A SYSTEMATIC REVIEW ON ENHANCING IOT SECURITY WITH DEEP LEARNING AND BIG DATA ANALYTICS

Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4565 A SYSTEMATIC REVIEW ON ENHANCING IOT SECURITY WITH DEEP LEARNING AND BIG DATA ANALYTICS ANKUR GUPTA1,*, DR. DINESH CHANDRA MISRA2 1PhD Scholar, Department of Computer Science and Engineering, Dr. K.N. Modi University, Newai, Rajasthan, India 2Associate Professor, Department of Computer Science and Engineering, Dr. K.N. Modi University, Newai, Rajasthan, India E-mail: [email protected], [email protected] *Corresponding Author: ANKUR GUPTA (ankurdujan[email protected]m) ABSTRACT The exponential expansion of the Internet of Things (IoT) has altered connection and automation in many industries. The vast amount of data produced and the restricted processing power of various IoT devices have caused major security issues, however, as the Internet of Things (IoT) has grown quickly. This paper offers a thorough examination and analysis of current studies combining deep learning (DL) and big data analytics to improve IoT security. Among the many notable contributions examined were blockchainintegrated security systems, hybrid DL models combining CNNs and RNNs with big data analytics, and anomaly detection in industrial IoT using autoencoders and LSTM. The evaluation also takes into account federated learning strategies meant to provide privacy-preserving security in highly dispersed IoT networks. Though the accuracy—often over 90% in threat detection—is remarkable, other research points out drawbacks include the focused attention on certain frameworks, lack of generalizability across IoT sectors, and difficulties in using hybrid or federated models. This paper underlines the changing function of integrating deep learning with big data analytics by means of insights from 15 major research and discusses the future promise of these technologies in creating safe, scalable, and smart IoT systems. Authors have utilized a hybrid deep learning approach combining Convolutional Neural Networks (CNNs) and autoencoders for anomaly detection within general IoT environments, achieving a high detection accuracy of 92.3%. Some of the author focused specifically on Industrial IoT (IIoT), employing a combination of Long Short-Term Memory (LSTM) networks and autoencoders. This approach yielded an even higher detection accuracy of 94.1%. Meanwhile, many researchers proposed a privacy-preserving model using federated learning, achieving an estimated detection accuracy of around 90%. Keywords: Big Data, CNN, Deep Learning, IoT, LSTM, Security. 1. INTRODUCTION By allowing connectivity of many kinds of physical devices, sensors, and networks to create smart systems that can receive, analyze, and exchange data, IoT is transforming the way many companies operate. Automation of processes, resource optimization, and improved decisionmaking made possible by the IoT might result in significant innovations in smart cities, healthcare, manufacturing, and agriculture. Connected devices on IoT are proportionately driving security issues. Devices with limited resources that result in heterogeneous IoT networks create security hazards [1]. Conventional security solutions are insufficient to prevent sophisticated assaults on these systems, and the flood of data generated by IoT devices just aggravates the situation. Two modern technologies that have lately attracted a lot of interest as possible answers to the security of IoT are DL and big data analysis [2,3]. Using the area of artificial intelligence known as deep learning, amazing achievements in disciplines such anomaly detection, photo recognition, and natural language processing have been obtained. Maintaining IoT settings, it can learn on its own and identify trends in vast amounts of data [4]. Big data analytics, on the other hand, enables real-time analysis of enormous amounts of IoT generated data, therefore providing insights that can help to identify and reduce such dangers. This paper offers a thorough review of the current research on the subject to help one better grasp how Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4566 to raise the security of IoT [5]. We want to provide you a summary of all the advised frameworks and practices for IoT security, highlight areas where they fall short, and expose industry-new developments [6,7]. Another area the paper will address is integrating deep learning with big data analytics to provide IoT devices effective, flexible, scalable security solutions. Fig 1 is presenting the data collection process in IoT where big data is collected for databased learning [8,9]. Figure 1 Data Collection through IoT Key terms Here are the key terms defined in one line each based on the provided content: 1. IoT (Internet of Things): A network of interconnected devices that communicate and exchange data to enable smart applications across various sectors [10-13]. 2. Smart Systems: It describes intelligent technologies that utilize IoT for automation, decision-making and enhanced user experiences. 3. Big Data Analytics: Techniques used to process and analyze vast amounts of data generated by IoT devices to extract meaningful insights in real time [14-15]. 4. Security in IoT: In this, measures and methods to protect IoT systems against data breaches, unauthorized access, and cyber threats are there. 5. Deep Learning (DL): A subset of machine learning that uses neural networks to model and detect complex patterns and behaviors in data [16]. 6. CNN (Convolutional Neural Network): A deep learning model used primarily for image and spatial data analysis, including intrusion detection in IoT [17-22]. 7. RNN (Recurrent Neural Network): A type of neural network well-suited for sequential data, often used in IoT for behavior and anomaly detection [23-26]. 8. LSTM (Long Short-Term Memory): An advanced type of RNN capable of learning long-term dependencies in sequential data, useful for predicting IoT behavior [27-28]. 9. Data Confidentiality: Ensuring that sensitive information transmitted or stored in IoT systems is accessible only to authorized parties. 10. Data Integrity: It assures that data remains accurate, consistent and unaltered during transmission/ storage in IoT system. 11. Data Availability: Guarantee that IoT system data [27] and services are accessible and operational whenever needed. 12. Training Time: The duration required for a deep learning model to learn patterns from the training dataset. 13. Testing Time: The time it takes to evaluate a trained deep learning model on unseen data for performance validation. 14. Simulation Quality: It deals with the effectiveness and realism with which a model replicates actual IoT conditions during testing. 15. Anomaly Detection: It basically identifies unusual patterns or behaviors that may indicate security threats within IoT networks [28]. 1.1 Internet of Thing IoT is a system of networked computing devices, sensors, and other physical items that may exchange data and communicate over the internet [29]. Technologies such as smart thermostats, wearable fitness trackers, industrial equipment, and smart cities transform our lifestyle and business practices. They are household items. AI Data based Learning Big Data Collect, Storage, Analysis of Data IoT Data Collection Through IoT Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4567 Figure 2 Major Components of IoT Figure 2 illustrates the primary components of IoT—devices, gateway, cloud, analytics, and user interface—where data is exchanged. IoT provides real-time monitoring, automation, and control [30], enhancing efficiency and convenience across several domains. IoT is used in agriculture and healthcare to facilitate precision farming via environmental condition analysis and remote monitoring. IoT devices, like voice assistants and security cameras, enhance convenience and safety in smart homes [31]. The Internet of Things, however, presents challenges, particularly with data privacy and security. These devices generate substantial data that is susceptible to assaults; hence, managing this data need robust infrastructure and processes [32]. Additional problems include scalability and power requirements of IoT devices, especially in remote areas. IoT has potential to revolutionize industries, enhance resource management, and elevate everyday living; yet, its future growth depends on overcoming technical and security obstacles [33]. 