A Hybrid Communication Approach for Metadata Exchange in Geo-Distributed Fog Environments

TL;DR

This work proposes HFCS, a novel hybrid communication system that combines hierarchical and peer-to-peer elements, along with edge pools, that outperforms a hierarchical and a P2P approach in task fulfillment at a slight cost to failure detection.

Abstract

Metadata exchange is crucial for efficient geo-distributed fog computing. Existing solutions for metadata exchange overlook geo-awareness or lack adequate failure tolerance. We propose HFCS, a novel hybrid communication system that combines hierarchical and peer-to-peer elements, along with edge pools. HFCS utilizes a gossip protocol for dynamic metadata exchange. In simulation, we investigate the impact of node density and edge pool size on HFCS performance. We observe a performance improvement for clustered node distributions, aligning well with real-world scenarios. HFCS outperforms a hierarchical and a P2P approach in task fulfillment at a slight cost to failure detection.

Figures (12)

• Figure 1: Architecture of HFCS
• Figure 2: Split of an edge pool: if the maximum number of edge pool members would be exceeded by inserting a new node, the edge pool is divided into seven subpools with individual nodes assigned to their closest subpool.
• Figure 3: The average volume of exchanged gossip messages decreases exponentially with the node distance due to the more scattered distribution of nodes across the edge pools. It also leads to an increasing standard deviation up to $d = 10$.
• Figure 4: With continental restrictions there is a distance $d_{max}$ for which $d_1 = d_2$. For any distance between nodes $d > d_{max}$, the maximization of the distance $d_1$ would simultaneously lead to a minimization of $d_2$.
• Figure 5: Geographical distribution of the message exchange volumes of nodes under various maximum node distances $d$ considering agglomerations and continental restrictions with continents assumed as rectangles. Low message exchange volumes of individuals nodes in agglomerations are caused by the split of edge pools into seven subpools which affects either an isolation of individual nodes, or a reduction of communication partners in the corresponding pool so that their message volume in total grows more slowly. Large maximum distances between nodes lead to a more scattered distribution of edge nodes to various pools which also causes a slower growth of the message exchange volume.