We provide a prototype system for a distributed storage model, the goal of this system is to store information in a p2p network of n nodes, and the original information could be recovered later in a computationally simple way from (1+e)k of nodes for some small e>0. For solving this problem, we employ a class of B-J codes with dimension 2 as base code and obtain a new class of LDPC code by using q-tuples to substitute the elements of , to modify and improve Raptor code. The improved Raptor Code has many advantages such as high decoding rate, flexibility in choices of the code length and rate of. As the benefits of the above characters, the storage system based on modified Raptor code has great improvement in decoding probability and parameter flexibility, and a simulation is also illustrated to prove our assumption.
A Comparison of Raptor Code Using LDGM and LDPC code
