Repeated uncoordinated information dissemination by flooding
Donald M. Topkis · Networks · 1995
Abstract A dynamic database in a communications network consists of a set of messages, where a sequence of different versions of each message is generated over time in a repeated and uncoordinated process and where the appearances of the various message versions are distributed among the nodes. The message versions are to be disseminated in the network so that each message version eventually resides at each node. Flooding is a distributed procedure for disseminating message versions. The use of flooding to disseminate repeated and uncoordinated message versions is an integral part of a generic adaptive routing mechanism similar to that used in ARPANET and proposed for the INTERNET and is an element of a method proposed for improving survivability in intra‐LATA networks. A model is formulated to analyze the performance of the dissemination by flooding of repeated and uncoordinated message versions. The worst‐case time complexity is established, showing that flooding is an optimal procedure in terms of this measure. Special properties of flooding in a tree are also established. A key parameter is a separator, which is a lower bound on the time interval between the initial generation of a particular version of any message and the initial generation of the next version of that message.