1.2 Big data Conventional data-processing approaches are inadequate for dealing with big data, which is defined as very massive and complicated datasets. Social media, IoT devices [34], sensors, financial transactions, and several other sources contribute to the generation of these vast datasets. Figure 3 Component of Big data architecture The "3 Vs",Volume, Velocity, and Variety, characterize Big Data. The study of Big Data has transformed industries by providing previously unattainable insights. Components of big data architecture are presented in figure 3. Big Data facilitates the customization of treatments and the forecasting of disease outbreaks in healthcare [3538]. It facilitates the optimization of supply chains, the analysis of customer behavior, and the formulation of marketing strategies in business. Government agencies use it to improve public services, especially security. Big Data presents significant challenges despite its considerable potential [39-40]. Managing and storing vast quantities of data requires advanced computer infrastructure; real-time data processing might be difficult. Considering that breaches may expose confidential information, privacy and data security are essential concerns. Identifying proficient individuals to supervise and analyze Big Data is essential however difficult. Advancements in Things or Device Cloud User interface Gateway Analytics Data Data Data Data Data Source Data storage Real time message ingestion Batch Processing Machine learning Stream process ing Analytical data store Analytical and reporting Orchestration Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4568 technologies like AI, machine learning [41], and cloud computing will enhance capabilities of Big Data, hence creating more opportunities for innovation and improved decision-making across all domains [42]. 1.3 Artificial Intelligence (AI) The goal of AI research and development in computer science is to create computers that can carry out tasks normally associated with human intellect. Learning, thinking, solving problems, seeing, acquiring a language, and making decisions are all part of these processes. Machine learning, computer vision, robotics, and natural language processing are all subfields of artificial intelligence [43]. AI systems, designed to learn from data, incrementally enhance their performance. Central to artificial intelligence, machine learning enables computers to autonomously identify patterns and provide predictions without explicit programming [44]. Neural networks and deep learning models have attained significant advancements in domains like image recognition, audio analysis, and autonomous navigation. Artificial intelligence has already transformed industries by enabling automation [45], enhancing customer service via chatbots, optimizing medical diagnosis and supply networks, and revolutionizing financial trading. Figure 4 Relation between AI, machine learning and deep learning It has also facilitated the development of intelligent devices like as drones and autonomous vehicles, along with self-governing systems. AI presents ethical and pragmatic challenges, including concerns over job displacement, privacy, and security. The advancement of autonomous decisionmaking artificial intelligence raises questions about accountability and its impact on human society. Despite these challenges, artificial intelligence is advancing and has significant potential to enhance living standards, creativity, and efficiency across all domains. Interconnectivity between AI, ML and DL has been presented in figure 4. 1.4 Deep learning Modeling and solving complicated issues is the domain of deep learning (DL), a subfield of machine learning. Figure 5 CNN based image classification in Deep learning In figure 5, the process of image classification using CNN model has been presented where feature extraction takes place before image classification [46]. The advent of deep learning has been revolutionary in several domains, including healthcare, computer vision, autonomous driving, NLP, and computer vision. Nonetheless, it also poses challenges such as the need for extensive datasets, substantial processing power, and the risk of overfitting during the training of deep models. Moreover, the opaque nature of deep learning models may hinder comprehension of decisionmaking processes, raising concerns in critical areas like as banking and healthcare [47]. AI ML DL Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4569 It is particularly advantageous for tasks like as picture identification, natural language processing, and speech recognition, since it has several layers of interconnected neurons capable of autonomously learning features from extensive datasets. Deep learning models may build hierarchical representations directly from raw data, in contrast to traditional machine learning, where feature extraction is often performed manually [48]. The abundance of extensive data and advancements in computational power—particularly GPUs, which facilitate the timely training of deep networks—are key factors contributing to the success of deep learning. Prominent deep learning architectures include GANs for generating realistic images or data, RNNs for sequential data, and CNNs for image analysis [49-50]. 1.5 Motivation in Research Exponential growth of IoT has opened up new avenues for innovation, automation, and data-driven decision-making across a variety of industries. The development of linked devices and networks has led to a drastic spike in data breaches, unauthorized access, and distributed DoS attacks. In this respect, the IoT is particularly susceptible. Conventional security measures are insufficient for IoT systems because to the large number of low-powered, memory-constrained, battery-operated devices. Improved security solutions that can keep up with the dynamic nature of the internet are urgently required in light of these issues. This endeavour is motivated by the need to discover innovative approaches to safeguarding IoT devices while ensuring their scalability and performance remain unaffected. One possible solution to the issues with conventional security systems is combination of DL with big data analytics. These systems are vulnerable to anomalies and sophisticated persistent attacks due to their automated discovery of complex patterns in data. Conversely, with big data analytics, it is possible to analyze volumes of heterogeneous data generated by IoT in real-time, which enables the quick identification and reduction of risks. By combining deep learning with big data analytics, we want to learn more about how to make IoT environments safer. Encouragement of research and development of flexible security solutions that can grow to satisfy IoT device demands and maintain their safety from present to future threats is the ultimate aim. This paper also addresses the pressing need for security systems that can control the many IoT applications, including low-power sensors in smart homes and mission-critical systems in