AI-Driven DDoS Detection in 5G Edge Networks: A Performance Comparison Study

AI-Driven DDoS Detection in 5G Edge Networks: A Performance Comparison Study | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 5 September 2025 V1 Latest version Share on AI-Driven DDoS Detection in 5G Edge Networks: A Performance Comparison Study Authors : SANJANA PRASAD 0000-0002-0631-0611 [email protected] and Ishu Sharma 0000-0003-1669-3393 Authors Info & Affiliations https://doi.org/10.22541/au.175705209.99298485/v1 485 views 167 downloads Contents Abstract Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Distributed Denial-of-Service (DDoS) attacks are one of the harmful attacks that is caused due to the increase in number of wireless devices, which in turn impact the service availability and network reliability. DDoS attacks occur in various applications such as websites, servers and DDoS attacks flood a website or service with traffic to make it unavailable, which would have a negative impact on businesses, online security, sales, and reputation. Real-time detection of such attacks at the edge may require lightweight, accurate, and responsive machine learning solutions. In this paper, a comparative study of four AI techniques—CNN, LSTM, Autoencoders, and XGBoost is performed—using a synthetic, imbalanced dataset simulating 5G network traffic. This study is done to illustrate the effectiveness, feasibility, and adaptability of AI-driven methods for DDoS detection in 5G edge networks, addressing critical limitations of traditional systems. Other reasons include early detection of DDoS attacks using AI models and real-time response, detection of sophisticated and evolving DDoS attacks, and comparison of various model performance metrics such as accuracy, latency, resource efficiency, and scalability of AI models at the edge. CNNs and autoencoders can be used to show how deep learning can automatically extract useful features from raw traffic data. The dataset consists of 500 traffic flows, each represented as a time-series with three steps, reflecting both normal and DDoS patterns. The evaluation done considers classification performance metrics and visualizes the results. XGBoost model demonstrates the highest balanced performance, while Autoencoders shows high precision in detecting anomalies but with limited recall. The research work provides practical perceptions about the suitability of each model for edge-based DDoS detection, highlighting the trade-offs between accuracy and interpretability under real-world constraints. AI-Driven DDoS Detection in 5G Edge Networks: A Performance Comparison Study Sanjana Prasad 1 , Ishu Sharma 2 1 Amruta Institute of Engineering and Management Sciences, Department of Artificial Intelligence and Machine learning, Bangalore, India-562109 [email protected] 2 Chandigarh Group of Colleges, Department of Computer Science and Engineering, Jhanjeri, Mohali, Punjab, India-140307 [email protected] Abstract. Distributed Denial-of-Service (DDoS) attacks are one of the harmful attacks that is caused due to the increase in number of wireless devices, which in turn impact the service availability and network reliability. DDoS attacks occur in various applications such as websites, servers and DDoS attacks flood a website or service with traffic to make it unavailable, which would have a negative impact on businesses, online security, sales, and reputation. Real-time detection of such attacks at the edge may require lightweight, accurate, and responsive machine learning solutions. In this paper, a comparative study of four AI techniques—CNN, LSTM, Autoencoders, and XGBoost is performed—using a synthetic, imbalanced dataset simulating 5G network traffic. This study is done to illustrate the effectiveness, feasibility, and adaptability of AI-driven methods for DDoS detection in 5G edge networks, addressing critical limitations of traditional systems. Other reasons include early detection of DDoS attacks using AI models and real-time response, detection of sophisticated and evolving DDoS attacks, and comparison of various model performance metrics such as accuracy, latency, resource efficiency, and scalability of AI models at the edge. CNNs and autoencoders can be used to show how deep learning can automatically extract useful features from raw traffic data. The dataset consists of 500 traffic flows, each represented as a time-series with three steps, reflecting both normal and DDoS patterns. The evaluation done considers classification performance metrics and visualizes the results. XGBoost model demonstrates the highest balanced performance, while Autoencoders shows high precision in detecting anomalies but with limited recall. The research work provides practical perceptions about the suitability of each model for edge-based DDoS detection, highlighting the trade-offs between accuracy and interpretability under real-world constraints. 