International Journal of Electronic Security and Digital Forensics (72 papers in press)

Regular Issues

• An intelligent method for detection and classification of Darknet traffic using sequential model along with Adam and stochastic gradient decent optimisers  
  by Ravi Sheth, Chandresh Parekha, Kinjal Sheth 
  Abstract: The clandestine nature of darknet activities poses a significant challenge to traditional cybersecurity measures, necessitating advanced techniques for effective detection and classification. Darknet traffic classification is very much needed now days as day by day the market of illegal and hidden services are being increased in the darknet. There are various machine learning-based approach has been proposed for the categorisation of darknet traffic but very few work has been done using the concept of deep learning. This research introduces an intelligent approach which leverages a sequential deep learning model to enhance the accuracy and efficiency of darknet traffic detection and classification. In the training phase, the model is exposed to a diverse dataset encompassing a wide range of darknet traffic patterns, ensuring its ability to generalise and recognise novel patterns in real-world scenarios. The proposed model has used sequential model along with the stochastic gradient descent (SGD) and Adam optimiser which successfully detect and classify the darknet traffic with the overall accuracy of 96.77%.
  Keywords: sequential model; Adam; stochastic gradient decent; SGD; darknet traffic; detection; classification.
  DOI: 10.1504/IJESDF.2025.10064345
   
  
• On classifying memory contents at page-level granularity: machine-learning approach  
  by Mohammed I. Al-Saleh, Akram Alkouz, Abdulsalam Alarabeyyat, Majed Bouchahma 
  Abstract: A significant challenge faced by digital investigators in the realm of law is performing digital media triage, which involves determining the relevant data that may aid in a criminal investigation. Effective triage can save time and improve investigative outcomes, particularly in memory investigation, as its contents are often scattered and diverse. Identifying and classifying file types in memory can be difficult due to the way the operating system’s paging scheme maps file contents into non-consecutive page frames in physical memory. This paper presents a machine learning approach to triage memory content at the page level, focusing on the classification of common file types within the context of law. The study conducted various experiments, and the results indicate that it is possible to accurately classify in-memory data into primary file categories, thus contributing to the field of digital investigation in accordance with legal processes.
  Keywords: memory forensics; digital media triage; machine learning; file type detection; digital forensics; classifying memory content.
  DOI: 10.1504/IJESDF.2025.10064346
   
  
• Machine learning-based cyber attack recognition model  
  by Leo John Baptist, Janani Selvam, Divya Midhun Chakkaravarthy 
  Abstract: Internet plays an essential role in the daily lives of individuals living in the contemporary world. Because of the volume of users, our private information runs the risk of being disclosed inadvertently somewhere else on the internet. The study of cyber security encompasses a wide range of topics, the most basic of which are the abuse of data and risks to internet security. The proposed system that performs an analysis of the dataset and determines if the data in question is typical or out of the ordinary. Following the completion of the dataset analysis, make an effort to recognise and forecast a cyber attack. The ensemble classification approach is used to determine the attack wise detection accuracy found by CADM. The categorisation of network traffic data has been done with the help of the gradient boosting and random forest algorithms. We achieved an accuracy level of 97.4%.
  Keywords: cyber attack detection; deep machine learning; DML; smart power grid; data processing.
  DOI: 10.1504/IJESDF.2025.10064475
   
  
• Exploring machine learning techniques for detecting anomalies in digital forensics: a survey  
  by Khawla Almutawa, Afef SELMI, Tarek Moulahi 
  Abstract: Cybercrime has recently and rapidly increased as a result of the extensive use of various digital devices. Digital forensic science, which was established to address issues of cybercrime, follows a defined approach to gathering digital evidence. In recent years, there has been a growing number of studies focusing on employing machine learning and deep learning in digital forensics applications. This research is motivated by the increasing importance of digital forensics and cybersecurity and the need for accurate and efficient tools to detect and mitigate potential security breaches and other forms of anomalous behaviour in digital systems. The purpose of this study is to conduct a literature review to obtain a comprehensive understanding of this topic, specifically deployed models, data preprocessing mechanisms, anomaly detection techniques, and evaluations. This review will help to identify gaps in the existing knowledge and potentially uncover new approaches to the topic. This review conducts a comprehensive literature review on anomaly detection in log files using ML and DL techniques. It will help to identify gaps in the existing knowledge and potentially uncover new approaches to the topic. Initial results indicate that deep learning methods show promise in effectively dealing with the intricate characteristics of log data.
  Keywords: digital forensics; DF; machine learning; ML; anomaly detection; log files.
  DOI: 10.1504/IJESDF.2025.10064574
   
  
• Matrix-based homomorphic encryption-using random prime numbers  
  by Sonam Mittal, Ketti Ramachandran Ramkumar 
  Abstract: Cloud service providers (CSPs) provide security to data during communication and storage, but data security during computation remains a challenge. Homomorphic encryption facilitates the computation of the encrypted data to guarantee the confidentiality and security of the client’s data. Most of the existing homomorphic encryption schemes work on bit-level plaintext and have various issues, such as large ciphertext size, impractical key size, noise growth, and more computational overhead. etc. The paper presents a homomorphic encryption algorithm based on integer numbers which uses the camouflage process and a matrix, filled with random prime numbers to transform the original plaintext to the next prime number and to have a more secure encryption scheme with reduced computational and memory overheads. The comparative analysis shows the memory requirement, ciphertext size, and key length as 4,896 bytes, 16.7 digits, and 3.6 digits for variant 1 and 4,855 bytes, 17.1 and 3.5 digits for variant 2 are less than existing standards.
  Keywords: homomorphic encryption; security; random prime numbers; substitutional matrix; camouflage process; cloud service providers; CSPs.
  DOI: 10.1504/IJESDF.2025.10065110
   
  
• Enhancement of data security in wireless sensors network: application in internet of things  
  by K. Sathiya Priya, C. Rajabhushanam 
  Abstract: As wireless sensor networks (WSNs) expand their application areas and become more widely utilised, the need for security in these networks becomes unavoidable and essential. The fundamental properties of wireless sensor networks do, however, impose limits on sensor nodes. These constraints include limited energy, processing capabilities, and storage capacity, among other things. The purpose of this article is to offer an outline of the problems about privacy and security that are linked with sensor networks. It has been shown that asymmetric key cryptography is not suitable for wireless sensor networks. However, in light of the development of new energy-efficient sensor nodes such as TelosB and others, researchers are exploring and assessing the influence that public key cryptography has on wireless sensor networks. At the time of data transmission, the proposed work achieves a 9.67% improvement in security, and it achieves an 11.38% improvement in security.
  Keywords: secure data aggregation in WSN; concealed data aggregation; homomorphic encryption.
  DOI: 10.1504/IJESDF.2025.10065111
   
  
• Constitutional Protections in Utilising Artificial intelligence Systems for Investigating and Inferring Crimes: A Comparative Study  
  by Ashraf Fathi Al-Rai, Nayel Musa AlOmran 
  Abstract: This paper explores the potential use of artificial intelligence AI) systems in investigating and deducing crimes and examines their impact on constitutional and legislative guarantees. With the rise of AI entities committing crimes using advanced systems. However, it can be used like any other technical system in the investigation and reasoning process. The problem arises in how to apply constitutional guarantees to the accused when AI systems investigate and deduce the crime. The paper reveals that there are no constitutional or legislative texts in Jordan related to the use of AI systems during the investigation and inference phase of crimes. Although many countries use AI systems for data and information collection, such as the United States of America, Britain, Germany, and the United Arab Emirates, they have not yet used them legally in the inference, investigation, and investigation processes through AI itself.
  Keywords: constitutional; guarantees; artificial intelligence; investigation; reasoning; crimes.
  DOI: 10.1504/IJESDF.2026.10065277
   
  
• Image encryption using deep learning : application of AI in medical Images  
  by Ravi Kishore Veluri , Sulakshana B. Mane, V. Sureka , K. Gokulkannan  
  Abstract: The Fourier frequency domain provides the opportunity to differentiate between the dominant frequency of each collection of pictures. After each group is stacked on top of the others, the ciphertext is scrambled. This process is repeated until the final ciphertext is constructed. Throughout the whole decryption process, deep learning is used in order to improve the speed at which the decryption process is carried out and the quality of the recovered image. In particular, the ciphertext that has been retrieved may be sent into the neural network that has been trained, and after that, the plaintext image can be immediately recreated. The results of experimental study indicate that the CC of the decrypted output may be more than 0.99 when 32 photos are encrypted by the process.
  Keywords: optical information security; deep learning; sinusoidal coding; frequency multiplexing.
  DOI: 10.1504/IJESDF.2026.10065826
   
  
• Advanced intelligent data hiding using video stego and hybrid convolutional neural networks  
  by Ravi Kumar  
  Abstract: The practice of steganography involves the concealment of confidential data inside other, seemingly innocuous files of the same or other sorts. The objective of this study is to create a stego technique that, when applied to a video clip, will successfully conceal a message inside its graphics. A model is developed for video steganography by developing a model to conceal video inside another video using hybrid convolutional neural networks (HyCNN). The second objective is to expand the size of the space that can be used for hiding, which has been accomplished via the use of CNN. The suggested model was trained using HyCNN on arbitrary pictures drawn from the ImageNet database. The findings also show that the system is able to produce excellent results in visibility and attacks, where the suggested approach is able to effectively mislead both the observer and the steganalysis software.
  Keywords: convolutional neural networks; hiding data; image stego; steganography; video stego.
  DOI: 10.1504/IJESDF.2026.10066025
   