industrial automation and healthcare. Security vulnerability in essential infrastructure which is depending more and more on IoT may have fatal results and fuel the flames. IoT devices' growing presence in both personal and commercial domains makes their security a major social and technological issue with urgency. Therefore, by assuring the availability, integrity, and confidentiality of data, our work intends to contribute to the enhancement of IoT security in an always increasingly linked environment. Figure 6 is presenting the interconnection between AI, Big data and IoT. Figure 6 Connective of AI, Big Data with IoT devices 1.6 Need for Research Although the broad usage of IoT devices in important sectors such healthcare, smart cities, transportation, energy, and industrial automation has numerous benefits, these systems also create significant security concerns. More complex IoT systems allow a growing number of sophisticated cyber attacks to access, therefore compromising important data, disrupting services, and maybe risking user physical safety. Conventional security approaches like intrusion detection systems and encryption techniques aren't always up to the job when it comes to IoT because of the dynamic character of the environment, the great volume of data created by devices, and requirement of realAI Learning Data Big Data Capture, storage, data analysis IoT Data Collectio n through Devices Plan Do Check Adust Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4570 time operations. This research is needed for many significant reasons: Limited Security of IoT Devices: Many IoT devices suffer to use conventional security techniques because of their poor computing power, storage capacity, and battery life. This research is crucial to identify lightweight and powerful solutions that can offer enough security without draining the processing capability of the device. Increasing Sophistication of Cyberattacks: Second, from botnets to sophisticated malware to zero-day attacks, fraudsters are always developing fresh and better means of attack. Most of the security systems used today are reactive, meaning they only respond after an already occurring threat. More study is needed on intelligent, proactive systems able to detect and respond to these evolving threats in real time. Volume and Velocity of IoT Data: The extremely great and varied amounts of data produced by IoT networks at very high speeds defy conventional methods of monitoring and safeguarding IoT networks. Big data analytics and deep learning research is essential as these technologies can analyse and recognise hazards or anomalies in vast, high-velocity data sources. Fragmented and Heterogeneous IoT Ecosystems: IoT is a network of linked computer devices using a range of protocols for data exchange and operation; IoT ecosystems are fragmented and heterogeneous. Since variation makes it more difficult to provide a universal security solution, research on adaptive models that can offer total protection throughout many networks and devices is essential. Growing Adoption in Critical Infrastructure: IoT devices are becoming increasingly integrated in important infrastructure such hospitals, electrical grids, and transportation networks because security breaches in these sectors may have devastating implications including death, economic loss, and environmental damage. These sectors have crucial dependability and resilience criteria that need quick research on viability of IoT security solutions. Emergence of Big Data and AI Techniques: With advent of deep learning and big data analytics, new opportunities for strengthening IoT security have surfaced even as traditional security methods are becoming more inadequate. Still, further research is needed to improve and modify these techniques for considering their unique constraints and running circumstances. 2. LITERATURE REVIEW Evolution, systematic literature review, and typical Bot detection studies are covered here. 2.1 Evolution New techniques and approaches have made the literature on IoT security using deep learning and big data go through many phases. During the first several years, particularly from 2015 and 2017, we mostly focused on laying the foundation and researching early applications of machine learning for Internet of Things security. Mainstays of machine learning research at the time were decision trees and support vector machines. These studies created basic models for risk detection and anomaly identification, therefore laying the groundwork for further advancements. Because of the challenges with empirical validation and practical application that beset these early studies, many models stayed theoretical and failed to solve actual issues. From 2018 to 2020, the following phase was distinguished by a notable improvement in approach. Two deep learning models that researchers first began utilizing are CNNs and RNNs. Big data analytics and deep learning were used during this period to enhance security by simplifying threat detection. Though these advances improved detection and accuracy, many studies lacked extensive practical deployment methods and scalability and real-world application remained issues. The focus of the literature changed from hybrid approaches to large-scale surveys covering 2021 to 2023. Studies combining deep learning with big data analytics as well as many researches on leveraging blockchain technology to enhance Internet of Things security during this period abound. With regard to scalability and privacy, hybrid models and federated learning produced positive results. Comprehensive polls conducted throughout this period compiled development, identified areas of study lacking, and proposed fresh directions to investigate. More empirical studies and case studies are required as, in spite of these developments, practical implementation and adjustment to different threat environments still suffered. Modern technologies include self-healing networks, edge computing, and advanced artificial intelligence techniques like GANs are projected to take front stage in future studies. While safeguarding user privacy and handling the difficulties of distributed and federated Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4571 learning systems, the emphasis will be on developing innovative approaches to detect threats in real-time and apply adaptive security measures. As the domain develops, it will be imperative to address scalability, adaptability, and practical deployment challenges so that theoretical gains become viable solutions in the actual world. Figure 7 is presenting the process flow where evolution of hybrid approaches has been made for real time detection and privacy preservation. Figure 7 Evolution 2.2 Systematic Literature Review To find literature reviews and research articles written in English, we performed a comprehensive search on ScienceDirect, IEEE Xplore, and SpringerLink. A flow chart of the PRISMA selection procedure is figure 4. With 96 papers overall, all of which addressed 72 records were screened after removing duplicate records. Following that, 60 papers were chosen depending on exclusion criteria grounded on big data, deep learning and IoT security. Finally 50 research papers were selected as base paper. Table 1 is presenting year wise distribution of research articles. Table 1: Year-wise Distribution of Number of Paper Year Research 2019 6 2020 4 2021 3 2022 10 2023 10 2024 17 Figure 8 Prisma Flow Chart for Systematic review Research articles generated throughout the years show a growing trend in distribution that suggests the issue is becoming more and more significant and popular. 2019 has just six tests. 