditing with only a small few of invalid files in the process. To solve this problem, this paper proposes a batch audit scheme with fast searching invalid files （FSBA） in cloud storage service. Through establishing the relationship of batch audit process, this method changes the calculation of right child nodes of binary search trees in order to reduce the number of Batch Verification （BV） in the whole the search process. We can get the results of the right nodes by lightweight computing using the intermediate results instead of running the heavy task of BV. If there is only one invalid file in a batch auditing, it may waste a lot of times by using the binary search method. Because it has to follow the search path to verify the validity of the nodes until the leaf node, which only contains the invalid file. By executing exponents test in batch audit, our method can finish the search of the sub-tree containing one invalid file only through an audit. It can effectively shorten the sub-tree search length. If there are more than one invalid file in the sub-tree, it can result in the failure of running exponents test. In order to significantly reduce the impact of the side effects, we propose the hybrid method called hybrid binary fast search （H-BFS）. According to the historical query information, the hybrid search method set t