An approximation to the greedy algorithm for differential compression of very large files

This paper presents a new differential compression algorithm that combines the hash value and suffix array technique. In this algorithm, hash values for every block of the reference file is computed. Next, suffix arrays on these block hash values are computed. This algorithm finds the longest matches for every offset of the version file. This algorithm depends upon the utilization of three new data structures, the block hash table, the quick index array, and the pointer array, which improves the run-time of the algorithm, and compress very large files.