  
• Securing wireless sensor networks using machine learning and blockchain  
  by K. Sathiya Priya , C. Rajabhushanam 
  Abstract: This study studies the prospect of improving the reliability and security of a Wireless Sensor Network (WSN) by using blockchain technology. The process starts with the gathering of routing data via the use of Q-tables in MATLAB. This is followed by the uploading of the data to the blockchain, where it is encrypted using SHA256 cryptography to ensure its safety. Proof of Work (PoW) and Proof of Authority (PoA) are the two consensus algorithms used to evaluate the robustness of the blockchain network. We carry out a Sybil attack on the blockchain network with the intention of determining how effective Proof of Work and Proof of Authority are in identifying and preventing breaches of security. The evaluation demonstrates that Proof-of-Work (PoW) offers greater security assurances, preserving the integrity, validity, and resilience of the blockchain against the attack, despite the fact that it requires more computing resources.
  Keywords: wireless sensor network; WSN; machine learning; blockchain; agriculture; security.
  DOI: 10.1504/IJESDF.2026.10066091
   
  
• Digital distinctiveness - a study and review on the methods that prevent identity hacking in metaverse  
  by Lakshmi Mansi Chada, S.P. Raja 
  Abstract: Digital identity is the most crucial element of any social technology, yet it is often the most vulnerable. The neurally connective nature of metaverse makes it very closely dependent on digital identity. While digital identity helps metaverse, by making it a user centric social technology, it also makes it quite vulnerable to identity hacking. Several crimes that occur in metaverse are interlinked with the theft of digital identity. The objective of this review paper is to conduct a comparative study on the currently existing methods that aim to preserve the digital identity of metaverse users. This comparative study will aim to understand each method based on two sets of parameters, which will give a clear report in terms of extent and availability of certain features that are necessary to prevent identity hacking. This study will also guide users in choosing the best suited preventive method for identity hacking, for their virtual environment. Finally, the outcome of this review is to discuss the functioning of each method and understand its role in preserving digital identity in metaverse and have a clear picture of the comparative study and choose the best method for this virtual environment.
  Keywords: metaverse; identity hacking; SSI; passwordless authentication; ZTA model; decentralised identity using blockchain; security; privacy.
  DOI: 10.1504/IJESDF.2026.10066334
   
  
• Behavioural cyber malware controller and blocker using block chain, machine learning, and VAPT  
  by Sulakshana B. Mane, Mohd Zafar Shaikh, Kiran Shrimant Kakade, Jayant Brahmane 
  Abstract: When talking about topics such as Digital India, the internet Era, or the internet of things, the role of providing people with a sense of safety is of the highest significance. Because we deal with such a huge number of different types of information, we are confronted with a wide variety of issues brought about by cyber infection. There is a kind of harmful software known as ransomware that may be discovered in internet. In addition, its influence will execute a variety of actions, including the theft of personal data, the misuse of data, and unauthorised access. When it spreads, it will encrypt your data and lock your machine. The purpose of this study is to explore the many different defensive techniques that might be used in response to ransomware attacks. In conclusion, the proposed algorithm got the best result in 66 ms, which was a classification accuracy of 99.41%.
  Keywords: digital India; cyber malware; ransom ware; security; block chain; preventative framework.
  DOI: 10.1504/IJESDF.2026.10066335
   
  
• Mortgage-based securities data hybrid encryption for financial data analysis  
  by Humashankar Vellathur Jaganathan  
  Abstract: Attribute-based encryption (ABE) is the most effective access control mechanism for protecting the cloud storage environment. The data of the data owner are separated into two distinct security levels and kept in various cloud providers according to our plan, which results in an increase in the degree of security that is provided by outsourcing data. Furthermore, our system, which is based on ciphertext-policy attribute-based encryption (CP-ABE), is able to not only offer a fine-grained access control for the data user, but it can also totally leverage the cloud side to permit outsourcing decryption. For the purpose of ensuring security, the FHE model achieved a gain of 88%, the AES approach achieved a gain of 81%, the RSA model reached 92%, and the PHE technique achieved 94%. A security level of 99.74% was obtained by the ALO-DHT model that was built. The ALO-DHT model that was built was successful in achieving 99% secrecy.
  Keywords: attribute-based encryption; multi-cloud; outsourcing decryption; access control.
  DOI: 10.1504/IJESDF.2026.10066461
   
  
• Host-based threat hunting framework for log analysis  
  by Parag Shukla, Sandesh Ajgekar, Jay Teraiya 
  Abstract: Sysmon is a Windows system service and device driver. It is designed to persistently monitor and record system activity in the Windows event log. Sysmon tool is a data source for host-based intrusion detection and it is open-source and free. Being a sophisticated logging tool for Windows, Sysmon lacks suspicious activity identification, log parsing and analysis capabilities. We also need to check the reliability of Sysmon in intrusion detection as an endpoint visibility tool. Hence, as a supporting tool we developed the Huntmon framework for log parsing and to provide some basic capabilities that might be crucial in any type of investigation. This framework is a general multi-purpose Sysmon log parser tool. Along with the Sysmon log parsing, the huntmon framework provides dynamic analysis of Sysmon logs. This tool is compared with other traditional tools with the same test objects. The outputs of both tools are discussed in this paper.
  Keywords: system activity monitoring; host-based intrusion detection; log analysis; VirusTotal Lookup; portable executable headers; portable executable strings; process execution block; dynamic analysis.
  DOI: 10.1504/IJESDF.2026.10066475
   
  
• The importance of administrative enforcement in protecting the family from digital domestic violence, analytical study between reality and hope  
  by Jehad D. Aljazi, Odai Mohammad Ali Al Heilat 
  Abstract: This research addresses the legal issues related to the emergence of digital domestic violence, and the consequences of it, which led to the occurrence of crimes that are a great danger to society and the family. Therefore, administrative enforcement authorities intervene to prevent and limit their occurrence. This research also aims to clarify the position of Jordanian and Moroccan legislation sometimes in addressing digital domestic violence. The main problem of this paper revolves around the extent to which the Jordanian legislation achieves the objective of protecting the family from digital violence. To solve this problem, we divided this research into two parts. In the first, we dealt with administrative control measures in digital domestic violence, and in the second, administrative deterrent mechanisms in domestic violence. The research reached important outcomes; fore mostly is that the penalties imposed by the Jordanian legislator are insufficient to achieve effective protection from digital domestic violence. We recommend developing alternative penalties with significant effects aiming to correct the behaviour of the perpetrator of domestic violence.
  Keywords: administrative penalties; measures; family; behaviour; administrative control; digital violence; technical development; fine; Jordan.
  DOI: 10.1504/IJESDF.2026.10066524
   
  
• Encryption-based safe cloud data storage using Merkle hash sum tree with message authentication code  
  by G. Gangoni Vani , Y. Ambica, Rohita Yamaganti, Aruna Varanasi 
  Abstract: Security is the most important thing and seen mainly in computer applications, especially in the data of cloud computing security on storage plays an important role. The cloud provides a desirable platform for cognitive smart cities to access user data, enabling them to adapt their current actions and learn from past experiences. The data in a cloud has minimum security, because of mutable establishment procedure of data integrity. In this study, the Merkle hash sum tree with message authentication code (MHST-MAC) approach is proposed for enhancing the security in the cloud. This approach supports the privacy-preserving public auditing to give a secure storage to cloud. The files in a data are requested by a data owner (DO) that audits with third-party auditor (TPA) as well as multi-owner authentication approach is concerned when alternate processes to authenticate a user. The results show that a proposed MHST-MAC approach delivers the performance metrics such as encryption as well as decryption time values about the 126 ms and 323 ms for 200 (KB) file size compared with existing methods such as Rivest Shamir Adleman (RSA) and ElGamal algorithms.
  Keywords: cloud storage; decryption; encryption; message authentication code; Merkle hash sum tree; third-party auditor; TPA.
  DOI: 10.1504/IJESDF.2026.10066615
   
  
• Leveraging advanced deep learning algorithms to combat fake news in Arabic media landscape  
  by K. Chitra, E. Srimathi , R. Rajpriya, Edwin Shalom Soji, R. Balamurugan, S.Silvia Priscila 
  Abstract: The increasing presence of false information online in today’s digital era can lead to societal issues such as political upheavals and the circulation of incorrect data. This study introduces a unique method to identify untrue reports in Arabic utilising advanced deep learning methods. We review existing literature on fake news detection and discuss the challenges specific to Arabic language processing. Our approach entails creating a deep learning architecture customised to Arabic language and extensively evaluating its performance. Results, graphical representations, tables, mathematical equations, and model efficacy discussions are presented. While our approach shows promising results, we also acknowledge its limitations and propose future research directions. However, the journey continues. Research and innovation are needed to address dataset scarcity, model interpretability, and adversarial attacks. Staying ahead of disinformation providers’ increasingly sophisticated strategies is crucial as the digital world advances. This study guides us and reminds us of the challenges ahead. With this researchs tools and insights, we can fight fake news better.
  Keywords: false information identification; advanced learning methods; Arabic tongue; linguistic data processing; neural systems; text classification; sentiment analysis; information warfare.
  DOI: 10.1504/IJESDF.2026.10066981
   
  
• The impact of cybercrimes on the achievement of sustainable development goals. Analytical study  
  by Muaath S. Al-Mulla, May Hammoud, Ahmed Fekry Moussa, Jamal Barafi 
  Abstract: The study aims to highlight the impact of cybercrimes on achieving the goals of sustainable development. Artificial intelligence has changed the methods of these crimes and exacerbated the problem of the lack of sufficient international cooperation to reduce their effects, as well as the loss of efforts towards the idea of cybersecurity. We followed a descriptive-analytical approach that begins by explaining the development of cybercrimes, from the stage of human control to the stage of auto-control by machines, to highlight the weakness of the law and the faltering international efforts to confront these crimes and mitigate their effects. The study provides evidence that current national laws and international agreements are not effective in combating these crimes and reducing their risks to sustainable development. Recommendations include unifying international efforts to conclude an international agreement obligating the parties to confront these crimes.
  Keywords: cybercrime; artificial intelligence; sustainable development; law; international cooperation.
  DOI: 10.1504/IJESDF.2026.10067069
   
