Many optimization techniques for networking protocols take advantage of topological information to improve performance. Often, the topological information at the core of these techniques is a centrality metric such as the Betweenness Centrality (BC) index. BC is, in fact, a centrality metric with many well-known successful applications documented in the literature, from resource allocation to routing. To compute BC, however, each node must run a centralized algorithm and needs to have the global topological knowledge; such requirements limit the feasibility of optimization procedures based on BC. To overcome restrictions of this kind, we present a novel distributed algorithm that requires only local information to compute an alternative similar metric, called Load Centrality (LC). We present the new algorithm together with a proof of its convergence and the analysis of its time complexity. The proposed algorithm is general enough to be integrated with any distance vector (DV) routing protocol. In support of this claim, we provide an implementation on top of Babel, a real-world DV protocol. We use this implementation in an emulation framework to show how LC can be exploited to reduce Babel's convergence time upon node failure, without increasing control overhead. As a key step towards the adoption of centrality-based optimization for routing, we study how the algorithm can be incrementally introduced in a network running a DV routing protocol. We show that even when only a small fraction of nodes participate in the protocol, the algorithm accurately ranks nodes according to their centrality.

中文翻译：

---

精确分布负载中心计算：算法、收敛和距离矢量路由的应用

许多网络协议的优化技术利用拓扑信息来提高性能。通常，这些技术核心的拓扑信息是中心性度量，例如介数中心性 (BC) 指数。事实上，BC 是一个中心性度量，在文献中记录了许多众所周知的成功应用，从资源分配到路由。但是，要计算 BC，每个节点都必须运行集中式算法，并且需要具有全局拓扑知识；这些要求限制了基于 BC 的优化程序的可行性。为了克服这种限制，我们提出了一种新的分布式算法，该算法只需要本地信息来计算替代的类似度量，称为负载中心性 (LC)。我们展示了新算法及其收敛性证明和时间复杂度分析。所提出的算法足够通用，可以与任何距离矢量 (DV) 路由协议集成。为了支持这一说法，我们在 Babel 之上提供了一个实现，这是一种现实世界的 DV 协议。我们在仿真框架中使用此实现来展示如何利用 LC 来减少节点故障时 Babel 的收敛时间，而不会增加控制开销。作为采用基于中心性的路由优化的关键步骤，我们研究了如何在运行 DV 路由协议的网络中逐步引入该算法。我们表明，即使只有一小部分节点参与协议，该算法也会根据节点的中心性准确地对节点进行排名。