Scalable and Distributed Cloud Continuum Orchestration for Next-Generation IoT Applications: Latest Advances and Prospects

1. Introduction

With the advent of the Internet of Things (IoT), the centralized cloud computing service delivery paradigm has been gradually transformed into a cloud continuum that includes edge and fog computing and heterogeneous IoT devices with varying computing and power capabilities. At the same time, the dawn of the 5G era requires advanced orchestration solutions that guarantee that the requirements of time-/mission-critical IoT-enabled applications are met. Under this setting, the cloud continuum is defined as the infrastructure that includes remote cloud datacenters, edge clusters, and IoT/end-devices with heterogeneous computing capabilities.

However, several challenges still need to be addressed regarding these capabilities. The orchestration of such complex applications requires, for example, heterogeneous resources spread across the cloud continuum. The core advantage of edge computing is the placement of computational resources at the network edge to alleviate the computation burden of IoT devices, which are also augmented with enhanced computational capabilities. In this context, the orchestration of such applications requires solutions for distributed service embedding, task offloading, resource autoscaling, and service migration in real time and in a scalable fashion. With the evolution of the IoT and the advent of 5G/6G and massive machine-type communications, extremely dense networks are expected to be created, placing even more strain on the available infrastructure across the cloud continuum.

2. Contributions

In this Special Issue, a range of articles tackle interesting challenges and propose novel solutions for various research problems involving the cloud continuum.

Hazrati et al. [1] focus on Information-Centric Networks (ICNs) and the caching problem. They denote data freshness and build a router-based mechanism to compute the cache hit rate and freshness score of the content stored in caches. Furthermore, they introduce an AI-based prediction mechanism to enhance caching efficiency. This proactive approach reduces unnecessary data replacements and ensures that the cache retains the most relevant and current data.

The authors of the second article [2] present a benchmarking framework for assessing deep learning (DL) models, which have been deployed and utilized across the cloud continuum for the delivery of intelligent IoT services. They focus on evaluating DL model scaling across three key performance axes during model inference: classification accuracy, computational overhead, and latency. Working in this direction, they employ several model structures from three popular DL application domains and utilize publicly available and pre-trained model architectures (BERT, EfficientNet, and MLP) and openly available datasets (i.e., GLUE-MRPC and ImageNet).

Sturley et al. [3] share a comparative study on virtualization and containerization in the computing continuum. They focus on these major methods for application deployment at the network edge and compare them, considering several criteria, such as compatibility based on user experience and the ease of installation/deployment, scalability based on the automatic elasticity facing the workload, and energy efficiency in terms of energy and computer resources.

The authors of the fourth article [4] explore the application of collaborative virtual and augmented reality in a cloud continuum context, focusing on designing, implementing, and verifying three reference architectures for deployment in five collaborative VR/AR use cases. These architectures consider the distribution of the computational workload between cloud and edge. The first processes all requests in the cloud, while the second balances the workload between cloud and edge infrastructure. Finally, the third architecture handles the workload entirely at the network edge.

Abinayaa et al. [5] focus on detecting Denial of Service (DDoS) attacks in wireless sensor networks. Their approach utilizes the CatBoost classifier (Cb-C) and the Lyrebird Optimization Algorithm (LOA). The latter is used to optimize the hyperparameters of the Cb-C, which is used as a classification methodology for predicting various types of attacks.

The paper by Kuaban, Godlove Suila et al. investigates the adoption of IoT technologies in the Silicon Mountain technology ecosystem in Cameroon [6]. Through a survey-based approach, the authors identify key barriers such as standardization issues, financial constraints, labor shortages, educational gaps, market challenges, government policies, security concerns, and inadequate power supply. Market challenges, including high marketing costs and socio-economic crises, are particularly significant. The authors suggest strategies such as enhancing tax policies, improving infrastructure, and promoting IoT education to overcome these challenges.

Shahid, Kamal et al. conducted a comparative analysis of routing protocols, namely, EIGRP, OSPF, and BGP, in IPv6-based load-sharing and link failover systems [7]. Using tools such as Wireshark, VMWare, GNS3, and Iperf3, this study evaluated network performance metrics, including convergence time, packet loss, jitter, and delay. The findings reveal that EIGRP achieves superior failover convergence and lower packet loss, while OSPF excels in managing link failures. The study’s route-mapping strategy and recommended BGP hold times provide practical guidance for network engineers in optimizing IPv6 infrastructure.

In their article, Karthick, Gayathri, and Glenford Mapp [8] introduce a Secure Service Ecosystem (SSE) to enable real-time, secure service migration in Intelligent Edge Environments for smart cities. The key mechanisms developed include a Resource Allocation Algorithm, a Resource Allocation Secure Protocol (RASP), and a Secure Service Protocol (SSP). By verifying the protocol’s safety with ProVerif, this study shows the potential of the SSE in enhancing performance and security in smart city applications.

The paper by Tsikteris Sean et al. [9] introduces a novel model for optimizing resource allocation and server selection in multi-access edge computing (MEC) environments. The proposed “TRUST-ME” model leverages a trust-based framework that evaluates the reliability of MEC servers based on both direct interactions and social feedback from other IoT devices, enabling more informed decision-making. An optimistic Q-learning algorithm is employed to allow IoT devices to autonomously select the most suitable MEC server.

Satish et al. [10] addresses the challenge of bufferbloat, which affects the performance of real-time applications such as video conferencing, cloud gaming, and IoT services. The authors propose a novel solution by integrating the Low Latency, Low Loss, and Scalable Throughput (L4S) architecture into the Asynchronous Advantage Actor-Critic (A3C) reinforcement learning algorithm within the FreeBSD networking stack. This approach involves a dual-queue coupled active queue management (AQM) system to manage queues and reduce packet loss, while A3C dynamically adjusts the base drop probability in response to network conditions.

A decentralized global registry system leveraging blockchain and NFT technology to address key challenges in digital asset management, such as security, interoperability, and scalability, is presented in the final article [11]. Through detailed system design and a comparative analysis with traditional centralized registries and existing blockchain-based solutions, this paper highlights the advantages of enhanced security, transparency, and efficiency. Additionally, the legal, ethical, and social implications of NFT-based tokenization are explored, particularly in intellectual property management.

3. Conclusions

This Special Issue on “Scalable and Distributed Cloud Continuum Orchestration for Next-Generation IoT Applications: Latest Advances and Prospects” showcases the latest contributions in the orchestration of cloud continuum- and IoT-based applications. The articles published in this Special Issue span a diverse range of topics, including information-centric networks, deep learning, cyber security, MEC resource orchestration, active queue management, comparative analyses of routing protocols, and smart city applications. These studies collectively highlight how the cloud continuum enables innovative applications and address technical challenges facing 5G and the IoT.