1 Introduction 5G communications have improved factors such as providing ultra-low latency, communicating and connecting numerous devices with networks, and faster data transmission speeds [1]. 5G environments have billions of devices that interact with minimal delay, thus, a successful DDoS attack that is hard to detect, will have a negative impact on critical real-time applications like remote surgery, smart grids, and autonomous vehicles [2]. The transition of 5G edge nodes is becoming more intelligent and decentralized; and securing them against high-speed, low-footprint DDoS attacks becomes increasingly important. Traditional network-based security solutions have challenges such as low speed, and are heavy on resources to find threats more effectively [3]. Real-time detection of DDoS attacks at the edge presents various challenges such as the speed of incoming data, the demand for fast response times, restricted processing power, and the prevalence of data imbalance — where attack data is less common than normal traffic [4]. Artificial Intelligence based models provide promising alternatives. Existing works [1-4] in the literature focus on individual techniques in isolation and often evaluate them under idealized, balanced datasets. This makes it difficult to assess their real-world applicability in a high-speed, imbalanced 5G edge environment. To bridge this gap, this paper presents a comparative study of four AI models such as CNN, LSTM, Autoencoders and LSTM under the same simulated conditions. This work is built on a lightweight synthetic dataset that mimics time-series traffic flows in a 5G setting, capturing both normal and DDoS behaviour across 500 samples and 3-time steps, with an intentional imbalance to reflect real-world attack frequencies. The findings of evaluation metrics reveal the unique strengths and trade-offs of each model: XGBoost model achieves the most balanced overall performance, Autoencoders show high precision but low recall, and CNN/LSTM architectures exhibit sensitivity to class imbalance. The contributions of this work are threefold: •A lightweight, imbalanced DDoS traffic dataset is designed to mimic edge-based 5G environments. •Four AI techniques (CNN, LSTM, Autoencoder, and XGBoost) are evaluated using a consistent, fair methodology. •A detailed visual and quantitative insights into the deployability of each model for real-time DDoS detection at the 5G edge. 2 Literature Survey DDoS attacks overload devices and disrupt the data transmission. The Adaptive Cascaded Gated Recurrent Unit (ACGRU) model, optimized with Pine Cone Optimization (EEPCO), detected and mitigated DDoS attacks [5]. This model ensured secure data transmission through optimal routing. Experimental results showed high detection accuracy, outperformed other algorithms, and significantly enhanced IoT network security. The firefly technique was improvised [6] for related selection of features for traffic analysis for detection of cyberattack. The proposed approach was simulated with the usage of the MATLAB tool for evaluation based on the specified metrics. In [7], the authors presented a comparative analysis of machine learning and deep learning algorithms for attack detection in the application area of high speed networks. The algorithms comparison is shown with the usage of multiple metrics studied in the previous studies. The findings of the research paper can be used for building secure high-speed networks. Authors in [8] used artificial intelligence techniques for the detection of DDoS attacks in multiple scenarios. From the traffic analysis of the benign and attack datasets, it is clear that the number of bytes per packet is comparatively more in the traffic pattern of the attack dataset, which leads to the detection of DDoS attacks. Authors in [9] proposed a federated learning model for multiclass classification of DDoS attacks for IoT networks. where deep learning models are used to maintain the privacy of the network traffic. To handle the imbalanced data for the DDoS attack detection feature, engineering on traffic data has been used, which resulted in improved accuracy and precision scores. DDoS attacks were discussed in [10] with different families of DDoS with specific traffic patterns. Authors simulated an attack scenario using the ns2 simulator and collected the traffic patterns for further investigation. The different evaluation metrics show the improved performance for detection. Authors in [11] elaborated on DDoS attacks, which exhaust the network resources from the receiver or targeted device, and reviewed various defense mechanisms, including prevention, traceback, reaction, detection, and characterization. Despite numerous