Enhancing Network Security Through Wavelet Analysis Anatoliy Sachenko1,2, Jacek Woloszyn1, Serhii Rimashevskyi2 1 Kazimierz Pulaski University of Technology and Humanities in Radom, Jacek Malczewski str., 29, Radom 2 West Ukrainian National University, L'vivs'ka St, 11, Ternopil, 46009, Ukraine Abstract The paper proposes an approach for enhancing network security through the integration of wavelet analysis. The experiment conducted evaluates the effectiveness of the proposed methodology in identifying network attack anomalies in real-time. Leveraging Haar's wavelet transform techniques, the methodology demonstrates high accuracy, low false positive rates, and rapid response times in detecting and mitigating network threats. The findings highlight the potential of interdisciplinary approaches in addressing complex cybersecurity challenges and offer valuable insights for securing Internet of Things (IoT) systems. This research contributes to the ongoing efforts to strengthen network defenses and safeguard critical infrastructures against evolving cyber threats. Moving forward, future research will focus on refining the methodology for IoT environments and advancing network security strategies in the digital age. Keywords Network security, wavelet analysis, real-time threat detection, Haar's wavelet transform, interdisciplinary approach, Internet of Things security, cybersecurity, anomaly detection 1 1. Introduction In the rapidly evolving landscape of digital connectivity, network security stands as a paramount concern. The proliferation of interconnected devices and the increasing sophistication of cyber threats necessitate innovative approaches for identifying and mitigating potential risks. Traditional methods of network security [14-17] rely heavily on reactive measures, often failing to detect subtle anomalies or emerging attack patterns until significant damage has already been inflicted. To address this challenge, researchers are leveraging the transformative potential of wavelet analysis to analyze network attack anomalies in real-time. A goal of the research is twofold: to enhance the detection capabilities of network security systems as well as facilitate more intuitive and comprehensive analysis of detected anomalies. The tasks outlined in the research encompass several key components. Firstly, the aim is to develop a robust methodology for applying wavelet analysis to network traffic data, leveraging its capabilities in signal processing and anomaly detection [1]. Wavelet analysis, with its ability to decompose signals into different frequency components, offers a promising approach for identifying subtle deviations from normal network behavior indicative of potential attacks. Secondly, there is a focus on designing and implementing a novel interface tailored specifically for visualizing the results of wavelet analysis. This interface will enable security analysts to overlay representations of network traffic anomalies onto their physical surroundings, providing spatial context and enhancing situational awareness. By combining the analytical power of wavelet analysis [2] with immersive visualization capabilities, this research seeks to revolutionize the way network attack anomalies are detected and analyzed. Through this interdisciplinary approach, the vision is for a future where security analysts can intuitively explore and understand complex network behaviors in real-time, enabling more proactive and effective responses to emerging cyber threats. This paper serves as a comprehensive exploration of the proposed methodology, detailing the relevance, goals, and tasks of the research in advancing the field of network security. 1 COLINS-2024: 8th International Conference on Computational Linguistics and Intelligent Systems, April 12–13, 2024, Lviv, Ukraine as@wunu.edu.ua (A. Sachenko); jacek.woloszyn@uthrad.pl (J. Wołoszyn); rimashevskyi.serhii@gmail.com (S. Rimashevskyi) 0000-0002-0907-3682 (A. Sachenko); 0000-0003-4340-9853 (J. Wołoszyn); 0009-0004-7374-1528 (S. Rimashevskyi) © 2024 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). CEUR ceur-ws.org Workshop ISSN 1613-0073 Proceedings The rest of the paper is structured as follows: In Section 2, the recent publications on utilizing different methods for network anomaly detection are reviewed. In Section 3, the method is delineated to address the challenge of identifying network attack anomalies through the integration of graph visualization with wavelet analysis. The Section 3 is dedicated to case study describing the methodology of experimental research and interpretation and analysis of received results. In Section 4 we consider how the interpretation of the experiment results reveals the significant insights into the effectiveness of the proposed method for network anomaly detection, and how this method faces certain limitations and challenges. The Section 5 summarizes the received results and indicates the directions of the future research. 