  
• Increasing network security using an enhanced hybrid deep intrusion detection model  
  by Jiacheng Wu, Tingting Jiang, Juan Li, Wujun Mei 
  Abstract: Internet of things (IoT) systems have recently seen a widespread use of machine learning (ML) methodologies for intrusion detection systems (IDSs). LSTM and GRU models, which are the RNNs, are used to identify the many kinds of threats that may occur in IoT systems. The Harris hawk optimisation and fractional derivative mutation methods are used in this study to perform feature choices. To evaluate the suggested technique, datasets that are accessible to the public were used. The empirical analysis revealed that the proposed method is superior to the other related approaches in accuracy and efficiency. The proposed model makes use of several databases. The proposed model attained a maximum accuracy of 100% in identifying attacks such as denial of service, exploits, generic, reconnaissance, and shellcode attacks respectively. This model provides a 99.7% accuracy rate.
  Keywords: intrusion detection systems; IDSs; machine learning; metaheuristics deep learning; long short-term memory.
  DOI: 10.1504/IJESDF.2026.10067347
   
  
• Securing renewable energy supply chains: challenges and opportunities for information security  
  by Al-Amin Abba Dabo, Amin Hosseinian-Far, Olalekan Adisa 
  Abstract: This paper explores the information security challenges and opportunities in renewable energy supply chains as they integrate advanced technologies from Industry 4.0 and Industry 5.0. Through a review of existing literature and industry insights, it identifies key sensitive data types at risk, such as intellectual property, operational data, and business strategies. The study highlights how the risk levels vary between Industry 4.0 and Industry 5.0 and examines the potential consequences of cyber-attacks, data breaches, and regulatory non-compliance. It also outlines opportunities to enhance security using advanced technologies, risk management strategies, and industry collaboration, offering valuable insights for stakeholders in the renewable energy sector.
  Keywords: renewable energy; supply chains; renewable supply chains; supply chain disruption; Industry 5.0; Industry 4.0; information security; cyber-attacks.
  DOI: 10.1504/IJESDF.2026.10067348
   
  
• Deep learning to detect social media fake news using sequence generative adversarial networks  
  by Saravanan Venkataraman, S. Albert Antony Raj, S. Belina V.J. Sara, S.Silvia Priscila 
  Abstract: Even if major efforts were made to verify the facts, the rising amount of fake news on social media, which has had a major influence on righteousness, confidence in others, and the community, persisted. This study offers SeqGAN to improve social media false news detection. By solving social media content’s unique problems, SeqGAN proves its sequence creation capabilities. The version uses SeqGAN’s generative energy to generate sensible text sequences and a discriminator network to identify bogus information narratives. Due to more accurate synthetic data, this unfavourable training method challenges the discriminator. The model’s ability to detect misinformation is tested here. It also evaluates SeqGAN on large datasets and uses multiple false news detection approaches to provide an overview. These findings illustrate the rise of bogus news, which could propagate on social media using SeqGAN. Experimental results reveal that the SeqGAN model outperforms standard false news detection approaches. The model is more sensitive to misinformation campaign linguistic and temporal quirks. The SeqGAN-based method works well in social media and online chat environments. Python’s model classifies instances with 98.5% accuracy and great proficiency, indicating its robustness in identifying truth.
  Keywords: sequence generative adversarial nets; SeqGAN; false news detection; social media; generative adversarial networks; misinformation campaigns.
  DOI: 10.1504/IJESDF.2026.10067349
   
  
• Cyberbullying detection and recognition using deep learning: SVM classification  
  by D. Maalini , I. Nandhini , S. Nelson , K. Umamaheswari  
  Abstract: Strong computational techniques are required for the growing importance of cyberbullying detection on social media platforms. A risk that some users may take use of these possibilities to humiliate, degrade, abuse, and harass other individuals. Research provides a comparative examination of various distinct deep learning procedures with the purpose of testing and evaluating the performance of deep learning methods in relation to a well-known worldwide Twitter dataset. The dataset in question is Twitter. Twitter is the dataset that you are referring to. The detection of abusive tweets and the discovery of remedies to the problems that are now being faced have both been accomplished via the use of attention-based deep learning algorithms. In order to extract the features, an application of the word2vec technique that was concatenated with CBOW was used. The proposed model achieves the accuracy rate of 90% for the given dataset.
  Keywords: cyberbully; RNN; CNN; LSTM; BiLSTM; word2vec; text classification.
  DOI: 10.1504/IJESDF.2026.10067637
   
  
• Unveiling Bitcoin transactions: a forensic framework beyond predefined artefacts  
  by Borase Bhushan Gulabrao, Digvijaysinh Rathod, Nitin Sharma, Aishwarya Tiwari 
  Abstract: This research presents a novel approach to Bitcoin forensics that transcends the limitations of predefined digital artefacts. This research delves into RAM and disk analysis associated with various Bitcoin wallet types. These wallets are studied on eight different parameters. The methodology includes structured and unstructured RAM analysis and disk analysis. This approach focuses beyond pre-defined artefacts and yields a comprehensive list of 33 digital artefacts linked to Bitcoin activities. Additionally, it identifies 46 crucial keywords that can empower LEAs to detect artefacts in seized digital devices. This research offers significant advancements in the field. The identified artefacts provide valuable insights into Bitcoin transactions, and the keywords serve as powerful tools for investigators. In conclusion, this study presents a groundbreaking framework for Bitcoin forensics. The study offers a more robust methodology for uncovering evidence of Bitcoin transactions. This research can significantly enhance investigative capabilities in the ever evolving realm of cryptocurrency crime.
  Keywords: Exodus; Electrum; ledger; OKX; public key; private key; passphrase; Trezor.
  DOI: 10.1504/IJESDF.2026.10067856
   
  
• The consequences of using autonomous weapon systems in light of international humanitarian law  
  by Naser Al Ali, Ziad Alwahshat, Vladimir Chebotarev 
  Abstract: Technological development, particularly in artificial intelligence, has led to the production of autonomous weapon systems. The use of this new type of weapon poses numerous challenges and issues regarding the suitability of the rules of international humanitarian law for these new means of warfare. This study aimed to explore the boundaries, prospects, and challenges of applying international humanitarian law to the use of autonomous weapon systems in armed conflicts, and to investigate their legitimacy. Therefore, the research aimed to address its primary inquiries, namely: what is the concept of autonomous weapon systems? What is the legitimacy of using autonomous weapon systems in armed conflicts? What is the legitimacy of using autonomous weapon systems in legitimate self-defence, and how can international humanitarian law be applied to the use of autonomous weapon systems in armed conflicts? What is the future of international engagement with autonomous weapon systems?
  Keywords: weapon systems; international humanitarian; killer robots; autonomous weapon; artificial intelligence.
  DOI: 10.1504/IJESDF.2026.10067920
   
  
• Digital technologies and legal regulation  
  by Nurlybek S. Nusipzhanov  
  Abstract: The purpose of the study was to clarify the need for a legal regulation of the digitalisation process in the field of civil relations. The methods that were used in the study are as follows: historical, statistical, analysis, induction, and comparative. The main results of the study should be highlighted as follows: analysis of the historical aspect of the development of digital technologies and their impact on the development of civil law; clarification of a number of specific terms, examination of their legislative interpretation; analysis of the state strategy of digital development of the Republic of Kazakhstan and highlighting positive aspects; analysis of the state of legislative regulation of the field and influence on the normal course of civil law relations in society, considering the risks of violation of authorship, confidentiality in the Republic of Kazakhstan and other countries; the positive world experience of regulating the relevant fields.
  Keywords: information technology; legislation; legal tech; digitisation; copyright issues; civil law.
  DOI: 10.1504/IJESDF.2026.10067921
   
  
• Blocking the internet and its impact on digital human rights and freedom  
  by Ruba Hmaidan, Tareq Al-Billeh, Ali Al-Hammouri, Abdulaziz Almamari, Mohammed Al Makhmari 
  Abstract: Many countries of the world suffer from the issue of blocking the internet. Meanwhile, some countries instituted protection against blocking the internet by considering that blocking the internet is an infringement of digital human rights that is enshrined in United Nations General Assembly resolutions. Blocking the internet can cause damage to health and education facilities and limit political participation. This blocking of the internet imposed by various governments in countries of the world causes serious harm to people’s human rights. Foremost among them is the right to freedom of expression and opinion. In that connection, it should be noted that the internet was blocked during election periods, protests, and the use of force by law enforcement agencies or the holding of examinations for some schools and universities. It should therefore be asked why the internet was blocked, which was usually accompanied by gross violations of digital human rights.
  Keywords: internet blocking; human rights; freedom of opinion; freedom of expression; digital rights; internet network.
  DOI: 10.1504/IJESDF.2026.10068035
   
  
• Machine learning in IoT digital forensics: a comprehensive review  
  by Abdullah Aldhayyaf, Samer Atawneh, Bushra Abdullah Shtayt 
  Abstract: Today, the pace of development of IoT devices has increased, which has led to their great use in various fields; therefore, ensuring their security against potential threats has become very important with the expansion of the IoT ecosystem. Digital forensics helps investigate and prevent cybercrimes. IoT forensics focuses on identifying the attack's origin on the IoT devices and networks. Machine learning algorithms can process vast amounts of data and identify cyberattacks that might go unnoticed by traditional methods. By analyzing data from IoT devices, these techniques can help in the early detection and prevention of cyber threats. This research seeks to explore the role of machine-learning methods in IoT digital forensics by providing a comprehensive look at the studies that dealt with this topic from multiple perspectives regarding the machine-learning techniques and applications in IoT forensics. Anomaly detection will also be discussed in this paper.
  Keywords: internet of things; IoT; digital forensics; DF; machine learning; ML; deep learning; DL; anomaly detection; security.
  DOI: 10.1504/IJESDF.2027.10068301
   
  
• Analysis of post-quantum cryptography algorithms in IoT communication  
  by S.V. Bhaskar, T. Gayathri , D. Chitra Devi, Ravi Kumar 
  Abstract: Quantum computing and cloud computing are two new technologies that have made a big difference in the computer business. These tools give us a lot of power and the chance to grow. This makes regular cryptography less safe, even though it is still useful. RSA and ECC are in danger because of things like Shors algorithm, which can speed up numbers a lot. The way quantum computing works could be bad for both of these. Now that this is known, these systems can be broken into, so we need encryption that quantum events cannot break. It is harder to keep data safe when you use cloud computing, so you need to setup strong security measures and rules for who can view what. The short integer alternative (SIS) problem could be a good fit for lattice-based security. The piece talks about unique public key encryption, a type of encryption that even quantum computers cannot break. ElGamal and SIS are where it comes from. In modern cryptography, the goal of this method is to keep you safe from both quantum and traditional dangers.
  Keywords: secret key exchange; SSI network; SIS problem; the offered scheme’s security.
  DOI: 10.1504/IJESDF.2026.10068375
   