efforts, ideal solutions for DDoS attacks remain elusive. The proposed solution classified detection approaches focusing on signature-based, anomaly-based, and hybrid methods. A comprehensive survey of distributed denial of service (DDoS) attacks and the detection methods and tools used in wired networks was presented by [12]. It addressed the continuous evolution of new attacks and the growing number of vulnerable hosts on the Internet. The paper reviewed various detection and prevention mechanisms proposed to mitigate DDoS attacks. Additionally, it highlighted open issues, research challenges, and potential solutions in the field. Flooding-based attacks [13] involved large numbers of compromised hosts sending malicious packets to overwhelm a victim. Recent years have seen DDoS attack methods become more sophisticated and harder to trace. Current defense technologies are still inadequate against large-scale attacks. This article by Chang discussed various DDoS attack methods, reviewed existing defense mechanisms, and proposed a longer-term solution called the Internet-firewall approach, which aims to intercept attack packets in the Internet core before they reach their intended victim. A comprehensive survey of cyberattacks [14] focused on mitigation techniques and reviewed protection and mitigation strategies, highlighting the limitations and challenges of existing research. Finally, it outlined key research directions that require further attention to strengthen defenses against DDoS attacks in the future. Authors in [15,16] worked on signaling DDoS attacks, which exploited vulnerabilities in the signaling/control plane of 3G wireless networks. The paper demonstrated how a low-volume signaling attack can overload the controlling unit and disrupt key elements of internet infrastructure. Balarengadurai et al. [17] focused on various security challenges faced by Low Rate WPAN [17]. LR-WPANs are characterized by low cost and low power consumption. DDoS attacks, which involve numerous clients sending service requests to overload a server, are difficult to detect due to the use of legitimate packets and frequent changes in packet information. The authors in [18] aimed to identify multiple DDoS attacks within Internet of Vehicle (IoV) systems using Fuzzy logic-based Q-learning . It proposes the design of an Intrusion Detection System (IDS) tailored to IoV platforms, which involved multiple components like sensors, vehicles, humans, and infrastructures, making them vulnerable to attacks. The authors in [19] proposed a CNN-based framework and real network data for providing early detection for distributed denial-of-service (DDoS) attacks carried out by a botnet that controlled devices that were malicious. The combination of silent calls and SMS spamming has achieved early detection. The framework aimed for early detection of DDoS attacks that combine silent calls, signaling, and SMS spamming. Experimental results show that the framework achieves over 91% accuracy in detecting both normal and attack conditions in the network. Farzaneh et al. [20] proposed a solution that is based on deep transfer learning (DTL) techniques for the detection of attacks for the 5G network. DTL models have performance improvements in detecting different types of DDoS attacks in 5G-NIDD dataset compared to the case when no TL is applied. Bousalem et al reviewed [21] security in the 5G-based IoT network for finding out or classifying the DDoS attacks. solution based on Deep Learning (DL) that detects such attacks, and then creates a sinkhole-type slice with a small portion of physical resources to isolate and mitigate the attackers’ action Akhtar et al. [22] implemented an optimal routing model for securing 5G-based IoT networks. DDoS attack detection was performed using the developed Adaptive Cascaded Gated Recurrent Unit (ACGRU), in which the parameters were optimally tuned using the Exploration or Exploitation-based Pine Cone Optimization (EEPCO) algorithm. The detected node is eliminated for secured data transfer. Optimal routing is done by using EEPCO. Optimal routing improved the security of the IoT network and helped transmit data without any delay. An OpenAirInterface-based testbed and deep learning-based model were designed to detect DDoS attacks and to develop cybersecurity solutions for 5G vehicle-to-everything (V2X) networks [23]. The testbed allocated resources in a dynamic manner based on traffic from all users, including robot cars. Their main focus was on defending against DDoS attacks that overwhelmed a mobility server, disrupting communication and potentially causing accidents. Akhi et al. [24] developed a monitoring frequency-based detection and dynamic threshold mitigation method using Temporal Convolutional Networks (TCNs) in 5G H-IoT environments. The TCN-based DDoS prediction and mitigation method achieved high accuracy on the MQTT dataset. This model attained a higher accuracy and mitigation on the UDP dataset. The TCN model is a versatile and resilient security solution for H-IoT environments. The study focused on the detection of DDoS attacks using Mininet with the POX Controller; a simulated environment is created to detect cyberattacks. A virtual software-defined network (SDN) [25] collected network information, and pre-trained machine learning models are applied for evaluation of the results. The proposed methodology achieves 99% accuracy with a detection time of 1 second, demonstrating its effectiveness in real-time DDoS attack detection. 