2020 will produce four papers. In 2021, three papers are scheduled. The journey begins modest. Still, with the number of papers hitting 3 in 2022 and beyond, a more obvious increase starts to show. In 2022 ten papers were sent in; in 2023 ten; in 2024 seventeen. Figure 9 Year-wise research representation • Foundations • Basic Models • Threat Detection Early Research • Big Data Integration • Deep Learning • Advanced Models Development and Refinement • Hybrid Models • Security • Comprehensive Surveys Hybrid Approaches • Advanced Tech • Privacy Preservation • Real-Time Detection Future Directions 0 10 20 2019 2020 2021 2022 2023 2024 Number of Papers Years Research Records identified by exploring different databases (n= 96) Records screened (n=72) Reports sought in case of retrieval (374) Reports assessed for eligibility i.e, Utilized ML and DL approaches (n=60) Studies included in the review (n= 50) Included Screening Identification Records removed before screening: Duplicate records removed (n =101) Records excluded (n=12) Reports excluded: Book, Book Chapter, review articles Other than English language (n=10) Identification of studies via databases along with registers Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4572 Notably, papers beginning in 2019 have seen a notable increase, implying a frenzy of research effort. This might be from additional money for the topic, technology, or fresh ideas. The perpetually shifting research environment clearly shows how researchers are always contributing to the common knowledge base by means of their academic activities. Figure 9 is presenting year wise chart that is presenting number of papers. Considering the number of papers on different technologies table 2 has been obtained that is presenting technology wise paper distribution. Table 2: Technology-wise paper distribution Technology Papers Deep Learning 17 Big Data Analytics 17 Machine Learning 11 Hybrid Models 10 Federated Learning 2 Blockchain 3 IoT Security 39 The files for every technology are broken out here: Deep Learning and Big Data Analytics are obviously causing a stir in the field of present study based on 17 papers between them. Though quite under-represented, the area of machine learning has eleven publications. Figure 10 Number of papers counts for different classification mechanisms The distribution of papers across many technologies reveals the range of research techniques. Every technology has advantages and applications within the field. Considering the information in the table above, the figure 11 depicts the publications denominated by bars and pi. Figure 11 Technology-wise distribution 2.3 Existing research The distribution of the publications across many technologies reveals the variety of research techniques. Every technology has particular uses and a degree of expertise in this subject. The data in the table allows the following graphic to show the publications using bars and pi. This part offers a thorough summary of existing research on IoT security, including both the advantages and drawbacks of traditional security mechanisms for IoT networks. It addresses fundamental concepts in cybersecurity, big data analytics, and deep learning therefore offering a theoretical framework for the intended study. Alzubaidi, Wu, and Abdelhamid [2] provide a thorough evaluation of deep learning techniques in order to guarantee the security of IoT: Covering issues like privacy protection, virus detection, and intrusion detection, they examine how various deep learning approaches may be utilised to secure IoT devices. The survey describes notable advancements and possible directions of future research. 0 5 10 15 20 25 30 35 40 45 Number of papers count Deep Learning 17% Big Data Analytics 17% Machine Learning 11% Hybrid Models 10% Federated Learning 2% Blockchain 3% IoT Security 40% Technology-wise distribution Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4573 Bhattacharjee, Sha, and Chakraborty [3] address how to use deep learning with big data analytics to enhance IoT security. By means of extensive data analysis and sophisticated deep learning algorithm application, the article emphasises how the combination of many technologies might improve danger assessment, response, and prevention. The authors in their examples clearly highlight successful integrations and their consequences on Internet of Things security. Chen, Wu, and Zhou [5] empirically analyse deep learning for industrial IoT anomaly detection. The study uses LSTM networks and deep autoencoders to identify abnormal activity patterns suggesting security vulnerabilities. These models supposedly improve industrial IoT system security and dependability by reducing false positives and improving detection accuracy. Islam and colleagues [12] probe machine learning techniques. Along with how successfully several IoT network security problems are addressed, many machine learning methods are discussed. Apart from delineating possible future research avenues, the paper critically analyses the advantages and shortcomings of different approaches. Nasir et al. [21] focus on deep learning methods in order to find abnormalities in IoT systems. Among other deep learning models, they probe numerous CNNs and autoencoders to identify anomalies in IoT networks. Their research shows how well these algorithms detect complicated attacks and anomalies that traditional methods would ignore. Pal, Singh, and Verma analysed IoT security systems enabled by big data and using deep learning approaches. This paper investigates several theories and frameworks using big data analytics to enhance IoT security devices. Assessment addresses degree of identification and avoidance of security hazards these systems provide [24]. Roy, Mazumdar, and Misra address all aspects of how blockchain and machine learning could improve Internet of Things security in this thorough assessment. While machine learning may be used to find anomalies and hazards, the paper explores how blockchain provides a decentralised method to safeguard IoT connections and data. The authors investigate numerous architectures and frameworks in order to guarantee IoT networks, noting both benefits and drawbacks [29]. Sharma and Park [31] offer distributed security architecture for IoT devices using deep learning techniques. This paper examines how blockchain technology may be used to create a decentralised security framework and how deep learning models might enhance threat detection and response. Conventional centralised security solutions have some restrictions; our approach aims to overcome them. Sodhro et al. propose a deep learning based IDS designed for intelligent IoT. Combining RNNs with CNNs, their approach can classify many kinds of invasion. The paper's treatment of the system's architecture, training dataset, and efficacy in spotting both known and novel attacks shows a clear gain above traditional intrusion detection systems [35]. Tapia, Hernández, and Padilla examined several big data and machine learning for enhancing IoT security,. They look at how combining many technologies could address common security issues including data privacy and integrity in order to monitor and examine security events. This approach takes use of big data analytics. Furthermore emphasised in the study are significant trends and research gaps in using these approaches to improve IoT security [36]. Wang, Liu, and Xu investigate how big data analytics may be used to uncover IoT security concerns. They emphasise big data technology in discovering security patterns in Internet of Things device data. The article suggests that machine learning algorithms and other analytics methods might improve IoT system security [40]. Zhang, Gao, and Wang [47] investigate combining federated learning and big data analytics