  
• Enhancing cyber threat prediction utilising digital forensics  
  by Qingli Zhang, Guohui Quan 
  Abstract: As we move into the Fourth Industrial Revolution, smart houses that use the Internet of Things (IoT) make digital sleuthing a lot more difficult. The study's goal is to look into these issues and create methods that can be used with a lot of IoT gadgets in smart homes. We come up with a complete method that includes open-source intelligence, application, network, and hardware studies. Our goal is to make this method work for all kinds of IoT devices and their different ways of storing data. A lot of tests were done on popular systems like Sony SmartThings, Aqara, QNAP NAS, and Hikvision IP cameras to make sure the suggested method works. The study makes big progress in the field of IoT digital forensics and sets the stage for more research into IoT cases that are more general. Datas from 7 countries are analysed in this article.
  Keywords: Internet of Things; IOT; incident investigation; smart home; digital forensics.
  DOI: 10.1504/IJESDF.2026.10068398
   
  
• An intelligent approach to detect and classify malicious URLs using multi-layer fine-tuned long short-term memory with Adam and RMSprop optimisers  
  by Ravi Sheth, Chandresh Parekha 
  Abstract: In view of the blazing speed to which cyber threats are growing, the urgent creation of reliable, dynamic, and intelligent technologies that help detect and counter malicious actions online is then the need of the day. The malicious URLs function as gateways that computers pass through, and this acts as a conduit for the distribution of malware, phishing campaigns, among other malicious activities. In this research, we propose an intelligent approach with a multi-layer fine-tuned LSTM, running RMSProp and Adam optimisers which are more powerful, employed for malicious URLs detection and classification. The LSTM structure is tuned by multiple levels to increase its ability to acquire information of systemic crawls and anomalies present on malicious URLs. This hybrid optimisation setup combines the support structure of Adam optimiser with the correctness notion of RMSprop made applicable through the determination of the high accuracy which reaches 0.967 (96.70%) and a strong performance.
  Keywords: long short-term memory; LSTM; URL; deep leaning; Adam; RMSprop; defacement; phishing; malware.
  DOI: 10.1504/IJESDF.2026.10068399
   
  
• Apple devices acquisition and forensic analysis  
  by Parag Shukla, Aditya Pratap 
  Abstract: Apple devices have become challenging to conduct forensic investigations primarily due to security mechanisms and the architecture used by Apple. Also, there has been limited exposure to the investigators as compared to the acquisition and analysis of Android & Windows devices. As part of this research paper, we will provide brief description of iOS and MAC internals including the internal filesystem, how the data is stored by applications, different boot modes and about the new security features implemented by Apple to encrypt the data stored within eMMC. In addition, we’ll look into different acquisition and analysis methods which are available for Apple devices that can be performed via open-source and commercial tools.
  Keywords: apple devices; acquisition; jailbreak; iOS; Mac; forensic investigation.
  DOI: 10.1504/IJESDF.2026.10068540
   
  
• Anomaly detection in digital forensics data using deep learning algorithm  
  by B. V. Santhosh Krishna , R. Pavithra , A. Seetha, Leo John Baptist Andrews 
  Abstract: As the internet of things (IoT) keeps evolving, it gets harder and harder to keep IoT networks and gadgets safe. Finding anomalies is a very important part of keeping the IoT secure. Machine learning (ML) techniques are a potential way to see strange things in the internet of things. The research field has a limited number of studies to find problems in the IoT. Different datasets from different places should be used with machine learning models to find things that do not fit. We used three well-known datasets for the comparison: IoT-23, NSL-KDD, and TON_IoT. Our data show that XGBoost did better than the SVM and the DCNN. It was right up to 99.98% of the time. It was also the best way to use computers because the model learned 717.75 times faster than the SVM and much faster than the DCNN.
  Keywords: anomaly detection; DCNN; internet of things; IoT; machine learning; ML; SVM; XGBoost.
  DOI: 10.1504/IJESDF.2026.10068541
   
  
• New security protocols of internet of things: improving the security in IoT  
  by N. Ashokkumar , M. Ananthi , M. Sri Geetha, V.P. Arul Kumar 
  Abstract: The industrial domains using IoT devices are expanding together with the market for IoT solutions and services. Security problems have been solved by researchers using machine learning to find intrusions at the network level. Using information from large source area datasets, transfer learning has been used to find dangerous traffic in internet of things systems that were not predicted. The problem is that most IoT devices work in small, different settings, like home networks, which makes it hard to pick good source domains for learning. This study gives us a plan for how to deal with this problem. Our suggested method says to choose a dataset as the source area for learning when it is hard to find a good dataset through pre-learning with transfer learning. Transfer learning is checked to see if it should be used so that the best way can be used in these situations.
  Keywords: internet of things; IoT; intrusion detection; transfer learning; wireless sensor networks; WSNs; security.
  DOI: 10.1504/IJESDF.2026.10068542
   
  
• Unveiling the power of machine learning: a deep dive into cutting-edge intrusion detection systems  
  by Afef Selmi , Salim El Khediri 
  Abstract: For cybersecurity, AI is one of the most important aspects as it detects variations, unusual activities, as well as potential threats. AI will strengthen the tools responsible for the harmonisation of protection, detection, and response, ensuring adaptability to the fast-changing IT landscape. AI technologies are coded into respective security tools to run the automation of replies among antivirus software, endpoint detection solutions, web safety channels, and firewalls. This research focuses on the role of machine learning (ML) methods in intrusion detection systems (IDS). Through extensive research on ML and DL applied to analyse flaws of logs, we identify gaps and innovative new approaches. First, the research notes how DL methods show positive results in dealing with the complexity of the logs. Through the careful assessment of ML-based approaches to IDS, AdaBoost, and comparison of performance indicators, we find both strong and weak aspects. We, however, want to offer experienced publications which will model the future development of cybersecurity. The evaluation in this paper goes beyond presenting the present stage of the employment of ML and DL in intrusion detection systems (IDS) to provide a guideline for future enhancement of IDS by way of new methodologies.
  Keywords: artificial intelligence; machine learning; cybersecurity; intrusion detection systems; IDS.
  DOI: 10.1504/IJESDF.2026.10068543
   
  
• Using machine learning for multiprocessor real-time systems scheduling analysis  
  by Lubna Sawaf, Walid Karamti 
  Abstract: Real-time systems (RTS) are increasingly prevalent, requiring multiprocessor architectures to meet strict timing constraints. Amidst the rise of machine learning applications, there’s a pressing need for nuanced schedulability analysis in RTS. This study focuses on evaluating the feasibility of scheduling real-time software periodic tasks on partitioned homogeneous multiprocessors. Introducing a hybrid model combining Decision Tree (DT) classification with Tabu Search (TS) optimisation, we aim to enhance precision in scheduling analysis. Additionally, our approach prioritises electronic security in realtime system analysis by streamlining SW/HW space exploration and reducing vulnerabilities. Evaluation using real satellite dataset validates the effectiveness of our model, underscoring the importance of security in modern RTS design.
  Keywords: real-time systems; RTS; scheduling analysis; tabu search; machine learning; decision tree; homogeneous multiprocessors.
  DOI: 10.1504/IJESDF.2026.10068575
   
  
• Digital forensics unleashed: Tether transactions on TRON blockchain dissected  
  by Borase Bhushan Gulabrao, Digvijaysinh Rathod, Nitin Sharma, Aishwarya Tiwari 
  Abstract: Stablecoins like USD Tether (USDT) have become popular among criminals due to their stability linked to the US dollar and anonymity. This paper explores the inner workings of Stablecoins, particularly USDT, to reveal associated risks and controversies. It also examines the TRON blockchain, which is popular platform for USDT transactions. The study includes practical experiments involving the transfer of USDT using hardware and desktop wallets. Comprehensive analyses of RAM, DISK, and network images are conducted to identify digital artefacts before and after transactions. A key finding is the concept of ‘digital artefacts preceding signatures’(DAPS), which are important keywords that help in detecting these artefacts. The research introduces a specialised tool hosted on GitHub, the DAPS extractor, to assist forensic investigators in finding and retrieving digital evidence related to USDT and TRON. The paper evaluates the effectiveness of this tool, demonstrating its potential to enhance forensic investigations significantly.
  Keywords: TRON; USD Tether; USDT; RAM; disk; public key address; transaction hash; blockchain.
  DOI: 10.1504/IJESDF.2026.10068620
   
  
• A hybrid deep learning method for URL spoofing in websites  
  by B. V. Santhosh Krishna , S. Vidhya , S. Krishnaveni, N. Ashokkumar  
  Abstract: In the 21st century, website uniform resource locator (URL) faking is still a way that phishing attacks are done. Hackers are still using URL faking to trick people who are not paying attention into giving out personal information on harmful websites. An important and well-known deep learning method is the convolutional neural network (CNN). Long-short-term memory (LSTM), on the other hand, has been used well in tough real-time situations because it can keep info for a long time. CNN and LSTM deep learning models are used together to see how well they can find fake website URLs. The goal is to use the best parts of both methods to create a more advanced faking URL detection system. We compared the suggested hybrid model to other models using two datasets. The UCL and PhishTank datasets were used to test the combined CNN-LSTM model, obtaining 98.9% and 96.8% respectively.
  Keywords: data collection; convolutional neural network; CNN; long-short-term memory; LSTM.
  DOI: 10.1504/IJESDF.2026.10068669
   
