Broadcast gossip algorithms for consensus

Tuncer Can Aysal, Mehmet Ercan Yildiz, Anand D. Sarwate, Anna Scaglione

Research output: Contribution to journalArticle

299 Citations (Scopus)

Abstract

Motivated by applications to wireless sensor, peer-to-peer, and ad hoc networks, we study distributed broadcasting algorithms for exchanging information and computing in an arbitrarily connected network of nodes. Specifically, we study a broadcasting-based gossiping algorithm to compute the (possibly weighted) average of the initial measurements of the nodes at every node in the network. We show that the broadcast gossip algorithm converges almost surely to a consensus. We prove that the random consensus value is, in expectation, the average of initial node measurements and that it can be made arbitrarily close to this value in mean squared error sense, under a balanced connectivity model and by trading off convergence speed with accuracy of the computation. We provide theoretical and numerical results on the mean square error performance, on the convergence rate and study the effect of the "mixing parameter" on the convergence rate of the broadcast gossip algorithm. The results indicate that the mean squared error strictly decreases through iterations until the consensus is achieved. Finally, we assess and compare the communication cost of the broadcast gossip algorithm to achieve a given distance to consensus through theoretical and numerical results.

Original languageEnglish (US)
Pages (from-to)2748-2761
Number of pages14
JournalIEEE Transactions on Signal Processing
Volume57
Issue number7
DOIs
StatePublished - 2009
Externally publishedYes

Fingerprint

Broadcasting
Ad hoc networks
Mean square error
Communication
Sensors
Costs

Keywords

  • Broadcasting
  • Distributed average consensus
  • Gossip algorithms
  • Sensor networks

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Signal Processing

Cite this

Broadcast gossip algorithms for consensus. / Aysal, Tuncer Can; Yildiz, Mehmet Ercan; Sarwate, Anand D.; Scaglione, Anna.

In: IEEE Transactions on Signal Processing, Vol. 57, No. 7, 2009, p. 2748-2761.

Research output: Contribution to journalArticle

Aysal, Tuncer Can ; Yildiz, Mehmet Ercan ; Sarwate, Anand D. ; Scaglione, Anna. / Broadcast gossip algorithms for consensus. In: IEEE Transactions on Signal Processing. 2009 ; Vol. 57, No. 7. pp. 2748-2761.
@article{2409a0fc382b423e86c51eb9755bad98,
title = "Broadcast gossip algorithms for consensus",
abstract = "Motivated by applications to wireless sensor, peer-to-peer, and ad hoc networks, we study distributed broadcasting algorithms for exchanging information and computing in an arbitrarily connected network of nodes. Specifically, we study a broadcasting-based gossiping algorithm to compute the (possibly weighted) average of the initial measurements of the nodes at every node in the network. We show that the broadcast gossip algorithm converges almost surely to a consensus. We prove that the random consensus value is, in expectation, the average of initial node measurements and that it can be made arbitrarily close to this value in mean squared error sense, under a balanced connectivity model and by trading off convergence speed with accuracy of the computation. We provide theoretical and numerical results on the mean square error performance, on the convergence rate and study the effect of the {"}mixing parameter{"} on the convergence rate of the broadcast gossip algorithm. The results indicate that the mean squared error strictly decreases through iterations until the consensus is achieved. Finally, we assess and compare the communication cost of the broadcast gossip algorithm to achieve a given distance to consensus through theoretical and numerical results.",
keywords = "Broadcasting, Distributed average consensus, Gossip algorithms, Sensor networks",
author = "Aysal, {Tuncer Can} and Yildiz, {Mehmet Ercan} and Sarwate, {Anand D.} and Anna Scaglione",
year = "2009",
doi = "10.1109/TSP.2009.2016247",
language = "English (US)",
volume = "57",
pages = "2748--2761",
journal = "IEEE Transactions on Signal Processing",
issn = "1053-587X",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "7",

}

TY - JOUR

T1 - Broadcast gossip algorithms for consensus

AU - Aysal, Tuncer Can

AU - Yildiz, Mehmet Ercan

AU - Sarwate, Anand D.

AU - Scaglione, Anna

PY - 2009

Y1 - 2009

N2 - Motivated by applications to wireless sensor, peer-to-peer, and ad hoc networks, we study distributed broadcasting algorithms for exchanging information and computing in an arbitrarily connected network of nodes. Specifically, we study a broadcasting-based gossiping algorithm to compute the (possibly weighted) average of the initial measurements of the nodes at every node in the network. We show that the broadcast gossip algorithm converges almost surely to a consensus. We prove that the random consensus value is, in expectation, the average of initial node measurements and that it can be made arbitrarily close to this value in mean squared error sense, under a balanced connectivity model and by trading off convergence speed with accuracy of the computation. We provide theoretical and numerical results on the mean square error performance, on the convergence rate and study the effect of the "mixing parameter" on the convergence rate of the broadcast gossip algorithm. The results indicate that the mean squared error strictly decreases through iterations until the consensus is achieved. Finally, we assess and compare the communication cost of the broadcast gossip algorithm to achieve a given distance to consensus through theoretical and numerical results.

AB - Motivated by applications to wireless sensor, peer-to-peer, and ad hoc networks, we study distributed broadcasting algorithms for exchanging information and computing in an arbitrarily connected network of nodes. Specifically, we study a broadcasting-based gossiping algorithm to compute the (possibly weighted) average of the initial measurements of the nodes at every node in the network. We show that the broadcast gossip algorithm converges almost surely to a consensus. We prove that the random consensus value is, in expectation, the average of initial node measurements and that it can be made arbitrarily close to this value in mean squared error sense, under a balanced connectivity model and by trading off convergence speed with accuracy of the computation. We provide theoretical and numerical results on the mean square error performance, on the convergence rate and study the effect of the "mixing parameter" on the convergence rate of the broadcast gossip algorithm. The results indicate that the mean squared error strictly decreases through iterations until the consensus is achieved. Finally, we assess and compare the communication cost of the broadcast gossip algorithm to achieve a given distance to consensus through theoretical and numerical results.

KW - Broadcasting

KW - Distributed average consensus

KW - Gossip algorithms

KW - Sensor networks

UR - http://www.scopus.com/inward/record.url?scp=67650188107&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=67650188107&partnerID=8YFLogxK

U2 - 10.1109/TSP.2009.2016247

DO - 10.1109/TSP.2009.2016247

M3 - Article

AN - SCOPUS:67650188107

VL - 57

SP - 2748

EP - 2761

JO - IEEE Transactions on Signal Processing

JF - IEEE Transactions on Signal Processing

SN - 1053-587X

IS - 7

ER -