Optimizing Resource Allocation and Energy Efficiency in Federated Fog Computing for IoT

Taimoor Ahmad 1 3 , Anas Ali 2 · 2025

Read Optimizing Resource Allocation and Energy Efficiency in Federated Fog Computing for IoT by Taimoor Ahmad 1 3 , Anas Ali 2 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-18100.txt.

Description

Abstract Address Resolution Protocol (ARP) spoofing attacks severely threaten Internet of Things (IoT) networks by allowing attackers to intercept, modify, or block communications. Traditional detection methods are insufficient due to high false positives and poor adaptability. This research proposes a multi-layered machine learning-based framework for intelligently detecting ARP spoofing in IoT networks.

Catalog record

CREATORTaimoor Ahmad 1 3 , Anas Ali 2
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE20 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-18100

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Optimizing Resource Allocation and Energy Efficiency in Federated Fog Computing for IoT Taimoor Ahmad 1 3 , Anas Ali 2 Abstract Address Resolution Protocol (ARP) spoofing attacks severely threaten Internet of Things (IoT) networks by allowing attackers to intercept, modify, or block communications. Traditional detection methods are insufficient due to high false positives and poor adaptability. This research proposes a multi-layered machine learning-based framework for intelligently detecting ARP spoofing in IoT networks. Our approach utilizes an ensemble of classifiers organized into multiple layers, each layer optimizing detection accuracy and reducing false alarms. Experimental evaluations demonstrate significant improvements in detection accuracy (up to 97.5%), reduced false positive rates (less than 2%), and faster detection time compared to existing methods. Our key contributions include introducing multi-layer ensemble classifiers specifically tuned for IoT networks, systematically addressing dataset imbalance problems, introducing a dynamic feedback mechanism for classifier retraining, and validating practical applicability through extensive simulations. This research enhances security management in IoT deployments, providing robust defenses against ARP spoofing attacks and improving reliability and trust in IoT environments. I Introduction The Internet of Things (IoT) has become a transformative technology, reshaping how devices interact, communicate, and share data across numerous applications, including healthcare, industrial automation, smart homes, and smart cities. IoT networks integrate billions of interconnected devices, each generating and consuming vast amounts of data. While IoT brings unprecedented convenience, efficiency, and innovation, it simultaneously introduces substantial security vulnerabilities. Among these vulnerabilities, Address Resolution Protocol (ARP) spoofing represents a significant threat due to its ease of execution and potentially catastrophic impacts on IoT network security [ 13 ] . ARP spoofing is a form of cyber-attack where an attacker sends forged ARP messages onto a local area network. The objective is to link the attacker’s MAC address with the IP address of another host, causing traffic intended for the legitimate host to be incorrectly routed to the attacker [ 14 ] . Consequently, this allows attackers to intercept, modify, or block communication within the network, leading to serious breaches of confidentiality, integrity, and availability. The simplicity of ARP combined with its lack of built-in security measures makes IoT networks particularly susceptible to ARP spoofing attacks. Traditional defenses against ARP spoofing involve static ARP tables and cryptographic solutions, which, while effective under certain conditions, do not scale well to large, dynamic IoT environments [ 15 , 1 ] . These methods often lead to increased overhead, reduced network performance, and fail to adequately address the dynamic and resource-constrained nature of IoT devices. Moreover, static ARP tables are cumbersome and difficult to maintain in large-scale networks, leading to management inefficiencies. Recent advancements in machine learning (ML) techniques offer promising avenues for addressing these challenges, providing adaptive and intelligent solutions for identifying and mitigating ARP spoofing attacks in IoT networks. For instance, supervised machine learning algorithms like Support Vector Machines (SVMs), decision trees, and neural networks have been employed for anomaly detection in network security, yielding positive results [ 16 , 2 ] . These approaches leverage historical data and network patterns to detect anomalies indicative of malicious activities. However, while machine learning techniques have demonstrated potential in enhancing IoT security, several limitations persist. First, IoT networks are characterized by heterogeneous devices generating varied and often unpredictable traffic patterns, complicating anomaly detection [ 17 ] . Second, IoT