A delay-tolerant network architecture for edge computing with applications in narrow band internet of things

Abstract

The increasing use of Internet of Things (IoT) applications had generated significant traffic delays and large amounts of data, impacting the delivery and efficiency of these applications. This necessitated faster response times and minimal delays in packet transmission. Fog devices, responsible for immediate data transmission, computation, and storage, were considered potential solutions to these challenges. However, research into fog and edge computing models was still in the early stages, requiring further exploration to unlock their potential for various IoT applications. This study aimed to determine the most suitable model for building highly available, fault-tolerant networks by designing and evaluating the performance of the CUBIC and BBR algorithms, proposing a novel delay-tolerant network (DTN) architecture for edge and fog computing, specifically designed to run IoT applications. At the heart of this solution was the design and evaluation of two rate-limiting algorithms, CUBIC (Cubic) and BBR (Bottleneck Bandwidth and Round-trip propagation time), on edge network nodes CUBIC is a TCP congestion control algorithm, named for the cubic function it uses to manage network congestion. These algorithms were integrated with bandwidth management techniques within a lightweight Kubernetes (K3s) cluster environment. Specifically, Narrowband Internet of Things (NB-IoT) using the SIM7020E module was employed in a K3s cluster for edge computing. A rate-limiting method with Cilium on a K3s cluster of six nodes acted as a rate limiter for layers 3 and 4 of the of the Open Systems Interconnection (OSI) model, using a bandwidth manager for K3s service pods on the network port to prevent Distributed Denial of Service (DDoS) and Internet Protocol (IP) flooding attacks. Quantitative methods were employed to evaluate the effectiveness of the proposed DTN solution. The evaluation model was designed from the engineering and design process as a research method. Data were collected using Prometheus and Fortio, a load testing tool, within a simulated IP flooding attack environment. Data analysis utilised information and system theories to test and validate the empirical data gathered for fog and edge networks, focusing on delay in fault-tolerant networks. The study made significant contributions across theoretical and practical domains. Theoretically, it introduced a new Delay-Tolerant Network (DTN) architecture specifically designed for Internet of Things (IoT) applications. Practically, it demonstrated the effectiveness of newly designed rate-limiting algorithms in reducing network delays and mitigating potential attacks that could cripple the network and the application running on it.