Network Graph Generation through Adaptive Clustering and Infection Dynamics: A Step Towards Global Connectivity

TL;DR

This paper introduces a comprehensive approach for deploying network infrastructure in remote areas, utilizing adaptive clustering, infection-inspired connection algorithms, and non-terrestrial backhaul nodes to improve connectivity and network performance.

Contribution

It presents a novel infection dynamics-inspired algorithm for connecting access points efficiently and explores the deployment of non-terrestrial backhaul nodes to enhance network coverage.

Findings

Improved connectivity ratio in remote area deployments

Reduced average hop count and backhaul length

Effective use of non-terrestrial backhaul nodes

Abstract

More than 40% of the world's population is not connected to the internet, majorly due to the lack of adequate infrastructure. Our work aims to bridge this digital divide by proposing solutions for network deployment in remote areas. Specifically, a number of access points (APs) are deployed as an interface between the users and backhaul nodes (BNs). The main challenges include designing the number and location of the APs, and connecting them to the BNs. In order to address these challenges, we first propose a metric called connectivity ratio to assess the quality of the deployment. Next, we propose an agile search algorithm to determine the number of APs that maximizes this metric and perform clustering to find the optimal locations of the APs. Furthermore, we propose a novel algorithm inspired by infection dynamics to connect all the deployed APs to the existing BNs economically. To support the existing terrestrial BNs, we investigate the deployment of non-terrestrial BNs, which further improves the network performance in terms of average hop count, traffic distribution, and backhaul length. Finally, we use real datasets from a remote village to test our solution.