datasets are often imbalanced, with significantly more benign traffic instances than malicious ones, leading to biased classifiers and higher false positive rates [ 18 ] . Third, existing ML solutions frequently fail to adapt quickly to evolving network dynamics, thus limiting their effectiveness against novel and sophisticated attack vectors [ 19 ] . Furthermore, anomaly detection in IoT environments frequently misclassifies benign anomalies as malicious due to the unique behaviors of IoT devices. For instance, intermittent connectivity, varied power usage patterns, and diverse operational protocols create false alarms in traditional anomaly detection mechanisms [ 10 ] . Thus, developing solutions specifically tailored for IoT networks that can efficiently distinguish genuine attacks from benign anomalies is critical. This paper explicitly addresses the challenge of accurately and efficiently detecting ARP spoofing attacks in IoT networks, minimizing false positives, and ensuring robust adaptability. Addressing this challenge is vital given the pervasive deployment of IoT devices across sectors where security breaches can have devastating real-world consequences. For example, in healthcare, compromised IoT devices could disrupt patient monitoring systems, leading to life-threatening scenarios. In industrial IoT settings, ARP spoofing could sabotage manufacturing processes, causing significant financial and operational losses [ 11 ] . To overcome these limitations, we propose an intelligent detection framework based on multi-layered ensemble machine learning techniques specifically designed for IoT networks. Our proposed model combines multiple classifiers into hierarchical layers, enhancing detection accuracy and minimizing false positives. Each layer in our ensemble approach is strategically designed to capture different aspects of IoT network behavior, leveraging decision trees for interpretability, random forests for robustness, and neural networks for adaptability and complex pattern recognition. Key differentiators of our approach compared to existing literature include: • A multi-layered ensemble classifier specifically optimized for heterogeneous IoT traffic. • An adaptive resampling method that effectively mitigates dataset imbalance, ensuring equitable representation of malicious and benign traffic. • A dynamic feedback loop mechanism for continuous classifier retraining, enhancing adaptability against evolving attack vectors. • Extensive simulation-based validation demonstrating significant improvements in detection accuracy, computational efficiency, and robustness compared to state-of-the-art techniques. The primary contributions of this research are: • Development of a robust and accurate multi-layered machine learning framework tailored explicitly for IoT environments, significantly enhancing ARP spoofing detection accuracy. • Introduction of an innovative adaptive resampling technique to address the dataset imbalance problem prevalent in IoT network traffic analysis. • Proposal and integration of a dynamic feedback mechanism that allows continuous learning and adaptation of classifiers to newly observed network behavior and threats. • Practical validation of our model through extensive comparative experiments against existing credible techniques, demonstrating superior performance in realistic IoT network scenarios. The remainder of this paper is structured as follows: Section II reviews recent relevant literature, highlighting gaps and motivating our proposed solution. Section III presents the detailed system model, including mathematical formulations, notation, and algorithmic descriptions. Section IV describes the experimental setup, simulation parameters, results, and detailed comparative analysis. Finally, Section V concludes the paper and outlines future research directions. II Related Work Research into ARP spoofing detection and mitigation in IoT environments has intensified as the adoption of IoT devices has rapidly expanded. This chapter critically reviews recent literature, highlighting contributions, limitations, and how our proposed work addresses these gaps. Kumar et al. [ 13 ] conducted an extensive survey focusing on ARP spoofing vulnerabilities specifically within IoT networks. Their comprehensive analysis highlighted the fundamental vulnerabilities in ARP protocols due to lack of authentication mechanisms. Although thorough, their work did not propose a definitive solution, instead recommending the exploration of adaptive detection techniques. Ouaddah et al. [ 14 ] examined cryptographic defenses against ARP spoofing. They introduced methods leveraging cryptographic keys for authenticating ARP responses, significantly reducing attack vectors. However, their approach required substantial computational resources, making it unsuitable for resource-constrained IoT devices, thus highlighting the need for more lightweight alternatives. Sharma et al. [ 15 ] proposed supervised

Read the complete text