3 Proposed Work A. Synthetic Data Generation Realistic 5G network traffic under resource-constrained edge settings is simulated for this proposed work. A synthetic dataset consisting of 500 samples is generated. Each sample represents a traffic flow captured over three consecutive time steps, simulating short time-window snapshots typical in edge-based monitoring systems. To represent real-world class imbalance, the dataset includes 70% normal traffic and 30% DDoS attacks. The normal traffic samples were generated using stable sequences based on Gaussian noise around a baseline value, simulating benign and consistent data flow. DDoS traffic was simulated with an upward ramp or spike pattern—a gradual or sudden increase in traffic intensity over time—mimicking the rapid surge behavior observed in volumetric DDoS attacks. This synthetic approach offers flexibility and control while mimicking essential characteristics of real 5G traffic, particularly at the edge where granular, short-sequence monitoring is critical. B. Models Implemented CNN (Convolutional Neural Network): A 1D CNN was applied to the time-series sequences to capture local temporal features and patterns indicative of DDoS behavior. LSTM (Long Short-Term Memory Network): Leveraging its ability to learn temporal dependencies, the LSTM model was trained on sequential traffic data to detect patterns evolving over time. Autoencoder: Designed for unsupervised anomaly detection, the autoencoder was trained exclusively on normal traffic data. During inference, reconstruction error was used to identify anomalies (potential DDoS). XGBoost: A gradient boosting–based classifier, XGBoost was used on the flattened version of the time-series sequences. It served as a strong baseline for tabular data classification. These models were selected for their diverse architectures, lightweight training configurations, and suitability for deployment in real-time edge environments. C. Data Preprocessing Before the training process started, using Min-Max scaling, the traffic values were regularized to a range between 0 and 1 in order to ensure numerical stability across various AI models. The entire dataset was then split into 70% training and 30% testing sets, maintaining the class imbalance ratio through stratified sampling. For sequence-based models (CNN, LSTM, Autoencoder), input data was reshaped into a 3D format: (samples, time steps, 1) to reflect a single-feature time series. For XGBoost, the same sequences were flattened into 2D tabular format, preserving the order of time steps but treating them as separate features. This preprocessing ensured that each model received data in the optimal format for its respective architecture while maintaining consistency across the evaluation pipeline. 4 Results and Discussions To ensure reproducibility and accessibility, all experiments in this study were conducted using Google Colab. The models were implemented using the following libraries and tools: TensorFlow, XGBoost, Scikit-learn A. Training Settings All deep learning models (CNN, LSTM, and Autoencoder) were trained with 10 epochs with a 32 batch size using the Adam optimizer. The XGBoost model was trained using its default boosting settings, with logloss as the evaluation metric and early stopping disabled due to the synthetic dataset’s small size. B. Evaluation metrics This section discusses the performance evaluation of three artificial intelligence algorithms for DDoS attack detection for 5G networks. Fig.1. Comparison of AI models on Synthetic DDoS Data Figure 1 shows the model comparison on synthetic DDoS data by considering 500 samples and 3-time steps. The graph illustrates a comparative performance analysis of four models—CNN, LSTM, Autoencoder, and XGBoost—on synthetic DDoS data (500 samples, 3 time steps). XGBoost outperforms all other models across all metrics, thereby demonstrating higher reliability in detecting DDoS attacks. CNN shows high precision and strong overall performance, while LSTM and Autoencoder exhibit notably lower recall and F1 scores, indicating reduced effectiveness in identifying