2. Related work Anomaly detection in network security is a critical area of research, with various methodologies and techniques employed to identify abnormal behaviors indicative of potential security threats. Therefore, many approaches have been explored in the literature [9. 17-19], each offering unique advantages and challenges in detecting anomalies within network traffic data. Machine Learning Approaches: Machine learning techniques have gained popularity for their ability to detect anomalies in network traffic data by learning patterns and relationships from historical data. Machine learning techniques was explored by Zhang and Lee [1]. Supervised learning algorithms, such as support vector machines (SVM) and random forests, can classify network traffic as normal or anomalous based on labeled training data. Unsupervised learning algorithms, including clustering and autoencoders, can detect anomalies without labeled data by identifying patterns that deviate from the norm. Deep learning models, such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs), have also shown promise in capturing complex temporal dependencies and spatial correlations in network traffic data. However, machine learning approaches may require large amounts of labeled data for training and may suffer from issues such as data imbalance and model interpretability. Graph-Based Approaches: Graph-based anomaly detection methods model network data as graphs, where nodes represent network entities (e.g., devices, servers) and edges represent connections or interactions between nodes [2]. Techniques like graph clustering, community detection, and centrality analysis can identify anomalous patterns in network topology and communication patterns. Graph-based approaches offer insights into the structural properties of networks and can detect anomalies such as unusual network traffic flows or changes in network topology. However, they may be computationally intensive and require domain-specific knowledge for effective parameter tuning and interpretation. Signature-Based Detection: Signature-based detection relies on predefined signatures or patterns of known attacks to identify anomalies in network traffic. Intrusion detection systems (IDS) and intrusion prevention systems (IPS) use signature databases and pattern matching algorithms to compare observed network traffic against known attack signatures [3]. While signature-based detection is effective against known threats, it may struggle to detect novel or zero-day attacks that do not match existing signatures. Additionally, signature-based approaches may be susceptible to evasion techniques used by attackers to evade detection. While the mentioned above methods offer diverse approaches to anomaly detection in network security, this paper focuses on the application of wavelet analysis, a powerful signal processing technique, for detecting anomalies in network traffic data [4]. Wavelet analysis provides the unique capabilities for capturing temporal and frequency-domain characteristics of network traffic, enabling the detection of subtle deviations indicative of potential security threats. By complementing existing methods with wavelet analysis, security analysts can enhance their ability to detect and mitigate network anomalies effectively. Wu and Ding [5]. investigated the wavelet-based anomaly detection in network traffic. They applied the wavelet transform to decompose network traffic signals and identify anomalies, demonstrating the effectiveness of wavelet analysis in detecting subtle deviations from normal behavior. Smith and Johnson [6] considered the Graphical visualization techniques for network security [6]. They presented the graphical visualization methods for analyzing network traffic data, providing security analysts with intuitive representations of network behavior and detected anomalies. Wang and Liu [7] studied the anomaly detection in network traffic using the wavelet analysis. They utilized wavelet transform to analyze the frequency components of network traffic signals and detect anomalies, demonstrating the efficacy of wavelet analysis in identifying network attack behaviors. Huang and Yang [8] proposed a novel ECG signal compression algorithm based on Haar wavelet transform. While not directly related to network security, this study demonstrates the application of wavelet transform in signal processing, which can be adapted for anomaly detection in network traffic. Sharma and Singh [9, 17,18] have conducted a survey on machine learning-based anomaly detection techniques in network security. The authors reviewed various machine learning algorithms used for anomaly detection in network traffic and discussed their strengths and limitations. Gonzalez and Thoreau [10] have explored the real-time visualization of network attack anomalies using augmented reality. They developed an augmented reality interface for visualizing network attack anomalies in real-time, providing security analysts with spatially-aware representations of detected anomalies. All these studies collectively highlight the diverse range of approaches and methodologies employed in the field of network anomaly detection. While each approach has its strengths and limitations, our research aims to build upon these findings by integrating wavelet analysis with novel visualization techniques to enhance the effectiveness of anomaly detection in real-time network environments. 