Research On Sampling Algorithms Of Data Integrity Verification In Cloud Storage

Cloud computing provides dynamic and scalable virtualized resources and services via theInternet, so it provides a new model for large data storage and management. More and more datais stored in the cloud server, so that their ownerscan enjoy this convenient service.However, dataowners have to relinquish their effective control over the fate of the data, so that the data stored inthe cloud storage service provided by untrusted servers is urgently faced with the challenges of theintegrity check problem, especially when data is stored in an untrusted cloud service. Dataintegrity verification has become an important means to solve this problem.Data integrity verificationunder cloud storage formass datacurrently uses the simple randomsampling method for sampling check. The specific implementation is, at first,to divide the file intoa certain number of datablocks, and then do select so as to respond to the underlying data’slost ordamagethat may occur while the data isfrequently updated. However, when dealing with datablocks’ lost or damage occur in a row, this sampling method issomewhat lagging behind. Thispaperproposes a random sampling-Markovdecisionprocess (MDP) verification method, whichisunder the modelof three party verificationensuring the publicity and security.By analyzing thestate of each individual data block and the correlation between the damaged blocks, therandom-Markov decision process strategy to extract data blocksis proposed up in order to solvethe problem that by simple random verification, it needs a long time to finddata blocks’ continuousloss or damage. Furthermore, in terms of the number of sample data blocks, the number ofverification rounds and the time interval between each verification (sampling) round, the methodproposed in this papercarried out further optimization.Finally, the theoretical analysis and experimental demonstrate that in the same conditions ofcalculation costand network transmission overhead, using diffusion strategy and MDP method to extract data blocksin the data integrity verification, inface of the continuous damaged data blocks,during the same experiment time,it can significantly reduce the calibration time verifying the samenumber of bad data blocks, and can detect more bad data blocks in the same time. And in face ofnon-continuous data blocks’corruption, it is similar withthe simple random method on theintegritytime required and the number of bad data blocksdetected.In a word, this data sampling algorithmproposed can improve the efficiency ofdata integrity verification.