  
• Early identification and prediction of ransomware attacks in transactions  
  by Swagata Sarkar, G. Yasika , M. Ramya, S. Alagumuthu Krishnan 
  Abstract: Blockchain technology is one of the most promising technologies. It can manage safe and genuine remote healthcare data across several clinics. It is simple to get healthcare services remotely, without having to physically visit the hospital, in order to receive necessary exams and reports. Nevertheless, security and cyberattacks are now a part of the working environment for digital healthcare systems. Despite this, a considerable percentage of healthcare data transactions are prevented from occurring while they are being processed on the network due to ransomware attacks, which remain a sophisticated vulnerability in block chain technology. Consequently, blockchain technology will be able to identify ransomware attacks at the code, data, and service levels (RBEF). The simulation results show that, in comparison to other blockchain technologies that are effective against ransomware, the RBEF saves money spent on processing healthcare data by ten percent and shortens transaction times by four to ten minutes. The proposed system achieves an accuracy rate of 98.3%.
  Keywords: blockchain; RBEF; ransomware; delays; sandbox; static and dynamic analysis.
  DOI: 10.1504/IJESDF.2026.10068714
   
  
• A novel method for intrusions detection in IoT enabled environment  
  by Ravi Kumar Saidala, Surekha Y., Lalitha Kumari Gaddala, Anjaneyulu Kunchala, Ramakrishna Reddy Mule, Ravi Kumar Tirandasu 
  Abstract: One of the most significant study areas in recent years has been the Internet of Things. It is suggested to use a supervised machine learning intrusion detection system (IDS) to identify IoT attacks with a high detection accuracy of 99.99% and an MCC of 99.97%. Using the minimum-maximum normalization technique for feature scaling, an efficient intrusion detection system (IDS) for the Internet of Things (IoT) is built to prevent information leakage on the test set. Because of this, it is necessary to provide a greater contribution to this context for the Internet of Things environment by assessing various AI-based algorithms on datasets that are capable of properly capturing the various aspects of the environment. Not only that, but we also looked at the effects of various approaches for feature engineering, such as correlation analysis and information gain.
  Keywords: internet of things; IoT; machine learning; deep learning; network security.
  DOI: 10.1504/IJESDF.2026.10068803
   
  
• Enhanced intrusion detection in smart grids through integrated pre-processing and classification techniques  
  by J. Jeyasudha, K. Sasikala  
  Abstract: The term “smart grid” refers to an updated electrical grid infrastructure that combines conventional power supply methods with modern sensing, communication, and control technology. Ensuring the security and dependability of the contemporary electrical infrastructure in smart grids requires effective intrusion detection. Making sure the grid infrastructure is secure becomes increasingly important as SG technologies become popular. The identification and avoidance of potential threats and attacks in SG environments is a critical function of IDS. In SG, intrusion detection is essential to maintaining system security. Using a deep learning (DL) classification model in conjunction with sophisticated pre-processing and feature extraction techniques, this study investigates a novel method of intrusion detection in SG datasets. The recommended method uses CNN for classification, AE, ICA, and PCA for feature extraction. This study investigates how CNNs, t-SNE, and feature extraction increase intrusion detection dataset accuracy, precision, and recall. CNN with t-SNE and autoencoder has the highest accuracy (92 %), precision (0.89), and recall (0.87). This hybrid technique protects SG infrastructures from cyberattacks by increasing detection. We utilise Python and Jupyter Notebook.
  Keywords: smart grid; SG; intrusion detection systems; IDS; convolutional neural network; CNN; independent component analysis; ICA; feature extraction; cybersecurity measure; microgrids development; deep learning; DL; auto encoders; AE.
  DOI: 10.1504/IJESDF.2026.10068804
   
  
• Forensic analysis of privacy and anonymity focused operating systems: Tails OS, Whonix and Qubes OS  
  by Ravirajsinh Vaghela, Parag Shukla, Sanjeev Varma Ragula, Naveen Chaudhary, Smit Bhanushali 
  Abstract: In the present digital landscape, privacy and anonymity have become essential concerns for individuals and companies alike. As a result, many operating systems emerged to facilitate privacy and anonymity. While these operating systems offer major privacy and anonymity advantages, they also pose considerable hurdles for digital forensic investigations. The very characteristics that preserve user privacy can hinder forensic investigations, making it harder for investigators to retrieve and evaluate data. In this research, acquisition and analysis of memory dumps and unencrypted network packet headers of such privacy and anonymity centric operating systems, is performed and potential artefacts have been identified that OS leaves in RAM and network which can be used as potential evidence at the court of law.
  Keywords: digital forensics; tails OS; Whonix; Qubes OS; anonymous OS; privacy; anonymity.
  DOI: 10.1504/IJESDF.2027.10068805
   
  
• Deep learning-based digital image forgery detection system  
  by A. Raajya Vardhini, Ravi Kishore Veluri, T. Veena , S. Aswini 
  Abstract: Information forensics experts need to be able to tell the difference between real photos taken with a digital camera and computer-generated images made by advanced graphics rendering engines. This is done to find out where the pictures came from and make sure the scenes they show are real. This paper gives two easy-to-use but effective ways to improve classification success in harsh circumstances. Each of these methods is based on gathering more data and combining predictions for events happening in different parts of the world. We'll talk more about each of these methods below. Our method might be easier to understand and use than others because it doesn't need as many computer tools. They also do a good job of putting things into the right categories. The suggested methods work because they were tested on sets of computer graphics pictures made by four well-known and cutting-edge graphics rendering engines.
  Keywords: deep learning; digital image; detection system; image forensic techniques.
  DOI: 10.1504/IJESDF.2027.10068812
   
  
• A universal forensics approach based on steganographic models: image manipulation detection  
  by Xiaoyan Liu, Ling Yang, Long Liu 
  Abstract: As a result of the proliferation of tools for manipulating images, a growing number of people are discovering that it is simple to modify the content of images. GPNet is the solution that we proposed in this study to overcome this difficulty. Through the use of transformer and CNN in parallel, GPNet is able to construct global dependence and effectively collect low-level information. In addition, we develop a powerful fusion module that we call TcFusion. This module is capable of combining feature maps that were produced by both branches in an efficient manner. It was determined that the combination of ResNet-50, PSO, and SVM produced the best model for the CXR dataset, with an accuracy of 99.76%. An autoencoder, PSO, and KNN were used in conjunction with one another to achieve the best level of accuracy possible for the MRI dataset, which was 9.51%.
  Keywords: image manipulation localisation; long-range modelling; two-stream network; feature fusion.
  DOI: 10.1504/IJESDF.2027.10068886
   
  
• Securing the internet of things: navigating complex cybersecurity threats and strategies for IoT applications  
  by Syeda Nazia Ashraf, Raheel Siddiqi, Fayyaz Ali, Shafique Ahmed Awan, Irfan Ali Kandhro 
  Abstract: The internet of things (IoT) integrates diverse components including systems, applications, data storage, and services, creating potential vulnerabilities for cyber-attacks as they continually provide services within organisations. Presently, the proliferation of software piracy and malware attacks poses significant risks to IoT security, potentially resulting in the theft of crucial information and subsequent economic and reputational damages. The current research has focused on the internet of things (IoT) revolution, with security and privacy emerging as the main concerns because of its adoption in vital areas. The IoT application and innovation are rapidly increasing, providing a wide range of facilities and solutions for industries in the fields of e-health, smart living, e-transport, and e-manufacturing. In this context, manufacturers and customers are concerned about the growing trend of cyberattacks on systems infrastructure, which is exacerbated by innate vulnerabilities. This examines the IoT cybersecurity landscape within the IoT domain, highlighting its security challenges. Additionally, we explore essential security requirements and techniques to mitigate these challenges. Finally, blockchain technology is examined as a recommended solution to bolster IoT security.
  Keywords: security; network analysis; IOT; cyber security; attacks detection; smart IoT; anomalies.
  DOI: 10.1504/IJESDF.2026.10069193
   
  
• Security analysis of cyber threats using digital forensics: explainable artificial intelligence  
  by Yanwei Xu, Ye Huang, Juan Luo, Xueyong Wan 
  Abstract: Artificial intelligence (AI) is now used to make software better in many places. Two of them are science and health. You can read and write about how works on explainable AI (XAI). With XAI, you can discover which parts of an AI model make it work differently. CF makes it easy to split files but hard to join them back together. We should group file bits as XAI told us to. SIFT is a new way to assemble things that we show you. This helps SIFT find things in a small part of a file. It does this by giving a number to each bit. The LIME and SHAP feature importance value has a bottom number for each feature. A multilayer perception model is created and improved to make multinomial classification better. The SIFT method was tested with fifty kinds of files, for a total of 47,482 files.
  Keywords: feature selection: XAI – feature relevance: LDA-Gibbs model: LDA-Gibbs theme model: SIFT – system overview.
  DOI: 10.1504/IJESDF.2027.10069234
   
  
• Rootkit hidden process detection in cloud computing: data extraction at hypervisor-level  
  by Tushar A. Champaneria, A. Arul Oli , Sunita Sachin Dhotre, S.D. Prabu Ragavendiran, S. Srinivasan  
  Abstract: The underlying hardware resources can be visualized by using hypervisors, which also make it possible for many operating systems to operate concurrently on the same infrastructure. This is done via the use of hypervisors. Rootkits are able to get access to the hypervisor with the help of the fact that it is located in the software stack at a higher privilege level than the operating systems. The approach that we have created is very sensitive to performance, and it was built with the objective of identifying rootkits in hypervisors from System Management Mode (SMM) while concurrently making use of the capabilities of SMI Transfer Monitor (STM). The creation of this strategy was prompted by the discovery of rootkits as the driving force behind it. When compared to other rootkit detection methods like mark-based systems and equipment-based recognition approaches, this procedure's accuracy is excellent and its rootkit recognition time is rapid.
  Keywords: rootkit hidden; prototype design and implementation; design of EPA-RIMM-V; system management mode; SMM; SMI transfer monitor; STM.
  DOI: 10.1504/IJESDF.2027.10069242
   
