Abstract-This paper introduces a Combinatory Optimization Problem (COP) which captures the performance in cooperation of a P2P Streaming Network, considered at the buffer level. A new family of strategies of cooperation is defined, which represents the order of query of pieces in a communication between peers. We define a measure of quality for each strategy, based on the most shocking video streaming parameters, namely the continuity and the start-up latency. Our aim is o find quasi-optimal strategies inside the proposed family, by means of on a model based on an Asymmetric Traveling Salesman Problem (ATSP). Finally, we solve this ATSP using an ACO (Ant Colony Optimization)-based algorithm. The results show substantial improvement with this approach, once compared with the ones obtained by previous defined strategies of cooperation.
Abstract-This work proposes a new piece selection strategy for improving latency and continuity in a P2P streaming network. Different piece selection strategies based on a simple and symmetrical model for sharing are considered. It is well known the scalability properties of Rarest First and low values of latency of Greedy. In this work these techniques have been revised, and a new family of strategies is proposed. The richness of this family is expressed in the Approximation Strategy Property, which shows that the shape of every feasible strategy can be approximated by one member of this family proposed. It is introduced a Permutation Finder Algorithm, which looks for strategies inside this family that achieve a tradeoff between continuity and latency, not found in previous related works.
Peer-to-peer networks are strongly based on cooperation. The users, called peers, communicate basically in a three-level based policy. In the first one, peers discover others interested in the same content, and is called swarm selection strategy (or swarming). Then, peers must select the best ones to cooperate, what is called peer selection strategy. Finally, peers cooperate sending pieces to each other, and the planning must attend the piece selection strategy. In this paper we propose an extension of a simple model based on cooperation for peer-to-peer video streaming networks. We assume that the swarming classifies peers according to their bandwidth. In this model we meet both the peer and the piece selection strategies, for simplified scenarios. The aim is to design network policies in order to achieve the highest continuity of video reproduction when peers reach a stationary state. We show that under full knowledge, the network can scale even under free-riding effects. At the same time, we provide theoretical results that reveal Rarest First has a poor performance in comparison with other techniques. Finally, we analyze the scalability in a worst-case scenario when a variable amount of special peers are included in the network.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.