The problem of maximizing the rate of a broadcast operation in peer-to-peer (P2P) systems that are bounded by a node degree is studied in this paper. The authors “address this problem by providing the first distributed solution that achieves near-optimal broadcast rate under arbitrary node degree bounds.” The solution mainly includes two distributed algorithms: (1) “a network-coding-based broadcasting algorithm” that uses the topology of the network to optimize the broadcast rate based on that topology, and (2) “a Markov-chain guided topology hopping algorithm that optimizes the topology.” Further, the paper proves that the proposed solution achieves optimality, “while previously proposed distributed algorithms obtain optimality only for P2P complete graphs.”

The paper is well written and flows well; the authors provided enough background and explanations for the different components and aspects of the paper. The different steps and their theoretical implications in building the proposed solution/algorithms are well described.

Even though the paper is targeting transmission control protocol/Internet protocol (TCP/IP) P2P-based network infrastructures, I am left wondering how the proposed solution behaves on high-performance computing (HPC)-based systems using InfiniBand (IB)/remote direct memory access (RDMA)-based networks. In other words, how can the proposed design be enhanced to take advantage of the hardware multicast support available with IB networks?

Further, as the evaluation of the proposed algorithms was a simulation, I am interested in the complexity involved in implementing and integrating them in a software framework.