Coding with low density block angular generator matrix

In this work we show that coding with a binary block angular matrix with some extra coupling rows can achieve similar results to codes that use other random binary matrices. Moreover, we also show that it is possible to use sparse block diagonal matrices and get the same results obtained with a dense equiprobable binary random matrix. The block diagonal form of the matrix grants an inherent parallelism for the decoding process. This characteristics can be used to speedup the decoding process with multicore architectures while reducing the communication between the cores. On storage applications it would be possible to perform the reconstruction of local data while minimizing the communication.