DONS: Dynamic Optimized Neighbor Selection for smart blockchain networks

Abstract

Blockchain (BC) systems mainly depend on the consistent state of the Distributed Ledger (DL) at different logical and physical places of the network. The majority of network nodes need to be enforced to use one or both of the following approaches to remain consistent: (i) to wait for certain delays (i.e. by requesting a hard puzzle solution as in PoW and PoUW, or to wait for random delays as in PoET, etc.) (ii) to propagate shared data through shortest possible paths within the network. The first approach may cause higher energy consumption and/or lower throughput rates if not optimized, and in many cases these features are conventionally fixed. Therefore, it is preferred to enhance the second approach with some optimization. Previous works for this approach have the following drawbacks: they may violate the identity privacy of miners, only locally optimize the Neighbor Selection method (NS), do not consider the dynamicity of the network, or require the nodes to know the precise size of the network at all times. In this paper, we address these issues by proposing a Dynamic and Optimized NS protocol called DONS, using a novel privacy-aware leader election within the public BC called AnoLE, where the leader anonymously solves the The Minimum Spanning Tree problem (MST) of the network in polynomial time. Consequently, miners are informed about the optimum NS according to the current state of network topology. We analytically evaluate the complexity, the security and the privacy of the proposed protocols against state-of-the-art MST solutions for DLs and well known attacks. Additionally, we experimentally show that the proposed protocols outperform state-of-the-art NS solutions for public BCs. Our evaluation shows that the proposed DONS and AnoLE protocols are secure, private, and they acutely outperform all current NS solutions in terms of block finality and fidelity.