Reduce Rounding Off Errors in Information Dispersal Algorithm

The Information Dispersal Algorithm (IDA) is an algorithm that can be used to store files securely. Files and data packets can be broken down at the bit level into several parts and then saved to separate nodes. IDA requires a matrix generator in the form of the Vandermonde matrix and the Cauchy matrix to split and reconstruct files. The matrix inversion process is required to reconstruct files. The implementation of inverse matrix calculation on the machine has limitations. It is the rounding off problem for inverse matrix elements. This causes the reconstructed file is not exactly the same as the file submitted. In this study the use of a triangular matrix consisting of the numbers 0 and 1 is examined. Another method also proposed using bitwise xor based operation. The use of 0-1 triangular matrix and bitwise xord file splitting can avoid the ocurrence of floating point number so that the rounding off mechanism is prevented and the reconstructed file can be exactly the same as the original file.