A Cost-Effective and Reliable Cloud Storage

Abstract

The project aims to provide a scalable, reliable and cost effective cloud storage solution based on a Low Density Parity Check (LDPC) code-based framework. The novelties of the project lie in the following aspects. Firstly, the proposed framework utilizes a new technique called dynamic parameterization so that the existing resources can be used more efficiently. Secondly, a tailored error correction code with localized property is specifically designed to minimize the cost occurred during encoding and decoding for the distributed storage system. Thirdly, a neuroevolution approach is proposed, combining artificial neural network learning algorithm with evolutionary method, to develop predictive models for dynamic resource allocation and performance optimization.