On Using Seeders for P2P Live Streaming

TL;DR

This paper investigates the feasibility and performance of using seeders in peer-to-peer live streaming, providing theoretical models and analyzing the efficiency of seeding strategies in such systems.

Contribution

It introduces the concept of live seeding in P2P streaming, formalizes its efficiency, and extends theoretical analysis to systems with multiple seeders considering overhead models.

Findings

Feasibility of live seeding demonstrated through theoretical analysis

Efficiency models for single and multiple seeders developed

Overhead impacts on seeding performance quantified

Abstract

Seeders (peers that do not request anything but contribute to the system) are a powerful concept in peer-to-peer (P2P). They allow to leverage the capacities of a P2P system. While seeding is a natural idea for filesharing or video-on-demand applications, it seems somehow counter-intuitive in the context of live streaming. This paper aims at describing the feasibility and performance of P2P live seeding. After a formal definition of "live seeding" and efficiency, we consider the theoretical performance of systems where the overhead is neglected. We then propose a linear overhead model and extend the results for this model, for a single seeder and for a set of seeders as well (it is not always possible to perfectly aggregate individual efficiencies in a given system).