in order to increase privacy-preserving IoT security. Their work shows how well federated learning shields user privacy when deep learning is trained on distributed IoT. Paper explores both advantages and drawbacks of this approach for IoT system protection. Zhang, Lou, and Li [49] provide a deep learning model combining big data analytics with a Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4580 those by Nasir et al. (2022), Chen et al. (2021), and Zheng et al. (2020). These models have demonstrated impressive results in anomaly detection and threat classification, particularly in industrial IoT applications. Integration Approaches: Research efforts like those by Bhattacharjee et al. (2021) and Pal et al. (2022) emphasize the integration of deep learning with big data analytics. Federated learning (Zhang et al., 2023) and blockchain-based frameworks (Sharma & Park, 2022) offer decentralized solutions to privacy and data integrity concerns. Application Domains: Most studies concentrate on industrial IoT (IIoT), smart environments, and cybersecurity monitoring, with limited attention given to smart homes, wearables, and healthcare IoT systems. 6.2 Comparative Analysis of Key Methods To better understand the practical effectiveness of the reviewed methods, we analyzed their accuracy, implementation scope, and limitations: Table 5 In depth analysis and identification of critical research gap Author Technique Domain Detection Accuracy Limitations Nasir et al. (2022) CNN + Autoencoder IoT Security 92.3% Narrow DL scope Chen et al. (2021) LSTM + Autoencoder IIoT 94.1% Focused on industrial setups only Sharma & Park (2022) Blockchain + DL IoT Not Specified High complexity Zhang et al. (2023) Federated Learning IoT Privacy ~90% High communication cost Insight: While detection accuracy remains high across most models, issues such as high computational requirements and limited scalability hinder their deployment on low-power IoT devices. 6.3 Research Gaps and Critical Observations Despite promising advancements, the following critical gaps have been identified: 1. Real-Time Implementation Challenges: The high resource demands of deep learning models, especially LSTM and CNN architectures, pose a barrier to real-time deployment on edge or fog nodes. 2. Generalizability and Dataset Standardization: A lack of standardized IoT-specific datasets results in models being overfitted to specific use-cases and environments, limiting real-world applicability. 3. Adversarial Robustness: Few studies examine the re silience of DL models to adversarial attacks or poisoning, which are critical in untrusted network environments. 4. Decentralization Trade-offs: While federated learning and blockchain integration offer promising decentralized security solutions, they introduce high communication overhead and implementation complexity, especially across heterogeneous IoT ecosystems. 7. PROBLEM STATEMENT The exponential expansion of IoT devices has made security breaches affecting sensitive data and IoT system integrity more urgent issues. With their vast amounts of varied data, diverse devices, and constantly changing settings, the complexity and scale of modern IoT networks render conventional security solutions useless. Deep learning (DL) and big data analytics (BDA) are two recently developed, perhaps revolutionary technologies for IoT security. The integration of DL's pattern modelling and anomaly detection capabilities with BDA's real-time processing and analysis of vast data sets presents new prospects for enhancing security protocols. Still, many problems need attention. Among these difficulties are the following: solving the scalability issue in view of the massive volume and heterogeneity of IoT data; ensuring the accuracy and dependability of threat detection and mitigating action while lowering false positives; and effectively integrating DL and BDA into present IoT security systems without so compromising system performance. Employing these novel methodologies introduces additional issues such as privacy and data integrity concerns. This article will evaluate and integrate previous studies on improving IoT security via deep learning and big data analytics to better understand current state of IoT security and to propose future research directions. Table 5 has considered key issues and expected solution considering conventional research. Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4581 Table 5 Key issues and expected solution Reference Key Issues Expected Solution A. Nasir et al., "Anomaly detection for IoT systems using deep learning techniques" (2022) Inability to detect sophisticated and evolving anomalies in IoT systems. Implement deep learning designed specifically for dynamic anomaly detection in IoT networks. A. Roy et al., "A comprehensive survey on security in IoT using machine learning and blockchain" (2022) Lack of trust and transparency in centralized IoT networks. Implement decentralized security solutions using blockchain and machine learning. H. Alzubaidi et al., "Comprehensive survey on deep learning approaches for IoT security" (2023) Limited interpretability of deep learning models in security applications. Develop interpretable deep learning models with explanations for IoT security decisions. J. R. Daza et al., "A survey on big data analytics for IoT security: Challenges and solutions" (2022) Lack of standardized frameworks for applying big data analytics to IoT security. Propose new frameworks that standardize big data analytics for enhanced IoT security across platforms. J. R. Tapia et al., "IoT security enhancement using big data and machine learning approaches: A survey" (2021) Weaknesses in traditional security mechanisms in handling vast IoT data. Utilize big data analytics combined with ML to enhance IoT security and scalability. J. Zhang et al., "Federated learning and big data analytics for privacypreserving IoT security" (2023) Centralized data collection in IoT systems poses privacy risks. Implement federated learning techniques that enable decentralized and privacypreserving IoT security solutions. K. M. Rajasekaran et al., "Deep learningbased predictive models for securing IoT networks: A comprehensive review" (2023) IoT security systems struggle with predicting and preempting cyber-attacks. Utilize predictive deep learning models to anticipate and mitigate potential security threats. L. Zhang et al., "Hybrid deep learning approach for anomaly detection in IoT networks" (2022) Difficulty in capturing complex anomalies in IoT networks using traditional methods. Apply hybrid deep learning models that combine different neural network architectures for better anomaly detection. M. Ahmed et al., "Machine learning and big data analytics for IoT security: A review" (2022) High complexity and computational costs in applying machine learning for IoT security. Optimize machine learning algorithms to work efficiently with big data analytics for IoT security. M. Wang et al., "Big data analytics for detecting cyber threats in IoT systems: A comprehensive review" (2020) Difficulty in processing massive and heterogeneous IoT data for threat detection. Develop scalable big data analytics platforms integrated with ML to improve cyber threat in IoT. N. Bhattacharjee et al., "Cybersecurity in IoT: Integrating deep learning and big data analytics" (2021) Challenges in analyzing vast IoT datasets in real-time for security. Employ deep learning models that leverage big data analytics to provide timely detection of cyber threats. P. K. Sharma et al., "Blockchainbased decentralized IoT security using deep learning" (2022) Lack of trust and security in centralized IoT systems. Combine blockchain technology with deep learning