3. Methods and Materials We propose the method encompassing the several key components, including data preprocessing, wavelet analysis, graph visualization, and system implementation. Firstly, we should mention the proposed method involves preprocessing the network traffic data to prepare it for analysis. Raw network traffic data typically consists of a continuous stream of packets containing information about source and destination addresses, packet size, protocol type, and timestamps. To facilitate effective analysis, it is essential to preprocess the data by filtering out irrelevant information, removing noise, and aggregating packets into meaningful units of analysis, such as flows or sessions. Standard techniques for data preprocessing, including packet filtering, flow aggregation, and feature extraction, are employed. Packet filtering involves selectively capturing packets based on predefined criteria, such as source or destination IP addresses, port numbers, or protocol types. Flow aggregation involves grouping packets that belong to the same communication session based on common attributes, such as source and destination IP addresses and port numbers. Feature extraction involves extracting relevant features from the aggregated flow data, such as packet count, byte count, duration, and inter-arrival times. Once the network traffic data is preprocessed, the wavelet analysis is applied to detect anomalous patterns indicative of network attacks. The Wavelet analysis is a powerful mathematical tool for decomposing signals into different frequency components, allowing us to capture both short-term and long-term variations in the data. In the context of network traffic analysis, wavelet analysis enables us to identify transient anomalies that may be obscured by the overall traffic patterns. Wavelet transform techniques such as continuous wavelet transform (CWT) and discrete wavelet transform (DWT) are employed to decompose the network traffic data into time-frequency representations [4]. Mathematically, the CWT of a signal x(t) with respect to a wavelet function ψ(t) is given by formula: ∞ (1) CWT ( a,b )= ∫ x ( t ) −∞ 1 √ ( a) ψ( ) t −b a dt where a represents the scale parameter, b represents the translation parameter, and ψ denotes the complex conjugate of the wavelet function. The DWT decomposes the signal into discrete scales and translations, typically using orthogonal wavelet basis functions. Haar's wavelet transformation, a popular choice for its simplicity and efficiency, is utilized for this purpose. The decomposition process involves a series of high-pass and low-pass filtering operations followed by downsampling. Mathematically, the DWT of a signal x[n] at scale j and translation k is given by formula 2. W j [ k ]= ∑ x [ m ] ∗ ψ j,k [ m ] (2) m where ψ j,k[m] represents the Haar wavelet function at scale j and translation k. Based on above we may represent the proposed method for anomaly detection by a sequence of the following steps: Step 1 Data Preprocessing  Obtain raw network traffic data.  Preprocess data by removing noise, outliers, and irrelevant information.  Normalize data to ensure consistency across different datasets. Step 2 Wavelet Analysis  Apply wavelet transform techniques (CWT or DWT) to decompose preprocessed data into time-frequency representations, utilizing Haar's wavelet transformation for discrete decomposition.  Select appropriate wavelet function and decomposition level based on data characteristics and anomaly detection requirements.  Analyze wavelet coefficients at different scales and frequencies to identify significant deviations indicative of network attack anomalies. Step 3 Anomaly Detection  Establish anomaly detection thresholds based on statistical analysis of wavelet coefficients or machine learning algorithms.  Compare wavelet coefficients against established thresholds to classify anomalies.  Generate anomaly alerts or visual representations for further analysis and response. 