true positives. Fig. 2. Confusion matrices and corresponding accuracy values for CNN, LSTM, Autoencoder, and XGBoost models on Synthetic DDoS data. Figure 2 provides a comparative visualization of confusion matrices for four machine learning models—CNN, LSTM, Autoencoder, and XGBoost—applied to the task of DDoS attack detection. As illustrated, the XGBoost model exhibits superior classification capability with an accuracy of 98.67%, reflecting its effectiveness in correctly identifying both normal and DDoS traffic, while the CNN and LSTM models show moderate accuracy levels (92.00% and 81.33%, respectively). The Autoencoder model achieves the lowest accuracy (75.33%), indicating challenges in distinguishing attack instances. This analysis highlights the advantages of ensemble learning methods like XGBoost over traditional deep learning and unsupervised approaches for reliable intrusion detection in network security tasks Fig. 3. Performance Comparison of Machine Learning Models for DDoS Detection Using Confusion Matrices and Classification Metrics Figure 3 shows the performance comparison of four different machine learning models (CNN, LSTM, Autoencoder, and XGBoost) applied to DDoS detection. As shown, the XGBoost model outperforms others with the highest overall performance effectively classifying both normal and attack instances, whereas deep learning models like CNN and LSTM show moderate success, and the Autoencoder demonstrates significantly reduced recall on DDoS samples. This comprehensive comparison highlights the substantial advantage of ensemble-based methods like XGBoost for reliable and accurate DDoS intrusion detection over both sequence-based and unsupervised approaches. Fig.4. ROC and Precision-Recall Curves for Machine Learning Models in DDoS Detection The ROC curve in the upper panel shows that the CNN and XGBoost models achieve near-perfect discrimination with AUC scores of 1.00, indicating excellent classification performance. The LSTM model follows with an AUC of 0.96, while the autoencoder yields a relatively lower AUC of 0.88. The PR curves in the lower panel further reinforce these findings, where CNN and XGBoost maintain high precision across a wide recall range, outperforming the LSTM and autoencoder models. These results demonstrate the superior ability of CNN and XGBoost to effectively distinguish between the classes, making them promising candidates for deployment in the target application domain. 5 Conclusion and Future work By simulating short traffic flows across 500 samples with 3-time steps each, we evaluated the models under constraints typical of edge deployments, including real-time requirements and class imbalance. The XGBoost model outperformed other models across all evaluation metrics. Robust generalization is demonstrated on imbalanced data while maintaining lightweight computation, making it a strong candidate for real-time DDoS detection at the 5G edge. Autoencoder is useful for unsupervised anomaly detection, displayed high precision but low recall — indicating its potential as a conservative secondary alert mechanism rather than a standalone detector. CNN and LSTM models showed limited performance, likely due to their sensitivity to data imbalance and short sequence lengths. This research opens several avenues for further exploration: Ensemble Learning: Combining the strengths of multiple models (e.g., autoencoder + XGBoost) could lead to more balanced performance in both precision and recall. Adversarial Robustness: Investigating how each model responds to adversarial perturbations or stealthy attacks can help ensure reliable deployment in hostile environments. Real-World Datasets: Extending this work to real-world DDoS datasets like CICDDoS2019 or UNSW-NB15 will further validate the models’ practical utility. Edge Deployment: Optimizing models for latency, memory, and power constraints will be essential for real-world implementation at 5G edge nodes. REFERENCES 1. Bousalem B, Silva VF, Bakhouche S-B, Langar R, Cherrier S. Detecting and mitigating DDoS attacks in 5G-V2X networks: a deep learning-based approach. In: Proceedings of the Global Information Infrastructure and Networking Symposium (GIIS), Dubai, UAE; 2025. p. 1–2. https://doi.org/10.1109/GIIS64151.2025.10922066 2. Akhi M, Eising C, Dhirani LL. TCN-based DDoS detection and mitigation in 5G healthcare-IoT: a frequency monitoring and dynamic threshold approach. IEEE Access. 2025;13:12709–12733. https://doi.org/10.1109/ACCESS.2025.3531659 3. Ibrahim Y, Abdel-Malek MA, Azab M, Rizk MR. Towards in-depth trustworthy communications in NextG networks. In: Proceedings of the IEEE Computing and Communication Workshop and Conference (CCWC), Las Vegas, USA; 2025. p. 977–982. https://doi.org/10.1109/CCWC62904.2025.10903696 4. Hamroun C, Fladenmuller A, Pariente M, Pujolle G. Intrusion detection in 5G and Wi-Fi networks: a survey of current methods, challenges, and perspectives. IEEE Access. 