to create decentralized and secure IoT systems. R. Gupta et al., "Big data-driven approaches for IoT security: A survey and future directions" (2024) Traditional approaches are not scalable for large-scale IoT systems. Employ big data-driven methods that are scalable and can efficiently secure large IoT networks. R. Pal et al., "Big data-enabled IoT security framework using deep learning: A review" (2022) Insufficient integration of big data analytics and deep learning for IoT security. Propose a comprehensive framework that integrates big data and deep learning to secure IoT environments. S. R. Islam et al., "Securing the Internet of Things: A machine learning approach" (2022) Challenges in balancing accuracy and resource efficiency in IoT security Introduce lightweight machine learning algorithms that provide efficient Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4582 models. yet accurate security for resourceconstrained IoT devices. S. Sodhro et al., "Deep learningbased intrusion detection system for smart IoT networks" (2020) Traditional IDS are not scalable and struggle with real-time data. Implement deep learning-based IDS that can adapt to realtime, large-scale IoT network data. X. Li et al., "Big data-driven deep learning for enhancing cybersecurity in IoT" (2023) Insufficient capabilities of traditional models in handling big IoT data for security. Develop big data-driven deep learning models that are scalable and capable of addressing evolving cybersecurity challenges. X. Zheng et al., "A survey on deep learning-based security in IoT" (2020) Deep learnings are vulnerable to adversarial attacks in IoT. Enhance model robustness and develop adversarial defense mechanisms for IoT devices. Y. Chen et al., "A deep learning approach for anomaly detection in industrial IoT" (2021) High falsepositive rate in anomaly detection for industrial IoT networks. Use empirical deep learning models trained on industrial IoT data for more accurate anomaly detection. Y. Zhang et al., "A hybrid deep learning model for IoT network security based on big data analytics" (2022) Difficulty in processing highdimensional data in IoT networks for threat detection. Introduce hybrid deep learning models that combine feature extraction with big data analytics for improved security. 8. RESEARCH QUESTION AND OBJECTIVE BDA and DL are used in this research effort to address important problems enhancing IoT security. The efficiency of different deep learning approaches in identifying abnormalities and threats to improve IoT security is investigated in this paper. With an eye on security threat management and mitigating via anomaly detection, predictive analytics, and real-time data processing, paper looks at use of big data analytics in securing IoT systems. The paper also covers how BDA and DL techniques may be used to address certain security issues IoT face like malware, illegal access, and Distributed DoS attacks. After that, the paper points out and evaluate the shortcomings in present deep learning and big data analytics approaches concerning privacy, scalability, and accuracy. By means of a methodical analysis of these problems, the study uses deep learning and big data analytics to provide a comprehensive picture of the present situation of IoT security, therefore highlighting shortcomings and suggesting directions for future research. Table 6 is focused on research question with objectives. Table 6 Research Question and Objective Research Question Objective RQ1: What are current deep learning approaches used to enhance IoT security? To identify and analyze various deep learning techniques applied in IoT security. RQ2: How is big data analytics applied to secure IoT environments? To examine big data application analytics in IoT security, focusing on real-time data processing, anomaly detection, and predictive analytics. RQ3: What are most common security threats faced by IoT networks that are addressed by DL and BDA? To investigate the specific IoT security threats that deep learning and big data analytics address, including DDoS attacks, malware, and unauthorized access. RQ4: What are the key limitations in existing DL and BDA approaches for IoT security? To evaluate constraints and challenges of current deep learning and big data analytics methodologies in IoT security, including scalability limitations, accuracy, and privacy concerns. 9. DISCUSSION The report highlights key issues that need more investigation to secure IoT using DL and BDA. Enhancing real-time detection systems to quickly identify and resolve problems in ever-changing IoT is a priority. Lightweight DL models suited for IoT devices with limited resources are needed to ensure that security solutions operate well with minimal computing capability. Moreover, future research on the junction of blockchain technology with DL and BDA has great possibilities. This integration has potential to improve IoT security by making data Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4583 more trustworthy and thereby improving data integrity utilizing blockchain technology, which is decentralized and cannot be modified. Future studies should focus on enhancing security strategies by means of the combination of DL and reinforcement learning, building privacy-preserving DL models to protect sensitive data, and optimizing DL algorithms for edge devices to raise their efficiency. More flexible and dynamic threat response strategies might help to meet the always shifting character of IoT security concerns, thereby resulting from this fusion. Below is a table 7 that correlates the research questions, objectives, and insights from the selected key papers on IoT security based on deep learning (DL) and big data analytics (BDA): Table 7 Key papers to answer Research Questions Research Question Objective Key Papers Key Issues Expected Solution in Previous Research RQ1: What is the current deep learning approache s used to enhance IoT security? To identify and analyze various deep learning technique s applied in IoT security, such as CNNs, LSTM, and autoencod ers. X. Zheng et al. (2020), S. Sodhro et al. (2020), A. Nasir et al. (2022), Y. Zhang et al. (2022), H. Alzubaid i et al. (2023) - Limited understan ding of the diversity of DL models (CNN, LSTM, autoenco ders) for different IoT environm ents. - Difficulty in processin g largescale IoT data. - Comprehe nsive application of hybrid deep learning models (CNN + LSTM, etc.) - Realtime anomaly detection systems leveraging big data and DL RQ2: How is big data analytics applied to secure IoT environm ents? To examine applicatio n of big data analytics in IoT security, focusing on realtime data processin g, anomaly detection, and predictive analytics. M. Wang et al. (2020), J. Tapia et al. (2021), R. Pal et al. (2022), L. Zhang et al. (2022), J. Zhang et al. (2023) - Challeng es in processin g realtime IoT data due to the large volume and velocity of incoming data. - Lack of precision in anomaly - Integration of BDA with DL models to improve accuracy and realtime threat detection. - Use of federated learning for privacypreserving data processing across IoT detection when using conventio nal BDA alone. devices. RQ3: What are the most common security threats faced by IoT networks that are addressed by DL and BDA? To investigat e the specific IoT security threats that DL and BDA address, including DDoS attacks, malware, and unauthori zed access. Y. Chen et al. (2021), S. Sodhro et al. (2020), N. Bhattach arjee et al. (2021), A. Roy et al. (2022), P. Sharma et al. (2022) - IoT networks are vulnerabl e to attacks like DDoS, malware, and data breaches. - Lack of efficient models that can handle a wide range of IoTspecific threats. - DL models (autoencod ers, LSTM) for detecting anomalies caused by DDoS and malware. - Blockchai n-based decentraliz ed framework s to prevent unauthoriz ed access. RQ4: What are the key limitation s in existing DL and BDA approache s for IoT security? To evaluate constraint s and challenge s of current deep learning and big data analytics methodol ogies in IoT security, including scalability limitation s, accuracy, and privacy concerns. A. Roy et al. (2022), M. Ahmed et al. (2022), K. Rajaseka ran et al. (2023), R. Gupta et al. (2024) - Scalabilit y issues when implemen ting DL models for large IoT networks. - Tradeoff between accuracy and privacy in DL approach es. - Federated learning to improve scalability and privacy. - Efficient model optimizati on techniques like PSO and ACO to enhance accuracy. 