A study of distributed storage systems with network coding in wireless networks

Increasing multimedia applications on the networks e.g., video and voice over the Internet, have attracted lots of interest to the distributed storage systems. Recently, lots of research has been performed on applying network coding to distributed storage systems. Although there are intensive theoretical studies in this field, very few practical simulations have been reported. We simulate distributed storage systems using network coding. In particular, we first use random linear network in reconstructing a file and in the repair process. We observe that probability of successful repair and successful downloading approaches to one for a large finite field size. In addition, we apply a lower complexity code and measure their probability of successful downloading, probability of successful repair, repair time and processing time. Our numerical results show for random linear codes, connecting to more storage nodes can substantially reduce the required finite field size. This leads to lower coding complexity.