  
• Approaches of critical infrastructure companies to recover from cyber-attack: insights from internal specialists and external information security auditors  
  by Iryna Leroy 
  Abstract: Companies operating in the PayTech and online e-commerce sectors play a crucial role in critical infrastructure, functioning within the dynamic digital landscape. This study focuses on the recovery process after cyber-attacks and examines the contrasting perspectives of internal and external professionals. The research reveals notable differences in the perceptions of recovery strategies between internal stakeholders such as investor relations, reputation management, and Chief Information Security Officers, representing companies belonging to critical infrastructure and external auditors, who provide just and emergency support and perform specific tasks. Importantly, the study underscores the current attitudes towards future information security strategies and their influence on the financial recovery and reputation of reliable companies following cyber incidents. This research contributes to the existing knowledge by shedding light on the perspectives of both a company's internal and external specialists involved in the recovery process and cyber resilience strategies in critical infrastructure sectors.
  Keywords: information security; information security assessment; digital; reputation management; cyber autonomy; cyber resilience.
  DOI: 10.1504/IJESDF.2027.10069372
   
  
• Digital forensic intervention in Android device privacy breach  
  by Santosh. M. Nandwana, Kiran Dodiya, Kapil Kumar 
  Abstract: In the modern virtual age, smartphones, particularly those jogging on Android, have become necessary for daily life, facilitating verbal exchange, enjoyment, and facts storage. With Android’s global dominance as a working machine, it is also a top target for cyber threats, primarily due to great privacy issues. This research, titled ‘Digital forensic intervention in Android device privacy breach’, is a crucial exploration of the intersection of digital forensics and Android safety. It aims to understand vulnerabilities in the Android platform that contribute to privacy breaches, verify the role of digital forensics in investigating those breaches, evaluate the effectiveness of current forensic systems, and communicate the criminal and ethical implications of digital forensic practices in protecting customer privacy. The literature review identifies OS fragmentation, app distribution methods, and permission versions of Android as the three main weaknesses. It also tackles the challenging circumstances of mobile tool forensics, specifically about maintaining data integrity and handling encryption. The analysis highlights the significance of robust virtual forensic techniques to lessen the impact of privacy breaches on Android devices, offering guidance to cybersecurity experts and legislators on enhancing the security and privacy of Android users in an increasingly virtualised world.
  Keywords: android security; digital forensic; privacy breach; OS vulnerabilities; mobile device forensic.
  DOI: 10.1504/IJESDF.2027.10069421
   
  
• A novel approach to enhance ATM cybersecurity: tailored YARA rules for ATM malware analysis  
  by Kiranbhai R. Dodiya, Kashyap Joshi, Kapil Kumar, Parvesh Sharma 
  Abstract: The growing ATM malware has begun to pose serious challenges for financial institutions to overcome. ATM malware has also adversely affected their working pace due to the changing technological environment. This study proposes a new method for detecting ATM malware based on custom YARA rules. We performed extensive behavioural analyses on diverse global ATM hardware/malware samples, which allowed us to identify unique malicious behaviours. We then employed these insights to formulate bespoke YARA rules for identifying risky behaviours related to ATM malware. We validated our approach on large datasets and found it accurate and robust in detecting malware. The custom YARA rules created in this work have a much higher detection rate and a lower false positive rate per detection than traditional detection techniques. They can enhance cybersecurity protections for financial institutions. The first step forward is to strengthen ATM security with the easy scalability that this research provides against advanced cyber threats to the banking infrastructure, along with helpful implementation for longer-term, permanent protection and stability.
  Keywords: ATM malware; YARA rules; behavioural analysis; cybersecurity; financial institutions.
  DOI: 10.1504/IJESDF.2027.10069494
   
  
• Advanced Android covert channel attacks with novel evasion methods  
  by Abhinav Shah, Digvijaysinh M. Rathod, Bharat Buddhadev, Jeet Rami 
  Abstract: In the field of secure communication, establishing concealed pathways is vital for maintaining confidentiality. This study introduces innovative approaches for covert communication through audio channels, employing advanced entropy encoding and dynamic strategies in Android. The proposed method explores creative means of embedding sensitive information (device ID, contact number, and SMS) within audio files to minimise detection. By integrating state-of-the-art entropy encoding, the system achieves effective data compression and encryption, ensuring resilience against unauthorised interception. Dynamic techniques are implemented to adapt to changing environmental conditions, enhancing the robustness of the covert communication system. The paper employs novel evasion methods and covert channels techniques, including Base64 encoding, fixed-length encoding, dynamic length encoding, and Huffman encoding in Android. Comprehensive experimental results demonstrate the effectiveness of the proposed methods in terms of covert channel capacity, reliability, and resistance to detection. The android application is developed using proposed methodology and security assessment is carried out. The paper concludes by introducing various entropy and dynamic techniques, comparing their results and highlighting outcomes in covert audio communication. Findings suggest that Base64 encoding, as an evasion technique, shows shorter encoding/decoding times, higher throughput, and increased bit-carrying capacity, enabling efficient transfer of sensitive data in an audio covert channel.
  Keywords: covert channel; cyber security; Android application.
  DOI: 10.1504/IJESDF.2026.10069654
   
  
• Revolutionising healthcare data exchange: a secure and patient-centric approach with blockchain technology  
  by R. Krishnamoorthy , K.P. Kaliyamurthie 
  Abstract: The sharing of healthcare data among institutions can be challenging due to potential incompatibility arising from heterogeneous data architectures. Additionally, inconsistent language usage in healthcare further complicates understanding. Despite potential agreement on structure and semantics, security and data consistency issues persist. Building up a uniform understanding record over an information sharing organise is troublesome, given the helplessness of centralised storehouses and specialist suppliers to cyberattacks. This paper proposes a Blockchain-based method for patient data exchange, relying on network agreement instead of a single, centralised source of trust. The primary objective is to safely and successfully exchange medical records within a data sharing network, emphasising uniformity, accessibility across institutions, and strong patient-dictated access restrictions. To enhance data utility and patient care, sharing data is crucial, provided in a comprehensible form for all relevant stakeholders to understand its structure and significance.
  Keywords: healthcare data; compatibility; data consistency; cyberattacks; data sharing network; uniformity; accessibility.
  DOI: 10.1504/IJESDF.2027.10069790
   
  
• Enhanced facial recognition of criminal identification system using machine learning approaches  
  by A. Jency, K.S. Thirunavukkarasu 
  Abstract: Criminal identification using machine learning has gained significant attention across various applications due to its potential benefits in fields such as security and personalisation. This research aims to develop accurate and efficient criminal identification systems by leveraging machine learning techniques. The primary objectives of this study include prediction, classification, access control, and personalisation in the context of criminal identification. Specifically, the focus is on prediction, using criminal identification convolutional neural network (CNN) algorithms to predict specific attributes such as age, gender, emotional state, and potential health conditions. By analysing facial features and expressions, machine learning models can offer valuable insights and predictive capabilities. This research introduces an innovative approach that combines shallow convolutional neural networks (SCNNs) with a local search strategy. This combination is designed to enhance the efficiency and accuracy of criminal identification by utilising local features, optimising model performance, and reducing computational complexity. The proposed approach is tested on various datasets, showing significant improvements in the accuracy of predictions for attributes such as age, gender, and emotional states. The shallow CNN-based local search effectively captures subtle facial features, thereby contributing to the precision of attribute predictions.
  Keywords: criminal identification; deep learning models; enhanced facial recognition; facial feature analysis; image classification techniques; machine learning algorithms; predictive analytics; shallow convolutional neural networks; SCNNs.
  DOI: 10.1504/IJESDF.2027.10069971
   
  
• Enhancing fake news detection using light gradient boosting machine and term frequency-inverse document frequency-based algorithms  
  by Ravi Sheth, Chandresh Parekha 
  Abstract: The internet and social media have transformed the dissemination of news, but also facilitated the dissemination of false information. Social media managers create and disseminate enormous amounts of information, some of which is false and unrelated to reality. This toxic disinformation has inflicted irreparable damage on societies, especially in times of crisis, such as terrorist attacks and natural disasters. To counter this, there is a need to rapidly detect rumours. Fake news identification is an important research area, as increasingly advanced communications technology and social media pose new challenges. This paper describes a Term Frequency-Inverse Document Frequency (TF-IDF) technology implementation for processing text attributes and determining whether news is real or fake. High accuracy is achieved using Light Gradient Boosting Machine (LightGBM) produced a very high accuracy of 99.84%. This research contributes to the design of efficient fake news detection techniques, solving an urgent problem in the current digital environment.
  Keywords: fake news; LightGBM; TF-IDF; machine learning; classification.
  DOI: 10.1504/IJESDF.2027.10070133
   
  
• Application of artificial intelligence: methods to detect the image changes in social media  
  by Yan Gao 
  Abstract: Active and passive methods are the two basic techniques to image modification detection that are described in the existing body of literature. In contrast to passive approaches, active techniques are proactive in nature, since they embed structures into photographs in order to enable future authenticity verification. The outputs from various networks are then fused together via the concatenation in order to determine whether or not the picture has been manipulated. This results in a complete detection framework that is more effective than the individual approaches that make up the framework. Our study presents a one-of-a-kind dataset that is the result of the combination of four datasets that are freely accessible to the public. This set of data includes photos that have been changed naturally and are very close to what happened in real life. It gets better at generalising across a lot of different ways of manipulating data.
  Keywords: digital image forensics; convolutional neural network; CNN; deep learning.
  DOI: 10.1504/IJESDF.2027.10070134
   
