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.

Exact Distributed Load Centrality Computation: Algorithms, Convergence, and Applications to Distance Vector Routing

Ghiro Lorenzo;Lo Cigno Renato
2020-01-01

Abstract

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.
File in questo prodotto:
File Dimensione Formato  
main.pdf

accesso aperto

Descrizione: Articolo
Tipologia: Documento in Post-print
Licenza: PUBBLICO - Pubblico con Copyright
Dimensione 3.15 MB
Formato Adobe PDF
3.15 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11379/537909
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 5
  • ???jsp.display-item.citation.isi??? 5
social impact