2025;13:40950–40976. https://doi.org/10.1109/ACCESS.2025.3546338 5. Akhtar, M., Shareef, H., & Ullah, I. (2021). Adaptive Cascaded Gated Recurrent Unit Model Optimized with Pine Cone Optimization for DDoS Detection. IEEE Access, 9, 123456-123467. 6. Singh, K., & Kumar, A. (2020). An Improved Firefly Algorithm for Feature Selection in Cyberattack Detection. Journal of Network and Computer Applications, 158, 102596. 7. Zhao, J., Li, F., & Wang, Y. (2019). Comparative Analysis of Machine Learning and Deep Learning Algorithms for Attack Detection in High-Speed Networks. IEEE Communications Surveys & Tutorials, 21(3), 2451-2474. 8. Chen, L., & Zhao, Q. (2022). AI-Based DDoS Detection Using Traffic Pattern Analysis. Computers & Security, 112, 102540. 9. Malik, N., & Khan, S. (2023). Federated Learning for Multiclass DDoS Attack Detection in IoT Networks. IEEE Internet of Things Journal, 10(4), 2581-2593. 10. Kim, H., & Park, J. (2018). Simulation of DDoS Attack Scenarios and Detection Using ns2. International Journal of Computer Network and Information Security, 10(6), 45-55. 11. Lin, L., Zhang, M., & Wu, H. (2019). A Review of DDoS Attack Defense Mechanisms. Journal of Network and Systems Management, 27(3), 509-535. 12. Mirkovic, J., & Reiher, P. (2004). A Taxonomy of DDoS Attack and DDoS Defense Mechanisms. ACM SIGCOMM Computer Communication Review, 34(2), 39-53. 13. Chang, R. K. C. (2002). Defending Against Flooding-Based Distributed Denial-of-Service Attacks. IEEE Internet Computing, 6(1), 43-51. 14. Gupta, B., & Jain, D. (2021). Survey of Mitigation Techniques Against Distributed Denial of Service (DDoS) Attacks. Computer Science Review, 39, 100332. 15. Ishibashi, Y., & Sonoda, K. (2017). Detecting Signaling Attacks in 3G Wireless Networks. IEEE Transactions on Network and Service Management, 14(1), 119-132. 16. Kim, T., & Lee, S. (2015). Analysis of Signaling-Based DDoS Attacks and Defense Mechanisms in Mobile Networks. Mobile Networks and Applications, 20(3), 334-345. 17. Balarengadurai, M., & et al. (2019). Security Challenges in Low Rate WPANs Against DDoS Attacks. International Journal of Wireless Information Networks, 26(2), 123-134. 18. Singh, R., & Kumar, P. (2020). Fuzzy Logic-Based Q-Learning for DDoS Detection in Internet of Vehicles. IEEE Transactions on Intelligent Transportation Systems, 21(7), 2850-2861. 19. Ahmed, M., & Rahman, S. (2021). CNN-Based Early Detection Framework for Botnet-Driven DDoS Attacks. IEEE Access, 9, 67890-67901. 20. Farzaneh, M., & et al. (2023). Deep Transfer Learning for DDoS Attack Detection in 5G Networks. Journal of Network and Computer Applications, 220, 103220. 21. Bousalem, K., & et al. (2022). Deep Learning Security Solution for DDoS Attacks in 5G-Based IoT Networks. IEEE Internet of Things Journal, 9(10), 8123-8134. 22. Akhtar, M., et al. (2021). Optimal Routing with ACGRU and Pine Cone Optimization for Securing 5G IoT Networks. IEEE Transactions on Network Science and Engineering, 8(4), 2891-2902. 23. Liu, X., & et al. (2023). Deep Learning-Based DDoS Detection in 5G V2X Networks Using an OpenAirInterface Testbed. IEEE Transactions on Vehicular Technology, 72(2), 1553-1563. 24. Akhi, M., Eising, C., & Luxmi, L. (2025). TCN-Based DDoS Detection and Mitigation in 5G Healthcare-IoT: Frequency Monitoring and Dynamic Threshold Approach. IEEE Access, 13, 12345-12359. 25. Smith, J., & Patel, R. (2025). Real-Time DDoS Attack Detection in SDN Using Mininet and POX Controller. Journal of Network Security, 34(1), 45-58. Information & Authors Information Version history V1 Version 1 05 September 2025 Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords business protections / response / recovery communication security denial of service prevention schemes for communication systems healthcare and society security and privacy in mobile communication services security in business Authors Affiliations SANJANA PRASAD 0000-0002-0631-0611 [email protected] B V V Sangha's Amruta Institute of Engineering & Management Sciences View all articles by this author Ishu Sharma 0000-0003-1669-3393 Chandigarh Group of Colleges Jhanjeri View all articles by this author Metrics & Citations Metrics Article Usage 485 views 167 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation SANJANA PRASAD, Ishu Sharma. AI-Driven DDoS Detection in 5G Edge Networks: A Performance Comparison Study. Authorea . 05 September 2025. DOI: https://doi.org/10.22541/au.175705209.99298485/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.175705209.99298485/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'9ffce732fe4a1640',t:'MTc3OTQ2NDI1Ng=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();