  
• Modified hybrid deep learning digital models with hierarchical-attention network models for legal judgement predictions  
  by M. Jaiikanth Manivel, G. Aswathy Prakash 
  Abstract: Computer-assisted decision-making in legal judgement forecasts has gained popularity due to the development of big data and AI technology. Some fundamental components of conventional judgement prediction systems are classification and feature modelling methods. However, feature models need significant specialised expertise and hand annotation labour. This paper uses the supreme court judgement prediction dataset as input data, which is pre-processed using stages like tokenisation, stop word removal, stemming, and lemmatisation mechanisms. Hence, the required features were extracted by pre-processed data using BoW, POS tagging, and TF-IDF. The word embedding is performed using the HAN model. Then, the judgement prediction is done using the hybrid deep learning model called H-Bi-LSTM-CNN. There are two main stages: the testing and training stages; based on training data, the testing is performed. The proposed models performance is then contrasted with the already-in-use methods using performance measures. Research demonstrates that our technique is more accurate than the existing technique for judgement prediction, with a 96.96% accuracy rate.
  Keywords: hierarchical-attention network models; judgement predictions; stemming and tokenisation; word embedding; modified hybrid deep learning digital models; lemmatisation; stop word removal.
  DOI: 10.1504/IJESDF.2027.10070203
   
  
• An enhanced model of secured data transmission between IoT and the cloud  
  by Shatakshi Kokate, Urmila Shrawankar 
  Abstract: The rapid increase in IoT devices has led to a surge in data traffic, making secure and efficient data transmission a critical challenge. The sensitive nature of IoT-generated data necessitates robust security measures to prevent data loss during transmission between IoT devices, the cloud, and end users. Existing solutions face significant drawbacks, including data leakage, compromised security, confidentiality issues, bandwidth constraints, high latency, and vulnerability to single points of failure. These vulnerabilities can be exploited by malicious agents to disrupt IoT networks. Blockchain technology, with its immutable, distributed, transparent, and secure ledger, complements fog computing by strengthening overall security. The proposed MedFogChain model integrates these two technologies to improve security, resource utilisation, scalability, system performance, and reduce transmission overhead and latency. This model proves particularly effective in the healthcare domain, where data sensitivity and the need for rapid, secure data transmission during emergencies are paramount.
  Keywords: data transmission; security; internet of things; IoT; fog computing; blockchain; cloud.
  DOI: 10.1504/IJESDF.2026.10070242
   
  
• Digital forensic analysis of gaming and social metaverse platforms  
  by Sumaya Mohammad Alshokeeran, Shema Mohammed Alenezi, Sultan Meshal Althaqeel, Kyounggon Kim, Sundaresan Ramachandran 
  Abstract: The metaverse has gained widespread popularity with gaming and social platforms, enabling users to interact through avatars and digital assets. While offering new opportunities, it also presents complex digital forensic challenges. This research develops a specialised methodology for digital forensic investigations in metaverse platforms, considering their unique characteristics. The study focuses on extracting forensic artefacts from virtual reality headsets (Meta Quest 2), Android phone, Windows computers, and cloud-based acquisitions from Roblox, Rec Room, Second Life, and Meta Horizons World social metaverse platforms. The methodology includes preparation, where the investigation scope is defined, platforms are studied, and data collection strategies are planned. In the collection phase, forensic tools acquire digital evidence from local storage, cloud services, and system logs. The analysis phase examines extracted data to identify patterns, user behaviours, and forensic evidence. Finally, the presentation and documentation phase compiles findings into a structured report, ensuring legal admissibility. The research successfully extracted key digital artefacts, including login records, chat logs, and virtual asset transactions. It highlights challenges such as the lack of standardised forensic procedures and encrypted system complexities, emphasising the need for advanced forensic tools tailored to virtual reality investigations.
  Keywords: metaverse; augmented reality; virtual reality; digital forensics.
  DOI: 10.1504/IJESDF.2027.10070630
   
  
• Detection and enhanced security against cyberbullying on social media using a hybrid deep learning framework  
  by Moushmee Milind Kuri, Ganesh R. Pathak 
  Abstract: Cyberbullying on social media is a widespread issue, prompting the need for an optimised hybrid deep learning (DL) framework for effective detection and prevention. Due to its complex and subjective nature, cyberbullying is difficult to label for training machine learning models, and manual annotation can be time-consuming and biased. This research develops a novel DL framework for detecting cyberbullying by pre-processing textual data through cleaning, tokenisation, and leveraging advanced techniques like BERT and Roberta for feature extraction. Roberta, with its refined training approach, outperforms BERT in sentiment analysis tasks. Findings reveal that non-bullying tweets have 98% mention levels while bullying tweets show consistently elevated levels. By combining BERT/Roberta for sentiment analysis, BIGRU networks for subtle cues, and CNNs for keyword identification, the framework enhances cyberbullying detection. Future applications include real-time monitoring, preventive measures, and further advancements in natural language understanding for social media interactions, promoting safer online environments.
  Keywords: hybrid deep learning framework; cyberbullying; social media platform; robustly optimised BERT pertaining approach; convolutional neural networks; CNNs; natural language processing; NLP.
  DOI: 10.1504/IJESDF.2026.10071010
   
  
• Cryptanalysis of cryptography failures and solutions to cyber-attacks  
  by Dhairya J. Vyas, Milind Shah 
  Abstract: In the digital age, safeguarding sensitive information against cyber threats is paramount, and cryptography plays a crucial role in this endeavor. However, failures in cryptographic systems can leave data vulnerable to sophisticated cyber-attacks. This study provides a comparative analysis of various solutions addressing cryptography failures that lead to such vulnerabilities. By examining different cryptographic algorithms and their susceptibility to attacks, the research highlights common failure points and evaluates the effectiveness of existing countermeasures. Through a thorough review of recent advancements and methodologies in cryptanalysis, the study identifies key areas where cryptographic practices are often compromised. The analysis not only underscores the importance of robust cryptographic frameworks but also offers insights into improving security protocols to prevent potential breaches. By drawing comparisons across multiple solutions, this research aims to enhance understanding and resilience in the face of evolving cyber threats, ultimately contributing to stronger and more secure digital communications.
  Keywords: security analysis; cryptographic system; advanced encryption standard; Rivest-Shamir-Adleman; electronic codebook; cipher block chaining; counter mode.
  DOI: 10.1504/IJESDF.2027.10071950
   
  
• An investigative strategy for the iPhone forced entry zero-click exploit for mobile espionage  
  by Kritarth Jhala, Nilay Mistry, Naveen Chaudhary 
  Abstract: While the majority of mobile malware attacks have historically targeted Android-based smartphones, recent attention has increasingly focused on Apple iOS devices. This study focuses on analysing iOS-based malware, addressing key stages of the attack lifecycle, including exploitation, detection, propagation, infection, and analysis. It uses hybrid analysis techniques and malware samples from social media and online banking applications. The analysis reveals a comprehensive approach to detecting malware evolution and presents a robust model for identifying future iOS-based attacks. Dynamic analysis is essential across infection, activation, payload delivery, operational algorithms, and propagation. By analysing phylogenetic relationships between different malware strains, the study predicts how new attack variants may evolve, strengthening the ability to proactively counter future threats.
  Keywords: mobile forensics; mobile malware analysis; mobile zero day attack; iOS Malware analysis.
  DOI: 10.1504/IJESDF.2027.10072148
   
  
• Balancing privacy rights and AI threats: analysis of the European Court of Human Rights case law  
  by Moza Adel Bin Tamim, Jamal Barafi 
  Abstract: The right to privacy is perceived and conceived in various ways, which creates a statute of ambiguity towards the concept of privacy. With the advancement of technology and the evolvement of artificial intelligence that is gradually taking over the human role in the legal sector, the notion of privacy needs further articulation. The prevalence of artificial intelligence will lead to serious privacy violations; therefore, it is necessary to establish legal frameworks and deal with potential issues to prevent these incidents and protect individuals' right to privacy from the risks posed by AI. Despite the European Convention on Human Rights acknowledging the right to privacy and emphasizing its protection, the results of relevant cases say otherwise. Cases examined by the European Court of Human Rights all shared the same fate, a precise innuendo that the legal system needs powerful grounds to build cases on when the subject is relevant.
  Keywords: right to privacy; artificial intelligence; personal data; the ECtHR; the ECHR; GDPT.
  DOI: 10.1504/IJESDF.2027.10072325
   
  
• Artificial intelligence empowered quantum-resistant cryptographic algorithms in information security of WSN for future-proof systems  
  by V. Bharathi , Inderpreet Kaur, S.N Manoharan, M. Kasiselvanathan , Kiran Sree Pokkuluri, R. Sivakumar  
  Abstract: The security of today's encryption methods is under serious threat. As quantum computing becomes more complex, public key scripts that support many security policies may be undermined by the development of quantum computers. This research proposes a new artificial intelligence (AI)-based system to improve the efficiency and security of QRCA in sensor networks (WSNs). Strong protection against quantum attacks is provided by a guidance technology called AI-QRCA. -WSN which combines the new QRCA with AI technology. This results in secure and efficient encryption and decryption. In this study an efficient encryption and decryption algorithm is proposed using Chameleon genetic algorithm. AI can also deploy a regular WSN security monitoring system by monitoring data traffic. The proposed method uses machine learning algorithms to reduce security risks and identify potential threats. To secure WSNs in the age of quantum computing, make use of the QRCA AI-QRCA-WSN Framework and AI's integrated capabilities.
  Keywords: artificial intelligence; AI; quantum-resistant cryptographic algorithms; QRCA; wireless sensor networks; WSNs; information security; cybersecurity; quantum computing; post-quantum cryptography.
  DOI: 10.1504/IJESDF.2027.10072341
   
  
• Intelligent signal processing and fuzzy system identification: research on communication security and fault detection strategies  
  by Lei Wang 
  Abstract: Automation, reliability, and clever control with low delay are all important parts of manufacturing and industry systems. With the help of the Industrial Internet of Things (IIoT), industrial systems may get a chance when it comes to safety, efficiency, and production. There is a pressing need to address hardware issues and malfunctions in the context of the Internet of Things (IoT). Accidents and financial losses may happen, production may suffer, and workers may be called in. Edge computing is used by the DASIF framework to run machine learning models that are very accurate and have low delay. The alert state is a first-stage counter measure meant to make contact inside the network more reliable. Multiple-path transmission and data backup are used together to make this possible. Both a replica of the IIoT network and a real-world petroleum information were used to figure out how well DASIF worked.
  Keywords: industrial internet of things; IIoT; machine learning; edge computing; internet of things; IoT.
  DOI: 10.1504/IJESDF.2026.10072881
   
