GellCast: Gossip-enhanced low-latency multicast for virtual English classroom

Low latency requirement, dynamic group members and limited out-degree of nodes pose great challenge on P2P solutions to virtual interactive learning environments. Most application level multicast protocols are heuristic algorithms of minimum spanning tree, aiming to reach the optimal parameters like latency concerning all nodes. However, few studies are focusing on providing stringent low perceived latency based on specific features of audio group communication. In this paper, we propose a gossip-enhanced multicast protocol called GellCast for a virtual English classroom to provide “natural” group discussions in problem solving. Making realistic presumptions on processing delay at each node, GellCast guarantees stringent low latency among discussion participants while providing relatively low latency among listeners. Hence, it provides satisfying low perceived latency to all users. We then present simulation results that show the feasibility and efficiency of the approach.