An improved Vandermonde Decoding PE for Distributed Storage System

An improved Vandermonde decoding PE for distributed storage system is proposed in this paper. The PE first give a decoding algorithm based on matrix partition and Lagrange interpolation method to split the complex matrix invention to small Vandermonde matrix's parallel parameter calculation. PE architecture is constructed by hardware working manner in Galois field of GF (2
⁸
). The PE is designed to work for both encoding and decoding with same module but different data and operation. The PE number is configurable to have best performance in different erasure codes “r” situations backward compatible by the control of data block distribution and Vandermonde erasure codes data selection. The hardware architecture is fully verified at RTL level and synthesized with GF 12nm technology lib and simulation results showed 60.6% to 79% speed advantages compared to Gaussian elimination.