Erasure-control coding for distributed networks

Channel coding has seen itself quickly emerge as a very promising technique in relatively new territory: distributed networks, where computing, communication and storage are extensively involved. In this environment, the distribution of rich media files to a large user population and the distribution of mission-critical data present special challenges for service providers and enterprises. Ensuring quality of service to these applications requires a combination of reliability, speed and scalability. Traditional packet-based networks endure poor end-to-end performance when packet-loss rates are high and a simple error-control strategy (through re-transmission) is applied. The paper will show how erasure-control coding, a special channel coding technique, plays an important part in promoting fault tolerance, and consequently quality of service across the network. Additionally, the construction of a class of maximum-distance-separable (MDS) array erasure codes is presented, which can be used in an adaptive way to efficiently control packet losses in a fluctuating network environment.