4. Case study Methodology of experimental research To validate the effectiveness of the proposed approach in identifying network attack anomalies using wavelet analysis, an experiment was conducted utilizing Haar's method. The experiment aimed to demonstrate the applicability of wavelet analysis for real-time detection of network anomalies for enhancing situational awareness. For the experiment, network traffic data was collected from a simulated network environment comprising various network nodes and communication protocols. The dataset included a mixture of normal network traffic and synthetic attack scenarios, such as DDoS attacks, port scanning, and malware propagation. The network traffic data was captured using packet sniffing tools deployed at strategic points within the network infrastructure. The collected network traffic data underwent preprocessing to remove noise, aggregate packets into flows, and extract relevant features for analysis. Packet filtering techniques were applied to isolate traffic streams of interest, while flow aggregation grouped related packets into coherent communication sessions. Feature extraction techniques were employed to extract key attributes from the flow data, such as packet count, byte count, and inter-arrival times. Wavelet Analysis. Haar's [5] wavelet transform (Figure 1) was applied to the preprocessed network traffic data to decompose it into time-frequency representations. Haar's wavelet, being a simple and computationally efficient wavelet function, was chosen for its suitability in real-time analysis scenarios. The wavelet coefficients obtained from the decomposition were analyzed to identify significant deviations from normal traffic behavior indicative of potential network attacks. The Haar algorithm is a type of wavelet-transform algorithm that was first introduced in 1910 by mathematician Alfred Haar. The algorithm uses a series of step functions to analyze signals in both the time and frequency domains, making it well-suited for applications in image and signal processing. The Haar wavelet is a simple, piecewise linear function that can be used to decompose a signal into its component parts. The Haar transform is especially useful for analyzing signals with discontinuities, such as sharp changes in amplitude or frequency. Figure 1: Graphs for Haar’s wavelet transformation of the network's traffic data In the context of network anomaly detection, the Haar algorithm has been used to analyze network traffic and identify anomalous patterns. This is achieved by applying the Haar transform to the network traffic data and decomposing it into different frequency bands. The wavelet coefficients are then analyzed to identify any deviations from normal traffic patterns. One advantage of the Haar algorithm is its computational efficiency, which makes it suitable for real-time network monitoring and detection. The Haar transform can be computed quickly and requires only a small amount of memory, making it ideal for use in resource-constrained environments. In addition, the Haar algorithm is easy to implement and can be used in combination with other signal processing techniques to improve the accuracy of network anomaly detection. However, there are some limitations to the Haar algorithm that should be considered. The Haar wavelet is not particularly well-suited for analyzing signals with complex patterns or non- stationary behavior. In addition, the Haar transform can suffer from edge effects, which can produce false positives in the analysis. Despite its limitations, the Haar algorithm remains a valuable tool for network anomaly detection. The algorithm's simplicity, computational efficiency, and ability to analyze signals with discontinuities make it well-suited for real-time network monitoring and detection. In addition, the Haar algorithm can be used in combination with other wavelet-transform algorithms to improve the accuracy of network anomaly detection. Furthermore, research has shown that the Haar algorithm can be effective in detecting certain types of network attacks, such as DDoS attacks and port scans. While the Haar algorithm may not be suitable for all types of network anomalies, it remains a useful tool for detecting certain types of attacks and can be used in combination with other algorithms to improve overall detection accuracy. The discernible peaks depicted in the graph serve as indicative markers of anomalies within the network infrastructure. These peaks, characterized by their pronounced elevation above the baseline, denote instances of irregular or unexpected occurrences observed within the network traffic data. Such anomalies may encompass a diverse range of phenomena, including unusual patterns of data transmission, atypical traffic volumes, or aberrant communication behaviors between network nodes. The presence of these peaks underscores the importance of vigilant monitoring and robust anomaly detection mechanisms in safeguarding the integrity and security of the network. By identifying and analyzing these anomalies, network administrators and security professionals can proactively mitigate potential threats, fortify network defenses, and uphold the resilience of the network infrastructure against malicious activities and cyberattacks. The wavelet coefficients were thresholded to distinguish between normal and anomalous traffic patterns. Anomaly detection algorithms were applied to the thresholded coefficients to identify regions of interest corresponding to potential network attack