Equilibrium Selection in Multistage Congestion Games for Real-Time Streaming

We model peer-to-peer real-time streaming as a multistage congestion game with strategies consisting of time-sequences of network nodes (or peers), and identify a large set of strategy profiles through which the whole content reaches all peers in a minimum number of time-periods. In order to make these profiles sustainable as equilibria, we provide strategy restriction mechanisms implementing those equilibria where both streaming duration and congestion are minimized. Their functioning is exemplified in a simple simulation environment. The potential and social cost of equilibrium without strategy restrictions are also investigated.