Abstract
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.
Highlights
P2P content distribution systems, known as file sharing, have become a popular and effective alternative to the classical server-client file sharing approach
̆ We propose a novel network coding scheme, referred to as super generation network coding (SGNC), that maintains the computational cost at reasonable level and boosts the robustness of the network by improving pieces diversity and availability
We consider that all the peers have same data rate and the download time can be measured in terms of number of rounds
Summary
P2P content distribution systems, known as file sharing, have become a popular and effective alternative to the classical server-client file sharing approach. The second important problem is that the nature of P2P content distribution networks is dynamic, and peers can depart suddenly which is known as dynamic peers participation or churn [6]. This in turn may affect the download time or or even prevent the completion of the download. In an effort to minimize the computational complexity, information pieces are partitioned into mutually exclusive subsets referred to as groups or generations, and coding is done only within a generation This approach scales down the encoding and decoding problem from the whole file size to the generation size. The concept of generations in network coding was first proposed by Chou et al in [11]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
