Browsing by Author "Ruomu Hou"

Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    LYRATA: Proof-of-Useful-Bandwidth Blockchain Against Persistent Adversaries
    (2026-11-05) Ruomu Hou; Yucheng Sun; Haifeng Yu
    Permissionless blockchains are subject to sybil attacks. They typically rely on Proof-of-X (such as Proof-of-Work), for sybil defense. Here "X" is a resource which the adversary only has limited supply. While effective, such Proof-of-X systems often result in significant resource waste. To avoid such waste, thiswork explores sybil defense based on in-protocol bandwidth consumption. Here in-protocol bandwidth consumption refers to bandwidth that is already consumed by the blockchain for providing the basic functionality of a distributed ledger. As our main contribution, this work proposes a novel blockchain LYRATA, with a simple yet elegant design. To our knowledge, LYRATA is the very first blockchain that can i) minimize resource waste by exploiting in-protocol bandwidth consumption, and ii) offer strong security guarantees under standard persistent adversaries. We have further implemented a research prototype of LYRATA. Our experimental results show that LYRATA also offers good practical end-to-end performance.
  • No Thumbnail Available
    Item
    Robust and Low-degree Overlay for Secure Flooding Against Resource-bounded Adversaries
    (2024-09-17) Yucheng Sun; Ruomu Hou; Haifeng Yu
    The security of large-scale blockchains relies on successful message flooding among the honest parties, on an overlay topology. The crux of doing such flooding successfully is to have a robust and low-degree overlay topology: Robust means that even if the malicious parties all refuse to relay messages, the remaining honest parties should still constitute a connected component in the overlay network. Low-degree means that the nodes in the overlay network should have relatively small node degrees. The central challenge of designing such robust and low-degree topology is that in permissionless blockchain context, the adversary is often bounded by resource (such as computation power or stake), rather than by the total number of malicious parties. We show that existing works of designing robust overlay against such resource-bounded adversaries all require excessively large node degrees (e.g., 25000 or more under real-world settings). As our main contribution, we propose a novel LOR overlay topology that is robust against such resource- bounded adversaries. Our design is the very first such design with practically-feasible node degrees (e.g., 200 to 400 under real-world settings