RC-Gossip: Information Freshness in Clustered Networks with Rate-Changing Gossip

TL;DR

This paper introduces RC-Gossip, a rate-changing gossip mechanism that adaptively directs information dissemination in clustered networks, significantly improving data freshness compared to traditional methods.

Contribution

It proposes a novel rate-changing gossip protocol and a renewal-reward analysis approach for evaluating information freshness in clustered networks.

Findings

RC-Gossip increases network freshness significantly.

Optimal cluster sizes enhance the effectiveness of RC-Gossip.

Renewal-reward analysis offers an alternative to SHS for studying gossip networks.

Abstract

A clustered gossip network is considered in which a source updates its information over time, and end-nodes, organized in clusters through clusterheads, are keeping track of it. The goal for the nodes is to remain as fresh as possible, i.e., have the same information as the source, which we assess by the long-term average binary freshness metric. We introduce a smart mechanism of information dissemination which we coin rate-changing gossip (RC-Gossip). Its main idea is that gossiping is directed towards nodes that need it the most, and hence the rate of gossiping changes based on the number of fresh nodes in the network at a given time. While Stochastic Hybrid System (SHS) analysis has been the norm in studying freshness of gossip networks, we present an equivalent way to analyze freshness using a renewal-reward-based approach. Using that, we show that RC-gossip significantly increases freshness of nodes in different clustered networks, with optimal cluster sizes, compared to traditional gossiping techniques.