Results 1 
6 of
6
Gossip versus Deterministic Flooding: Low Message Overhead and High Reliability for Broadcasting on Small Networks
"... Rumor mongering (also known as gossip) is an epidemiological protocol that implements broadcasting with a reliability that can be very high. Rumor mongering is attractive because it is generic, scalable, adapts well to failures and recoveries, and has a reliability that gracefully degrades with t ..."
Abstract

Cited by 41 (0 self)
 Add to MetaCart
Rumor mongering (also known as gossip) is an epidemiological protocol that implements broadcasting with a reliability that can be very high. Rumor mongering is attractive because it is generic, scalable, adapts well to failures and recoveries, and has a reliability that gracefully degrades with the number of failures in a run. However, rumor mongering uses random selection for communications. We study the impact of using random selection in this paper. We present a protocol that superficially resembles rumor mongering but is deterministic. We show that this new protocol has most of the same attractions as rumor mongering. The one attraction that rumor mongering hasnamely graceful degradationcomes at a high cost in terms of the number of messages sent. We compare the two approaches both at an abstract level and in terms of how they perform in an Ethernet and small wide area network of Ethernets.
Adaptive Broadcasting With Faulty Nodes
, 1996
"... We consider broadcasting from a faultfree source to all nodes of a completely connected nnode network in the presence of k faulty nodes. Every node can communicate with at most one other node in a unit of time and during this period every pair of communicating nodes can exchange information packe ..."
Abstract

Cited by 15 (1 self)
 Add to MetaCart
We consider broadcasting from a faultfree source to all nodes of a completely connected nnode network in the presence of k faulty nodes. Every node can communicate with at most one other node in a unit of time and during this period every pair of communicating nodes can exchange information packets. Faulty nodes can receive information but cannot send it. Broadcasting is adaptive, i.e. a node schedules its next communication on the basis of information currently available to it. We study worst case running time of faulttolerant broadcasting algorithms in two models: in the wakeup model only nodes already having the source message can call other nodes and in the unrestricted model all faultfree nodes can call. In the first model we give an optimal algorithm working in time k +dlog(n \Gamma k)e. In the second model we give an algorithm working in time O(log 2 n), whenever the fraction of faulty nodes is bounded by a constant smaller than 1. Key words: adaptive, algorithm, broadc...
Reliable Broadcasting
 Disc. Appl. Math
, 1994
"... Broadcasting is a process of information dissemination in a communications network whereby a message, originated by one member, is transmitted to all members of the network. By adding some redundant calls to the broadcasting scheme, the completion of the broadcast can be guaranteed in the presence o ..."
Abstract

Cited by 4 (2 self)
 Add to MetaCart
Broadcasting is a process of information dissemination in a communications network whereby a message, originated by one member, is transmitted to all members of the network. By adding some redundant calls to the broadcasting scheme, the completion of the broadcast can be guaranteed in the presence of faulty components. We investigate the implications of transmission failures on broadcasting. In particular, we consider broadcasting when the number of transmission failures is bounded by a constant. We determine the time required to guarantee a broadcast in this model. We also study the number of links required in networks which allow reliable broadcasting. 1 Introduction A graph G = (V; E) represents a communications network in which the vertices in V correspond to the members of the network and the edges in E correspond to communication links connecting pairs of members. In broadcasting, one member of the network, the originator, has a message which is to be transmitted to all of the o...
Broadcasting With Linearly Bounded Transmission Faults
"... We consider broadcasting with a linearly bounded number of transmission failures. For a constant parameter 0 ! ff ! 1 we assume that at most ffi faulty transmissions can occur during the first i time units of the communication process, for every natural number i. Every informed node can transmit inf ..."
Abstract

Cited by 4 (0 self)
 Add to MetaCart
We consider broadcasting with a linearly bounded number of transmission failures. For a constant parameter 0 ! ff ! 1 we assume that at most ffi faulty transmissions can occur during the first i time units of the communication process, for every natural number i. Every informed node can transmit information to at most one neighbor in a unit of time. Faulty transmissions have no effect. We investigate worstcase optimal broadcasting time under this fault model, for several communication networks. We show, for example, that for the nnode line network this time is linear in n, if ff ! 1 2 , and exponential otherwise. For the hypercube and the complete graph, broadcasting in the linearly bounded fault model can be performed in time logarithmic in the number of nodes. y Instytut Informatyki, Uniwersytet Warszawski, Banacha 2, 02097 Warszawa, Poland. This work was done during the author's stay at the Universit'e du Qu'ebec `a Hull as a postdoctoral fellow. z D'epartement d'Informatiqu...
Broadcasting With a Bounded Fraction of Faulty Nodes
, 1995
"... We consider the problem of broadcasting a message from one node of a completely connected (n+1)node network to all other nodes. We adopt the 1port halfduplex model in which every node can communicate with at most one neighbor in a unit of time and during this period one of the communicating node ..."
Abstract

Cited by 4 (1 self)
 Add to MetaCart
We consider the problem of broadcasting a message from one node of a completely connected (n+1)node network to all other nodes. We adopt the 1port halfduplex model in which every node can communicate with at most one neighbor in a unit of time and during this period one of the communicating nodes can only send information and the other can only receive it. The source is faultfree but all other nodes are subject to permanent failures. A faulty node can receive the message but cannot relay it. The fraction of faulty modes is bounded by a constant. We consider nonadaptive broadcasting algorithms working correctly in the presence of faulty nodes and investigate their worstcase running time. We also show lower bounds for broadcasting time under this scenario. Our main result is the construction of a faulttolerant broadcasting algorithm whose running time is less than 1.73 times larger than optimal, for sufficiently large n. y Instytut Informatyki, Uniwersytet Warszawski, Banacha 2,...
APPROVAL
, 2000
"... ii When communication lines in an interconnection network are prone to random transmission failures, the time required to broadcast a message throughout the network from a given source vertex is uncertain. We view this broadcast time in such unreliable networks as a random variable dependent on th ..."
Abstract
 Add to MetaCart
ii When communication lines in an interconnection network are prone to random transmission failures, the time required to broadcast a message throughout the network from a given source vertex is uncertain. We view this broadcast time in such unreliable networks as a random variable dependent on the strategy employed by individual nodes to inform their neighbours, and we seek to describe those strategies which minimize its expected value. Two major results are the complete characterization of optimal deterministic broadcast strategies in unreliable star networks, and proof that each arborescence network has a unique optimal randomized strategy. Our introductory treatment of this problem combines aspects of Markov chains, convexity theory, and the coupon collectorâ€™s problem. We hope and expect these tools to prove valuable as work continues on several open problems. iii Acknowledgments