  
• Tort liability for online publishers  
  by Sohib Al-Shurman, Ahmed Al-Bnian, Mosab Al-Shurman 
  Abstract: The liability of online publishers has received great consideration as it is closely linked to freedom of expression and respect for the privacy of others, both of which are guaranteed by the Constitution. Unmoderated online publishing may violate individuals' privacy, defame them, or promote and spread ideas that harm society. This research aims to examine the civil liability of online publishers that require compensation for damages to third parties and the adequacy of the general rules addressing the liability of the online publisher for the published content. The study reached a set of conclusions and recommendations, the most important of which is that the Jordanian legislator should include a text in the Electronic Crimes Law and the Printing and Publishing Law that precisely defines the liability of the online publisher for the content he created or users uploaded.
  Keywords: online publisher; e-publishing offence; tort liability; publishing methods; civil liability.
  DOI: 10.1504/IJESDF.2027.10072928
   
  
• A hybrid machine learning approach to improve IoT forensics for the identification of attacks in the IoT environment  
  by F. Mashiya Afroze , V. Poornima  
  Abstract: Data is exchanged between network-connected devices due to the rapid growth of the internet of things (IoT). The widespread interconnectedness of IoT devices causes various security issues. The rapid growth of IoT devices brings many benefits, but also new security and forensics issues. Due to the massive amount of data generated by the billions of IoT devices, digital investigators and practitioners confront major challenges when investigating cybercrimes in a timely and forensically sound manner. The project aims to establish a framework for forensic investigation on resource-constrained IoT utilising forensic technology and machine learning to detect various assaults. This study tests DNNs with IoTF to detect assaults using operating system logs. Operating system operation characteristics are listed in a dataset. The DNN and classification training model are developed using these data. DNN training parameters might be challenging to determine. Traditional training methods delay convergence and stagnate at local minima. This study recommends training a DNN, SSOA with optimal settings. SSOA-DNN is compared to KNN, RF, SVM, DT, LDA, and NB ML classifiers. The following metrics assess ML model effectiveness: four factors: accuracy, precision, recall, and F-Measure. The results reveal that SSOA-DNN surpasses other ML classification algorithms in IoT Analysis with 96.37% accuracy.
  Keywords: internet of things forensics; IoTF; machine learning; attack prediction; deep neural networks; DNNs; Salps swarm optimisation algorithm; SOA; machine learning; IoT devices.
  DOI: 10.1504/IJESDF.2027.10073363
   
  
• Advanced forensic analysis of Tails operating system and its implication to cybercrime in deep and dark web  
  by Ngaira Mandela, Amir Aboubakr Shaker Mahmoud, Animesh Kumar Agrawal, Nilay R. Mistry 
  Abstract: The Tails operating system, renowned for its emphasis on privacy and anonymity, has become a preferred choice for individuals seeking to safeguard their online activities. Tails OS design centres around providing a secure environment that leaves minimal traces, thereby attracting privacy-conscious users. However, this very secure design also entices cyber criminals operating in the digital landscape, to use Tails to perpetrate illicit activities, creating new challenges for digital forensics practitioners in their pursuit of extracting evidence. This paper conducts an exhaustive forensics of Tails operating system, aiming to uncover the digital remnants left behind during its utilisation. By analysing the RAM, network, disk, and virtualisation, employing a range of activities and forensic tools, many artefacts are unearthed that provide insight into user interactions within the Tails environment. This research contributes to our understanding of the interplay between privacy preservation and digital evidence recovery, shedding light on the complexities of investigating a privacy-focused operating system like Tails.
  Keywords: Tails operating system; amnesic incognito live system; privacy-focused OS; anonymity; cybercrime; digital forensics; deep and dark web.
  DOI: 10.1504/IJESDF.2025.10061945
   
  
• Tweet sentiment analysis: leveraging weighted word embedding to evaluate criticality and security of events  
  by Irfan Ali Kandhro, Ali Orangzeb Panhwar, Asma Touqir, Asadullah Burdi 
  Abstract: Currently, one of the most important sectors globally is social media. It has been reported that about 74% of people on the planet utilise social media. This has spurred a lot of social media study. Sentiment analysis of real-time social media data for security purposes is one such helpful application. In this paper, we present a weighted word embedding approach for classification of tweets related to education. The proposed approach utilised weighted GloVe word embedding with a sequential neural network (SNN) architecture to enhance the accuracy of sentiment classification. Moreover, this paper incorporates the weighted word embeddings, assigning varying weights based on their relevance to sentiment analysis. In the empirical analysis, we evaluate the predictive performance of different word embedding schemes, including SNN word embedding, GloVe, and DIM, with deep neural network architectures. The performance of the proposed approach is measured using precision, recall, F1 score and accuracy. The experimental results showcase promising performance, effectively capturing valuable insights into public opinion during the pandemic. The combination of TF-IDF weighted GloVe word embedding and the SNN architecture proves to be a robust approach for sentiment analysis, providing accurate sentiment classification reviews on Twitter.
  Keywords: security; alerts; sentiment analysis; education tweets; GloVe; word embedding; deep learning.
  DOI: 10.1504/IJESDF.2025.10069807
   
  
• A proposed framework for the detection of cyber threats using open-source intelligence tools in real-time  
  by Ravi Sheth, Chandresh Parekha 
  Abstract: In today's fast-changing digital world, getting accurate and relevant information is crucial for cybersecurity, competitive analysis, and research. With open-source intelligence (OSINT) technology, social media and digital platforms have become significant data sources. Our research will compare different OSINT systems for data collection. We'll examine various OSINT tools, categorise them based on functionality, usability, and effectiveness, including web scrapers, social media analytics, domain analysers, and search engines. Each tool will undergo evaluation based on data collection capabilities, result accuracy, module availability, and user-friendliness. It aims to assist individuals – be they sufferers, academics, or practitioners – in making informed decisions when selecting open-source intelligence (OSINT) tools for information extraction. We have provided a detailed comparison of their capabilities and limitations, along with ethical considerations, for using OSINT tools effectively and responsibly in various applications and the proposed cyber threat alert system in this paper.
  Keywords: template OSINT; information gathering; web scanner; search engine; cyber security.
  DOI: 10.1504/IJESDF.2025.10064154
   
  
• Data security by means of cryptography and image processing – a deep learning based method  
  by Pooja Sharma, Anurag Patel 
  Abstract: Traditional picture encryption methods make use of rounds of diffusion and confusion in order to create an explicit trade-off between the degree of security and the length of time it takes to decode an image. Deep learning is a method that has the potential to be used in the production of responses that are acceptable for present problems in photo encryption systems. We explore the advantages and disadvantages of each of these research approaches and develop conclusions that are relevant to our findings. In the second step of the process, the various methods are compared and analysed in terms of the cryptographic properties of the recovered photos and the quality of the cipher images that are produced. Finally, conclusions are formed from comparing and analysing the deep learning technique in end-to-end encryption and decryption systems. These findings serve as a platform for future research.
  Keywords: deep learning; image encryption; cryptographic attacks; encryption keys; style transfer.
  DOI: 10.1504/IJESDF.2025.10065036
   
  
• Big data security using homomorphic encryption: application in finance  
  by Tingxin Jiang 
  Abstract: Techniques that protect privacy make it possible to utilise private information without compromising the confidentiality of the information. The use of homomorphic encryption algorithms offers unique ways that make it possible to do computations on encrypted data while still preserving the secrecy of the information that is being protected. The use of homomorphic encryption methods is also discussed in relation to a security framework for big data analysis that is designed to protect individuals' privacy. After that, we will proceed to provide a comparison of the properties that have been discovered in relation to the common homomorphic encryption tools that are now accessible. Analysis is performed on the outcomes of the installation of a variety of different homomorphic encryption toolkits, and a comparison is made between the various performances of each of these kits. The proposed model has an accuracy rate of about 93.75%.
  Keywords: big data; encryption algorithms; homomorphic encryption; privacy preserving; machine learning.
  DOI: 10.1504/IJESDF.2025.10066462
   
  
• AI chatbots: security and privacy challenges  
  by Manju Lata, Vikas Kumar 
  Abstract: The increasing use of artificial intelligence (AI) chatbots in different domains has surely developed the competence, customer capabilities and engagement. On the other hand, the AI chatbots regularly handle sensitive data, making them striking objects on behalf of the malicious actors. Present work describes the prominent security and privacy issues related to the use of AI chatbots, in order to plan the mitigation strategies. An all-inclusive approach is required to handle the security and privacy issues to ensure the transparent data practices, robust security techniques, ethical improvement procedures, and compliance with significant regulations. Since the AI chatbots continue to advance at a faster pace, a pre-emptive approach is crucial for reliable and protected amalgamation into sustainable digital lives. By probing and mitigating security and privacy challenges, this paper contributes to the reliable development and deployment of AI chatbots, with adoption of more secure and trustworthy conversational AI environment.
  Keywords: AI chatbots; security; privacy; data security; mitigation strategies; challenges; AI environment.
  DOI: 10.1504/IJESDF.2025.10065035
   
  
• Exploring novel encryption approaches for safeguarding heterogeneous data  
  by Dhruvi Karansinh Zala, Mohammad Akram Khan 
  Abstract: In today's global landscape, data security faces complex challenges due to diverse data forms and transmission structures. This study explores safeguarding data through unconventional encryption methods, stressing the need for adaptable techniques. It begins by assessing traditional encryption's performance with diverse data types, highlighting security gaps in uniform encryption. The research advocates for encryption that handles data complexities, proposing strategies aligned with heterogeneous data's variability. Practical tests and simulations inform these novel encryption algorithms, aiding in advancing security protocols for diverse scenarios. Ultimately, this research contributes to evolving data security in heterogeneous environments, emphasising the importance of robust encryption in managing varied data formats and structures in an interconnected world.
  Keywords: encryption; data security; image; audio; video; text; heterogeneous data; privacy preservation; hybrid encryption; multi-level encryption.
  DOI: 10.1504/IJESDF.2025.10064447