英文摘要: |
As the supporting architecture of cryptocurrencies, blockchain is also showing its promising potential to be applied to mobile devices. Being WiFi supported, mobile devices nowadays are able to perform ad-hoc local communications in multi-hop manners, thus can potentially extend cryptocurrency service into areas without stable Internet connection. However, traditional blockchain architecture suffers from the intolerably high block propagation latency and communication cost which will be even trickier if adopted in wireless multi-hop networks (WMNs). This paper aims to reduce this latency and cost by leveraging node weights to distinguish the feedback speed of different nodes and constructing a low-length multicast tree to select a subset of nodes with higher feedback speed to participate in the block verification. Thus, block propagation is depicted by the minimum length multicast tree, intrinsically the Steiner Tree Problem. The primary challenge lies in that the WMNs only allow local communication and the distributed consensus protocol of the blockchain makes a predetermination of receiver nodes impossible. We design our algorithm via a toward source'' Steiner tree approach in favor of the distributed environment. Tree construction proceeds by progressively enlarging the searching areas until the accumulated weight of receiver nodes reaches a threshold. Our algorithm provably returns a tree length fairly close to that of the optimal Steiner tree with latency O(root nlog n) (where n represents the number of nodes in the network), the best so far. Furthermore, our algorithm is empirically validated to provide instructive insights for efficient block propagation with applications in Bitcoin network. |