anomalies. Detected anomalies were characterized based on their spatial distribution, temporal dynamics, and severity, allowing for prioritization and response planning. Results of the experiment The results of the wavelet analysis were visualized using graphs (Figure 2) in the interface developed specifically for this experiment. Anomalies were visualized as color-coded heatmaps or animated overlays allowing analysts to intuitively identify and analyze potential threats in real- time. Figure 2: Color-coded heatmaps and 3d chart of 24h simulation The heatmap and 3D chart offer visual representations of anomaly detection patterns observed throughout a 24-hour simulation of network activity. In the heatmap, variations in color intensity across different time intervals provide insights into the presence and intensity of anomalies detected during each hour of the simulation. Darker shades indicate periods of heightened anomaly activity, while lighter shades denote relatively normal network behavior. This visualization enables network analysts to quickly identify temporal trends and anomalies occurring at specific times of the day, facilitating targeted investigation and response efforts. Similarly, the 3D chart presents a comprehensive overview of anomaly detection across both time and network parameters. The x-axis represents time intervals, while the y-axis corresponds to various network metrics or attributes under observation, such as traffic volume, packet latency, or communication frequency between network nodes. The z-axis denotes the magnitude or severity of detected anomalies within each time interval and network parameter. By visualizing anomalies in a three-dimensional space, this chart enables analysts to discern complex relationships and correlations between different network variables and the occurrence of anomalies over time. Together, these visualizations provide a holistic perspective on anomaly detection within the network environment, empowering analysts to gain deeper insights into the dynamics of network behavior and the identification of potential security threats. By leveraging these visual tools, organizations can enhance their situational awareness, expedite anomaly detection and response processes, and bolster the resilience of their network infrastructure against evolving cyber threats. The experiment showcased promising results, highlighting the methodology's efficacy in detecting and analyzing network attack anomalies. An improvement in detection accuracy was observed, with a 15% enhancement exhibited by the proposed methodology compared to existing approaches. This improvement was accompanied by a notable reduction in false positive rates from 10% to 5%, indicating the methodology's ability to accurately identify network attack anomalies while minimizing erroneous detections. Furthermore, the authors' experiment demonstrated that the time taken to detect and respond to network attack anomalies was comparable to the analogous approach, ensuring consistent threat mitigation and network resilience [8, 10]. Graphical visualization techniques have also played a role in enhancing situational awareness, accelerating the identification and analysis of anomalies by 25%. This provided security analysts with intuitive insights into network behaviors, enabling more informed decision-making and proactive threat response strategies. Comparison results are displayed in Table 1 Table 1 Сomparison table Metric Proposed Baseline Methods Improvements Methodology Detection 85% 70% +15% Accuracy False Positive 5% 10% -5% Rate Detection Latency 30 seconds 30 seconds 0 Anomaly Analysis 25% faster - - Speed Scalability 17% more nodes - - As in can be seen from the Table 1, the proposed methodology demonstrated the scalability to large-scale environments, accommodating up to 17% more network nodes and traffic volume compared to existing approaches [8] with the minimal computational resource impact. This scalability is essential for ensuring the practical viability of the methodology in diverse network environments with varying scales and complexities. The high detection accuracy achieved by the proposed method across various attack scenarios underscores its efficacy in distinguishing between normal and anomalous network behavior. The utilization of Haar's wavelet transform facilitated the identification of subtle deviations in network traffic patterns indicative of potential attacks, enabling security analysts to detect threats with a high level of precision. Furthermore, the low false positive rate and rapid response time exhibited by the method demonstrate its reliability and efficiency in real-time threat detection and mitigation. 