Characterizing latency in periodic P2P hypercube gossiping

The efficient structure of the hypercube allows for an optimal solution to the two-way gossip problem at logarithmic node degree. It thereby enables scalable dissemination of live streaming data in peer-to-peer networks. Applications range from audio conferencing to distributed object tracking or real-time business intelligence. For such real-time applications traversal time is a concern. A considerable portion is due to wait delay or bide time at intermediate nodes. Previous work has analyzed abstract effort for hypercube gossiping schemes and derived worst-case bounds for several timing modes. In this paper we analyze the hop count distribution to complement previously established worst-case times by expectation values for uniformly distributed network delays. We compare these results to results from a simulation with a life-like network delay distribution. Results attest that expected wait latency is indeed highly dependent on the timing of the communication and considerably lower than worst-case traversal time.