Advanced Compression Technique for Random Data
Blaise M Crowly · International Journal on Information Theory · 2015
This paper proposes a technique to compress any data irrespective of it's type.Compressing random data especially has always proved to be a difficult task.With very less patterns and logic within the data it quickly reaches a point where no more data can be represented within a given number of bits.The proposed technique will allow us to compress irrespective of it's pattern or logic within, and represent it.Further it will permit the technique to be again applied to the already compressed data without having any change in the possible compression ratio.While data can't be compressed further than a limit the technique will rely on representing the data as a position in any computationally easy number series that extends to infinity provided it have high enough deviation among it's digits.Only few markers that are linked to the position is saved, rather than representing the original data.The system then use these markers to guess the position and derive the data from it.The procedure is however, computationally intensive and as of now can raise questions of data corruption but with more computing power, efficient algorithm and proper data integrity checks it will be able to provide very high compression ratios in the future .