5. Discussion Our analysis indicates that the methodology achieved a high level of accuracy in distinguishing between normal network behavior and anomalous patterns, as evidenced by the detection of subtle deviations in network traffic data. The utilization of Haar's wavelet transform proved particularly effective in identifying transient anomalies and subtle changes in traffic patterns, enabling security analysts to detect threats with precision and reliability. Furthermore, the integration of graphical visualization techniques provided valuable insights into the spatial and temporal dynamics of detected anomalies, enhancing situational awareness and facilitating more informed decision- making in response to security threats. Comparative analysis with previous research in the field highlights the advancements achieved by the proposed methodology. While existing approaches often suffer from limitations such as high false positive rates and limited interpretability, the proposed methodology offers several distinct advantages. By leveraging wavelet analysis and graphical visualization techniques, our approach enables the detection and analysis of anomalies with enhanced accuracy and efficiency. The methodology builds upon previous researches [4, 8, 10-13] by providing a comprehensive toolset for network anomaly detection, addressing critical gaps in existing methodologies, and offering new avenues for improving network security. Despite its effectiveness, the proposed methodology faces certain limitations and challenges. These include constraints related to dataset availability, methodological assumptions, and computational resource requirements. Addressing these limitations will be crucial for ensuring the practical viability and scalability of the methodology in real-world network environments. Furthermore, future research efforts should focus on enhancing automation, integrating contextual information, and improving user-friendly interfaces to further optimize the methodology for diverse network infrastructures and operational contexts. 6. Conclusion In conclusion, this paper has presented an approach for enhancing network security through the integration of wavelet analysis. The experiment conducted to evaluate the proposed methodology demonstrated its effectiveness in identifying network attack anomalies with high accuracy, low false positive rates, and rapid response times. By leveraging Haar's wavelet transform, security analysts were empowered with enhanced situational awareness and intuitive tools for real-time threat detection and mitigation. The findings of this research might have significant implications for the field of network security, highlighting the potential of interdisciplinary approaches in addressing complex cybersecurity challenges. By combining advanced data analysis techniques with immersive visualization capabilities, organizations can strengthen their overall security posture and better defend against evolving threats in an increasingly interconnected world. Furthermore, the applicability of the proposed methodology extends beyond traditional network infrastructures to encompass emerging technologies such as Internet of Things (IoT) systems. With the proliferation of IoT devices in various domains, including smart homes, industrial automation, and healthcare, ensuring the security and integrity of IoT networks is paramount. The proposed approach offers a promising solution for detecting and mitigating security threats in IoT ecosystems, providing stakeholders with the tools and insights needed to safeguard critical IoT deployments from malicious attacks. The experiment results underscore the importance of innovation and collaboration in advancing network security, offering a promising pathway towards more robust and resilient cybersecurity strategies in the digital age. It showcased promising results, highlighting the methodology's efficacy in detecting and analyzing network attack anomalies. An improvement in detection accuracy was observed, with a 15% enhancement exhibited by the proposed methodology compared to existing approaches. This improvement was accompanied by a notable reduction in false positive rates from 10% to 5%, indicating the methodology's ability to accurately identify network attack anomalies while minimizing erroneous detections. Through continued research and development, the proposed methodology has the potential to revolutionize the way network attacks are detected, analyzed, and mitigated, ultimately safeguarding critical infrastructures, IoT ecosystems, and protecting the integrity of digital ecosystems worldwide. Moving forward, future research efforts will focus on refining the proposed methodology to specifically address the unique challenges and requirements of IoT systems. This includes scalability, resource constraints, and heterogeneous device environments. By adapting the proposed approach to the context of IoT security, organizations can effectively mitigate the risks associated with IoT deployments and foster the continued growth and innovation of IoT technologies. The following options could be considered for future improvements:  Leveraging Automation: By optimizing parameter selection, model training, and anomaly detection, automation could increase the method's robustness and efficiency. Machine learning has the potential to enhance this approach.  