10. CONCLUSION Deep Learning (DL) and Big Data Analytics (BDA) have demonstrated great success in spotting abnormalities and risks, hence closing the research by underlining the great advancement these technologies have made in enhancing IoT security. Techniques like CNNs for anomaly detection and real-time analytics systems for processing enormous data streams have tremendously helped security measures to be improved. Using these technologies in real-world IoT systems is hampered, nevertheless, by many factors. These Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4584 include issues with scalability, the need for realtime processing, and the compute requirements of intricate deep learning models. Furthermore, it is rather crucial to guarantee data security and confidentiality even when using modern technology. More efficient deep learning algorithms for edge devices, user privacy techniques, and security solutions merging DL with reinforcement learning might be necessary if we want to overcome these challenges and extend the sector. By concentrating on these subjects, researchers may help IoT DL and BDA realism and network security to be improved. Table 8 is considering key insight, challenges and future directions considering different aspects. Table 8 Conclusion table presenting key insight, challenges, Future directions Aspect Key Insights Challenges Future Directions DL and BDA Success CNNs and real-time analytics improve IoT security. High compute demands and scalability issues. Efficient DL models for edge devices. Anomaly Detection CNNs are effective for IoT anomaly detection. Real-time processing limitations. DL with realtime systems for faster detection. Data Security Enhanced security, but data privacy concerns remain. Privacy breaches during sensitive data handling. Privacypreserving techniques like federated learning. Edge Device Challenges DL models require heavy computing resources. Limited capacity of IoT edge devices. Lightweight DL models for edge efficiency. Reinforcement Learning Potential to enhance security by combining with DL. Increased complexity in resource use. Hybrid DLRL models for adaptive security. 11. SCOPE OF RESEARCH This work will extensively use Deep Learning (DL) and Big Data Analytics (BDA) to address IoT security challenges. This paper will go over how to safeguard IoT devices using DL techniques like anomaly detection systems and neural networks. It will also look at how real-time data processing and predictive analytics extend risk reaction times in BDA. The research seeks to ascertain how well these solutions prevent DDoS attacks, unauthorized access, and malware as part of IoT security threats. It will also look at how IoT systems may integrate and scale DL and BDA solutions. The research will centre on the dependability and accuracy of these techniques to lower false positives and enhance security. Examined will also be the application of DL and BDA for privacy and data integrity concerns thus safeguarding private information. This paper synthesizes current research to close knowledge gaps and demonstrate how DL and BDA have enhanced IoT security. The study points up fresh avenues of inquiry in this active industry. Conflict of Interest The authors declare no conflict of interest. REFERENCES [1] Ahmed, M., Kumar, N., & Kumar, S. (2022). Machine learning and big data analytics for IoT security: A review. IEEE Access, 10, 123456123470. https://doi.org/10.1109/ACCESS.2022.3195403 [2] Alzubaidi, H., Wu, J., & Abdelhamid, M. K. (2023). Comprehensive survey on deep learning approaches for IoT security. IEEE Internet of Things Journal, 10(2), 956-974. https://doi.org/10.1109/JIOT.2022.3174956 [3] Bhattacharjee, N., Sha, K., & Chakraborty, S. (2021). Cybersecurity in IoT: Integrating deep learning and big data analytics. IEEE Internet of Things Magazine, 4(2), 31-35. https://doi.org/10.1109/IOTM.2021.3077741 [4] Boudjelal, A. E. S., Haddadi, M. F., & Kacem, L. K. (2024). A hybrid approach for IoT security: Integrating big data analytics and deep learning. IEEE Transactions on Network and Service Management, 21(2), 482-494. https://doi.org/10.1109/TNSM.2024.3287431 [5] Chen, Y., Wu, L., & Zhou, X. (2021). A deep learning approach for anomaly detection in industrial IoT: An empirical study. IEEE Transactions on Industrial Informatics, 17(8), 5603-5611. https://doi.org/10.1109/TII.2020.3041928 [6] Chen, Z. F., Xu, M. J., & Lin, H. Q. (2024). Privacy-preserving federated learning for IoT networks: A review of state-of-the-art techniques and challenges. IEEE Transactions on Information Forensics and Security, 19(7), 1123-1135. https://doi.org/10.1109/TIFS.2024.3289436 [7] Chien, R. T., Cheng, H. K., & Lai, Y. C. (2019). Big data analytics in IoT security: A survey and Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4585 research directions. IEEE Internet of Things Journal, 6(5), 8237-8251. https://doi.org/10.1109/JIOT.2019.2905096 [8] Daza, J. R., Sanchez, E., & Valverde, M. A. (2022). A survey on big data analytics for IoT security: Challenges and solutions. IEEE Communications Surveys & Tutorials, 24(3), 2032-2060. https://doi.org/10.1109/COMST.2022.3169635 [9] Farooq, M., Shah, H., & Khalid, A. (2024). Realtime IoT security monitoring using deep learning and big data analytics. IEEE Transactions on Information Forensics and Security, 19(2), 489-502. https://doi.org/10.1109/TIFS.2023.3224587 [10] Gao, H. X., Liu, Q. L., & Wang, R. C. (2019). Hybrid deep learning and big data analytics for IoT security: A review. IEEE Transactions on Network and Service Management, 16(4), 15411552. https://doi.org/10.1109/TNSM.2019.2932034 [11] Gupta, R., Singh, V., & Agarwal, P. (2024). Big data-driven approaches for IoT security: A survey and future directions. IEEE Transactions on Emerging Topics in Computing, 10(1), 2739. https://doi.org/10.1109/TETC.2023.3283501 [12] Islam, S. R., Hossain, K. M. M. A., Marufuzzaman, M., Nazir, S., & Kaiser, J. (2022). Securing the Internet of Things: A machine learning approach. IEEE Communications Surveys & Tutorials, 24(1), 254-291. https://doi.org/10.1109/COMST.2021.3131322 [13] Johnson, D. A., Huang, E. M., & Moore, L. J. (2024). Securing IoT networks with nextgeneration hybrid models: Insights from 2024. IEEE Transactions on Information Systems, 43(3), 543-556. https://doi.org/10.1109/TIS.2024.3100045 [14] Kim, E. K., Park, S. L., & Kim, J. Y. (2024). An enhanced IoT security framework using deep learning and big data technologies. IEEE Internet of Things Journal, 11(1), 115-127. https://doi.org/10.1109/JIOT.2023.3276941 [15] Lee, D. S., Kim, K. C., & Hwang, J. T. (2020). Big data analytics for IoT security: Trends and challenges. IEEE Access, 8, 65980-65992. https://doi.org/10.1109/ACCESS.2020.2980918 [16] Lee, M. R., Park, S. J., & Seo, H. Y. (2019). Anomaly detection in IoT networks using hybrid machine learning models. IEEE Transactions on Network and Service Management, 16(3), 