Research On Parallelization Encoding And Decoding Of Cyclic Codes(BCH)

With the rapid development of information technology, the amount of informationthe world is growing at an unprecedented rate, how to store and handle such a hugeamount of data, new technologies and methods is developed one by another, cloudstorage and MapReduce (a parallel computing framework) are representative of thesetechnologies. Cloud storage is a storage system which built on a distributed filesystem, it has many characteristics: large capacity, diversity, timeliness, accuracy etc.Distributed File System is combined of many distributed nodes, the traditionalvalidation and correction method(Such as: Parity、Hamming parity etc) of storage cannot meet the requirements of distributed file system. How to verify and correct errorthe data more effectively in distributed file system is even more urgent. In this paper,main topic focuses on parallelized encoding and decoding of BCH code on distributedfile stored data in large data environments.This paper mainly from the perspective of linear block code encoding and decod-ing parallelization, make use of the characteristics of block code, it’s include tow asp-ects:(1) Parallelization of block code encoding;(2) Parallelization of block codedecoding; In this paper the problems of storage scheme of current distributed filesystem is in-depth studied, and with the shortage of capacity calibration and correc-tion, a fault tolerance mechanisms which based on BCH cycle code is proposed. Thismechanisms can solve the problems of the lack of parity and error correction mechan-ism in Distributed File System and the wasted space when multiple copies are stored.On basis of analyzing the characteristics of big data and block codes, a parallel encod-ing and decoding algorithm which based on lookup table is proposed, encoding anddecoding scheme is designed based on the MapReduce parallel-framework andencoding and decoding scheme which base on pieces and lookup-table is designed, arealization of the scheme is proposed.Finally, this paper analyzes the performance of the scheme. At result of analysis,encoding and decoding performance has been greatly improved, the problem of datavalidation and error correction of large data effectively solved.