Issues in Peer-to-Peer Networking: A Coding Optimization Approach

In this paper we consider a linear optimization approach for studying download finish times in peer-to-peer networks that allow but do not require coding. We demonstrate that using the network coding framework simplifies analysis even in scenarios where the optimal solution does not require coding. For example, we use the network coding framework to disprove the claim of Ezovski et al. that in the absence of coding, the sequential minimization of file download times minimizes the average finish time over all users. We also use this framework to study the effect of requiring reciprocity, a typical feature of incentive-compatible protocols. Lastly, we show that for a dynamically changing network scenario, coding can provide a robust and optimal solution that outperforms routing.