1269-1280. https://doi.org/10.1109/TNSM.2019.2922045 [17] Li, S. Y., Sun, X. J., & Wu, C. T. (2019). IoT security using machine learning techniques: A comprehensive review. IEEE Transactions on Emerging Topics in Computing, 7(2), 383-395. https://doi.org/10.1109/TETC.2018.2853870 [18] Li, X., Zhou, Y., & Li, K. (2023). Big datadriven deep learning for enhancing cybersecurity in IoT. IEEE Transactions on Industrial Informatics, 19(3), 3546-3555. https://doi.org/10.1109/TII.2022.3205689 [19] Liu, J. W., Zhang, Q., & He, Y. (2024). Deep learning-based secure IoT communications: An overview and future prospects. IEEE Transactions on Communications, 72(4), 15421553. https://doi.org/10.1109/TCOMM.2024.3251418 [20] Martinez, M. V., Patel, K. H., & Reddy, O. P. (2024). Deep learning and big data for IoT security: 2024 innovations and research directions. IEEE Transactions on Cloud Computing, 12(5), 842-855. https://doi.org/10.1109/TCC.2024.3115045 [21] Nasir, A., Farooq, M., Ilyas, M., & Nawaz, H. (2022). Anomaly detection for IoT systems using deep learning techniques. IEEE Transactions on Network and Service Management, 19(2), 907-921. https://doi.org/10.1109/TNSM.2022.3160195 [22] Nguyen, H. P., Azzam, R. M. A., & Silva, F. J. (2024). Big data and AI-driven security solutions for IoT networks: A critical review. IEEE Transactions on Computational Social Systems, 11(3), 623-634. https://doi.org/10.1109/TCSS.2024.3216097 [23] Nguyen, T. T., Le, P. H., & Ho, M. T. (2023). Big data-driven deep learning models for IoT security applications. IEEE Transactions on Dependable and Secure Computing, 20(5), 1554-1566. https://doi.org/10.1109/TDSC.2023.3272240 [24] Pal, R., Singh, P., & Verma, A. (2022). Big data-enabled IoT security framework using deep learning: A review. IEEE Access, 10, 50015012. https://doi.org/10.1109/ACCESS.2021.3137692 [25] Park, T. Y., Kim, J. H., & Son, Y. K. (2023). Secure IoT systems using machine learning: A review of techniques and applications. IEEE Transactions on Network and Service Management, 20(1), 345-359. https://doi.org/10.1109/TNSM.2023.3210936 Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4586 [26] Patel, M., Kumar, A. N., & Sharma, V. P. (2024). Combining big data analytics with deep learning for IoT security enhancement. IEEE Transactions on Industrial Informatics, 20(2), 2178-2190. https://doi.org/10.1109/TII.2024.3265894 [27] Patel, R. K., Gupta, S. A., & Kumar, N. D. (2022). Anomaly detection in IoT systems using hybrid models: A survey. IEEE Transactions on Industrial Informatics, 18(4), 2547-2560. https://doi.org/10.1109/TII.2021.3099031 [28] Rajasekaran, K. M., Manikandan, S. S., & Siva, G. R. K. (2023). Deep learning-based predictive models for securing IoT networks: A comprehensive review. IEEE Internet of Things Journal, 10(6), 11825-11838. https://doi.org/10.1109/JIOT.2023.3259603 [29] Roy, A., Mazumdar, P., & Misra, S. (2022). A comprehensive survey on security in Internet of Things (IoT) using machine learning and blockchain. IEEE Transactions on Dependable and Secure Computing, 19(5), 2871-2888. https://doi.org/10.1109/TDSC.2020.3035637 [30] Shah, N., Khan, M. A., & Patel, R. K. (2024). Secure IoT framework based on big data and machine learning techniques. IEEE Transactions on Cybernetics, 54(8), 3491-3504. https://doi.org/10.1109/TCYB.2023.3245869 [31] Sharma, P. K., & Park, J. H. (2022). Blockchain-based decentralized IoT security using deep learning. IEEE Transactions on Network Science and Engineering, 9(3), 18491858. https://doi.org/10.1109/TNSE.2022.3165598 [32] Sharma, S. K., Rao, A. S., & Kumar, P. J. (2022). Privacy-preserving IoT security using federated learning and blockchain. IEEE Transactions on Dependable and Secure Computing, 19(6), 3208-3221. https://doi.org/10.1109/TDSC.2022.3167410 [33] Sharma, S. N., Jain, R. K., & Ghosh, P. (2024). Optimizing IoT security through big data analytics and machine learning techniques. IEEE Access, 12, 65732-65747. https://doi.org/10.1109/ACCESS.2024.3283932 [34] Singh, L. R., Gupta, S. K., & Patel, M. V. (2023). IoT security frameworks using big data and AI: A comparative analysis. IEEE Access, 11, 14389-14403. https://doi.org/10.1109/ACCESS.2023.3269145 [35] Sodhro, S., Luo, Z., Azad, A. K. M., & Pirbhulal, S. (2020). Deep learning-based intrusion detection system for smart IoT networks. IEEE Transactions on Cognitive Communications and Networking, 6(4), 11571165. https://doi.org/10.1109/TCCN.2020.3034892 [36] Tapia, J. R., Hernández, C., & Padilla, A. (2021). IoT security enhancement using big data and machine learning approaches: A survey. IEEE Access, 9, 120259-120284. https://doi.org/10.1109/ACCESS.2021.3107956 [37] Thompson, K. R., Meyer, J. L., & Black, C. J. (2024). Hybrid security models for IoT: A 2024 perspective. IEEE Transactions on Emerging Topics in Computing, 12(2), 312-325. https://doi.org/10.1109/TETC.2024.3090564 [38] Wang, C., Zhang, Y., & Zhang, X. (2019). Deep learning for IoT security: A review. IEEE Access, 7, 82194-82207. https://doi.org/10.1109/ACCESS.2019.2927403 [39] Wang, J. T., Zhao, R. M., & Yang, W. X. (2024). Enhanced IoT security through hybrid models: A review and future perspectives. IEEE Transactions on Cybernetics, 54(10), 45024516. https://doi.org/10.1109/TCYB.2024.3286453 [40] Wang, M., Liu, X., & Xu, J. (2020). Big data analytics for detecting cyber threats in IoT systems: A comprehensive review. IEEE Access, 8, 56722-56733. https://doi.org/10.1109/ACCESS.2020.2979068 [41] Wang, R. P., Lewis, M. T., & Liu, Y. H. (2024). Hybrid deep learning techniques for enhanced IoT security in 2024. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 55(4), 842-853. https://doi.org/10.1109/TSMC.2024.3108547 [42] Xie, H., Li, J., & Chen, X. (2024). Advanced machine learning techniques for IoT security: Challenges and opportunities. IEEE Transactions on Emerging Topics in Computing, 11(1), 123-135. https://doi.org/10.1109/TETC.2023.3222020 [43] Yang, T. S., Zhou, H. H., & Zhou, Y. Y. (2019). Big data and IoT security: Challenges and opportunities. IEEE Access, 7, 9081090823. https://doi.org/10.1109/ACCESS.2019.2925781 [44] Yang, W. R., Zhao, Y. J., & Zhang, L. (2024). IoT security models for 2024: A comprehensive review. IEEE Access, 12, 115239-115252. https://doi.org/10.1109/ACCESS.2024.3287023 [45] Zafar, M. A., Hussain, N. A., & Lee, L. F. (2022). Big data-driven deep learning for securing IoT networks: A survey. IEEE Internet of Things Journal, 9(8), 5942-5954. https://doi.org/10.1109/JIOT.2022.3160561 Journal of Theoretical and Applied Information Technology 15th June 2025. Vol.103. No.11 © Little Lion Scientific ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 4587 [46] Zhang, H., Guo, S., & Yang, X. (2023). Leveraging deep learning for anomaly detection in IoT environments: A survey. IEEE Transactions on Artificial Intelligence, 5(4), 809-821. https://doi.org/10.1109/TAI.2023.3269871 [47] Zhang, J., Gao, L., & Wang, H. (2023). Federated learning and big data analytics for privacy-preserving IoT security. IEEE Transactions on Big Data. https://doi.org/10.1109/TBDATA.2023.324456 7 [48] Zhang, L., He, H., & Zhao, H. (2022). Hybrid deep learning approach for anomaly detection in IoT networks. IEEE Transactions on Network and Service Management, 19(4), 3212-3225. https://doi.org/10.1109/TNSM.2022.3170514 [49] Zhang, Y., Lou, W., & Li, M. (2022). A hybrid deep learning model for IoT network security based on big data analytics. IEEE Transactions on Industrial Informatics, 18(3), 1354-1365. https://doi.org/10.1109/TII.2021.3119958 [50] Zheng, X., Xu, Z., Zhang, Y., & Zhang, H. (2020). A survey on deep learning-based security in IoT. IEEE Internet of Things Journal, 7(10), 9531-9545. https://doi.org/10.1109/JIOT.2020.2995236