Super Generation Network Coding for Peer-to-Peer Content Distribution Networks

In peer-to-peer (P2P) content distribution systems, network coding is known as a helpful method for increasing the content availability, accelerating the download process, and robustness against churn. Originally, dense network coding (DNC) has been proposed and theoretically considered as an optimal solution. However, due to its huge computational overhead, it is not viable for real-world systems. Subsequently, sparse, generations, and overlapped generations network coding schemes are proposed as possible alternatives but at reduced performance compared to that provided by the DNC. Further in this article, an improved applicable network coding scheme for P2P content distribution systems referred to therein as Super Generation Network Coding (SGNC) is proposed. SGNC maximizes the generation size so that it is as close as possible to the optimal size without adding computational overhead. Theoretical analysis and experimental work show that SGNC outperforms classical and all previous coding based schemes for P2P content distribution systems in terms of content availability, download time, overhead, and decodability for all piece scheduling policies.