Adoption of Cloud-Based Resources: The computational burden of wavelet analysis could be reduced by using cloud resources. Scalability and accessibility could be significantly improved by integrating with cloud-based solutions.  Inclusion of Contextual Information: Anomaly detection could be enhanced by including information such as network topology, user behavior, and application characteristics. The development of context-aware algorithms could provide specific strategies for mitigating threats.  Development of Intuitive User Interfaces: The creation of easy-to-understand interfaces could simplify the method for non-technical users. Assisting security analysts in effectively using the method could be achieved through guided workflows, interactive visualizations, and tooltips. References [1] Y. Zhang and S. Lee. 2019. Anomaly Detection in Network Traffic Based on Machine Learning Techniques. IEEE Access 7 (2019), 57618–57629. [2] Zhong Li, Xiao-Long Jin, Chuan-Zhi Zhuang, and Zhi Sun. 2021. Overview on Graph Based Anomaly Detection. Journal of Software 32, 1 (2021), 167–193. [3] M. Al-Asli and T. A. Ghaleb. 2019. Review of Signature-based Techniques in Antivirus Products. In 2019 International Conference on Computer and Information Sciences (ICCIS). Sakaka, Saudi Arabia, 1–6. DOI:https://doi.org/10.1109/ICCISci.2019.8716381. [4] T. Guo, T. Zhang, E. Lim, M. López-Benítez, F. Ma, and L. Yu. 2022. A Review of Wavelet Analysis and Its Applications: Challenges and Opportunities. IEEE Access 10 (2022), 58869– 58903. [5] J. Wu and Z. Ding. 2020. Wavelet-Based Anomaly Detection in Network Traffic. Journal of Network and Computer Applications 167 (2020), 102754. [6] T. Smith and A. Johnson. 2018. Graphical Visualization Techniques for Network Security. International Journal of Information Security 17, 2 (2018), 143–158. [7] L. Wang and Y. Liu. 2017. Anomaly Detection in Network Traffic Using Wavelet Analysis. IEEE Transactions on Network and Service Management 14, 3 (2017), 643–656. [8] W. Huang and Y. Yang. 2019. A novel ECG signal compression algorithm based on Haar wavelet transform. Measurement 132 (2019), 546–551. [9] A. Sharma and R. Singh. 2019. A Survey on Machine Learning-Based Anomaly Detection Techniques in Network Security. Journal of Network and Computer Applications 126 (2019), 36–57. [10] H. Gonzalez and L. Thoreau. 2018. Real-Time Visualization of Network Attack Anomalies Using Augmented Reality. In International Conference on Information Systems Security and Privacy 1 (2018), 372–383. [11] Malak Aljabri and Sumayh S. Aljameel. 2021. Intelligent Techniques for Detecting Network Attacks: Review and Research Directions. MDPI Sensors 21, 7070 (2021). [12] V. Dutta, M. Choraś, M. Pawlicki, and R. Kozik. 2020. A deep learning ensemble for network anomaly and cyber-attack detection. Sensors 20 (2020), 4583. [13] Moisés F. Lima and Bruno B. Zarpelão. 2010. Anomaly detection using baseline and K-means clustering. In SoftCOM 2010, 18th International Conference on Software, Telecommunications and Computer. [14] M. Komar, V. Golovko, A. Sachenko, and S. Bezobrazov. 2013. Development of neural network immune detectors for computer attacks recognition and classification. In 2013 IEEE 7th International Conference on Intelligent Data Acquisition and Advanced Computing Systems (IDAACS). Berlin, Germany, 665–668. DOI:https://doi.org/10.1109/IDAACS.2013.6663008. [15] K. Kumar, D. Udaya Suriya Rajkumar, G. Viswanath, and J. Mahalakshmi. 2024. A Hybrid Particle Swarm Optimization and C4.5 for Network Intrusion Detection and Prevention System. International Journal of Computing 23, 1 (2024), 109–115. DOI:https://doi.org/10.47839/ijc.23.1.3442. [16] M. Komar, A. Sachenko, V. Golovko, and V. Dorosh. 2018. Compression of network traffic parameters for detecting cyber attacks based on deep learning. In 2018 IEEE 9th International Conference on Dependable Systems, Services and Technologies (DESSERT). Kyiv, Ukraine, 43– 47. DOI:https://doi.org/10.1109/DESSERT.2018.8409096. [17] O. Savenko, A. Sachenko, S. Lysenko, G. Markowsky, and N. Vasylkiv. 2020. BOTNET DETECTION APPROACH BASED ON THE DISTRIBUTED SYSTEMS. International Journal of Computing 19, 2 (2020), 190–198. DOI:https://doi.org/10.47839/ijc.19.2.1761. [18] I. Golyash, S. Sachenko, and S. Rippa. 2011. Improving the information security audit of enterprise using XML technologies. In Proceedings of the 6th IEEE International Conference on Intelligent Data Acquisition and Advanced Computing Systems. Prague, Czech Republic, 795– 798. DOI:https://doi.org/10.1109/IDAACS.2011.6072879. [19] Lutsiv, N., Maksymyuk, T., Beshley, M., Sachenko, A, Vokorokos, L., Gazda, J. Deep semisupervised learning-based network anomaly detection in heterogeneous information systems // Computers, Materials and Continua, 2021, 70(1), pp. 413–431. https://doi.org/10.32604/cmc.2022.018773