Probabilistic heuristics for disseminating information in networks

TL;DR

This paper compares probabilistic algorithms for information dissemination in networks, proposing a heuristic that improves efficiency by balancing message and time complexities, supported by mathematical analysis and simulations.

Contribution

It introduces a new probabilistic heuristic for network flooding that enhances cost-effectiveness over existing methods.

Findings

The heuristic reduces message complexity compared to flooding.

Mathematical analysis confirms the heuristic's efficiency.

Simulations demonstrate improved trade-offs in random-graph models.

Abstract

We study the problem of disseminating a piece of information through all the nodes of a network, given that it is known originally only to a single node. In the absence of any structural knowledge on the network other than the nodes' neighborhoods, this problem is traditionally solved by flooding all the network's edges. We analyze a recently introduced probabilistic algorithm for flooding and give an alternative probabilistic heuristic that can lead to some cost-effective improvements, like better trade-offs between the message and time complexities involved. We analyze the two algorithms both mathematically and by means of simulations, always within a random-graph framework and considering relevant node-degree distributions.