TY - GEN
T1 - When infinite flow is sufficient for ergodicity
AU - Touri, Behrouz
AU - Nedić, Angelia
PY - 2010
Y1 - 2010
N2 - We consider the consensus and ergodicity for a random linear discrete-time system driven by stochastic matrices. We focus on independent models with certain properties, and we study the ergodicity and consensus of such random models through a novel property, termed infinite flow property. Our key result is the establishment that for a class of independent random models, this property is a necessary and sufficient condition for ergodicity. Using this result, we show that the ergodicity of these models and the ergodicity of their expected models are the same. The result provides us with new tools for studying various aspects of dynamic networks and beyond. We demonstrate the potential use of our key result through several different applications. In particular, we apply it to provide a generalization of the randomized gossip algorithm and to study a consensus over a dynamic network with link failures. Also, we use the result to investigate necessary and sufficient conditions for the ergodicity of an equal-neighbor average algorithm on Erdös-Rényi random graphs. Finally, we demonstrate that our result can be employed to provide an alternative proof of the second Borel-Cantelli lemma.
AB - We consider the consensus and ergodicity for a random linear discrete-time system driven by stochastic matrices. We focus on independent models with certain properties, and we study the ergodicity and consensus of such random models through a novel property, termed infinite flow property. Our key result is the establishment that for a class of independent random models, this property is a necessary and sufficient condition for ergodicity. Using this result, we show that the ergodicity of these models and the ergodicity of their expected models are the same. The result provides us with new tools for studying various aspects of dynamic networks and beyond. We demonstrate the potential use of our key result through several different applications. In particular, we apply it to provide a generalization of the randomized gossip algorithm and to study a consensus over a dynamic network with link failures. Also, we use the result to investigate necessary and sufficient conditions for the ergodicity of an equal-neighbor average algorithm on Erdös-Rényi random graphs. Finally, we demonstrate that our result can be employed to provide an alternative proof of the second Borel-Cantelli lemma.
KW - Borel-Cantelli lemma
KW - Ergodicity, product of random stochastic matrices
KW - Gossip algorithm
KW - Linear random model
KW - Link failure
UR - http://www.scopus.com/inward/record.url?scp=79953131935&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=79953131935&partnerID=8YFLogxK
U2 - 10.1109/CDC.2010.5717769
DO - 10.1109/CDC.2010.5717769
M3 - Conference contribution
AN - SCOPUS:79953131935
SN - 9781424477456
T3 - Proceedings of the IEEE Conference on Decision and Control
SP - 7479
EP - 7486
BT - 2010 49th IEEE Conference on Decision and Control, CDC 2010
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 49th IEEE Conference on Decision and Control, CDC 2010
Y2 - 15 December 2010 through 17 December 2010
ER -