See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/382947230 Bridging the Digital Divide: Empowering Nigerian Universities through Technological Advancements in Academic Libraries Conference Paper · May 2024 CITATIONS 0 READS 3,879 3 authors, including: Bolaji David Oladokun Federal University of Technology, Akwa Ibom 150 PUBLICATIONS   585 CITATIONS    SEE PROFILE Tella Adeyinka University of Ilorin 167 PUBLICATIONS   3,195 CITATIONS    SEE PROFILE All content following this page was uploaded by Bolaji David Oladokun on 08 August 2024. The user has requested enhancement of the downloaded file. https://www.researchgate.net/publication/382947230_Bridging_the_Digital_Divide_Empowering_Nigerian_Universities_through_Technological_Advancements_in_Academic_Libraries?enrichId=rgreq-4aa462bce4b60bd80444698a027a4b4b-XXX&enrichSource=Y292ZXJQYWdlOzM4Mjk0NzIzMDtBUzoxMTQzMTI4MTI3MDUxMzIxMUAxNzIzMTAyMzUwNjI4&el=1_x_2&_esc=publicationCoverPdf https://www.researchgate.net/publication/382947230_Bridging_the_Digital_Divide_Empowering_Nigerian_Universities_through_Technological_Advancements_in_Academic_Libraries?enrichId=rgreq-4aa462bce4b60bd80444698a027a4b4b-XXX&enrichSource=Y292ZXJQYWdlOzM4Mjk0NzIzMDtBUzoxMTQzMTI4MTI3MDUxMzIxMUAxNzIzMTAyMzUwNjI4&el=1_x_3&_esc=publicationCoverPdf https://www.researchgate.net/?enrichId=rgreq-4aa462bce4b60bd80444698a027a4b4b-XXX&enrichSource=Y292ZXJQYWdlOzM4Mjk0NzIzMDtBUzoxMTQzMTI4MTI3MDUxMzIxMUAxNzIzMTAyMzUwNjI4&el=1_x_1&_esc=publicationCoverPdf https://www.researchgate.net/profile/Bolaji-Oladokun?enrichId=rgreq-4aa462bce4b60bd80444698a027a4b4b-XXX&enrichSource=Y292ZXJQYWdlOzM4Mjk0NzIzMDtBUzoxMTQzMTI4MTI3MDUxMzIxMUAxNzIzMTAyMzUwNjI4&el=1_x_4&_esc=publicationCoverPdf https://www.researchgate.net/profile/Bolaji-Oladokun?enrichId=rgreq-4aa462bce4b60bd80444698a027a4b4b-XXX&enrichSource=Y292ZXJQYWdlOzM4Mjk0NzIzMDtBUzoxMTQzMTI4MTI3MDUxMzIxMUAxNzIzMTAyMzUwNjI4&el=1_x_5&_esc=publicationCoverPdf https://www.researchgate.net/profile/Bolaji-Oladokun?enrichId=rgreq-4aa462bce4b60bd80444698a027a4b4b-XXX&enrichSource=Y292ZXJQYWdlOzM4Mjk0NzIzMDtBUzoxMTQzMTI4MTI3MDUxMzIxMUAxNzIzMTAyMzUwNjI4&el=1_x_7&_esc=publicationCoverPdf https://www.researchgate.net/profile/Tella-Adeyinka-2?enrichId=rgreq-4aa462bce4b60bd80444698a027a4b4b-XXX&enrichSource=Y292ZXJQYWdlOzM4Mjk0NzIzMDtBUzoxMTQzMTI4MTI3MDUxMzIxMUAxNzIzMTAyMzUwNjI4&el=1_x_4&_esc=publicationCoverPdf https://www.researchgate.net/profile/Tella-Adeyinka-2?enrichId=rgreq-4aa462bce4b60bd80444698a027a4b4b-XXX&enrichSource=Y292ZXJQYWdlOzM4Mjk0NzIzMDtBUzoxMTQzMTI4MTI3MDUxMzIxMUAxNzIzMTAyMzUwNjI4&el=1_x_5&_esc=publicationCoverPdf https://www.researchgate.net/institution/University-of-Ilorin?enrichId=rgreq-4aa462bce4b60bd80444698a027a4b4b-XXX&enrichSource=Y292ZXJQYWdlOzM4Mjk0NzIzMDtBUzoxMTQzMTI4MTI3MDUxMzIxMUAxNzIzMTAyMzUwNjI4&el=1_x_6&_esc=publicationCoverPdf https://www.researchgate.net/profile/Tella-Adeyinka-2?enrichId=rgreq-4aa462bce4b60bd80444698a027a4b4b-XXX&enrichSource=Y292ZXJQYWdlOzM4Mjk0NzIzMDtBUzoxMTQzMTI4MTI3MDUxMzIxMUAxNzIzMTAyMzUwNjI4&el=1_x_7&_esc=publicationCoverPdf https://www.researchgate.net/profile/Bolaji-Oladokun?enrichId=rgreq-4aa462bce4b60bd80444698a027a4b4b-XXX&enrichSource=Y292ZXJQYWdlOzM4Mjk0NzIzMDtBUzoxMTQzMTI4MTI3MDUxMzIxMUAxNzIzMTAyMzUwNjI4&el=1_x_10&_esc=publicationCoverPdf Copyright ©2024 Online ISSN: 2645-2960 Print ISSN: 2141-3959 Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) Copyright ©2024 ii Online ISSN: 2645-2960 Print ISSN: 2141-3959 Conference Organising Committee Prof A. L. Azeez Dean Prof. Oluwakemi C. Abikoye Chairperson Prof. R. G. Jimoh Co-chair1 Dr. A. O. Bajeh Co-chair2 Mrs. Rotimi T. Bamidele Secretary Subcommittee Chair Publication and Review Dr. A. O. Bajeh Finance Dr. Rafiat A Oyekunle Logistics Dr. L.K. Mustapha Technical Programme Committee Dr. J.B. Awotunde Registration Dr. S.O. Onidare Hotel & Accommodation Dr. N. A. Balogun Publicity Dr. A.O. Babatunde Website Dr. I.D. Oladipo Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) Copyright ©2024 iii Online ISSN: 2645-2960 Print ISSN: 2141-3959 Preface This book contains the Proceedings of research articles presented at the 3rd International Conference on ICT for National Development and Its Sustainability held from Tuesday 21st – Friday 4th May 2024 at the University of Ilorin, Ilorin, Kwara State, Nigeria. The conference was organized by the Faculty of Communication and Information Sciences, University of Ilorin. Over 50 research manuscripts were submitted for the conference. A keynote speech and two lead papers were presented at the conference. Apart from Nigeria, submissions were received from countries such as Niger, Malaysia and South Africa. All manuscripts were subjected to a double-blind peer review process. Before the review, every manuscript was subject to Turnitin plagiarism checker. Manuscripts with a similarity index of not more than 25% were processed for review. The articles which appear in this Proceedings are revised versions of the accepted manuscripts after the review process. I wish to appreciate all who contributed in one form or the other towards making this Proceedings a success. Thank you. Prof. A. L. Azeez Dean Faculty of Communication and Information Sciences University of Ilorin Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) Copyright ©2024 iv Online ISSN: 2645-2960 Print ISSN: 2141-3959 Table of Contents S/N Article Page 1 An Improved Machine Learning-Based Framework for Fake News Detection Oluwakemi Christaina Abikoye, Taofeek Akintunde Abdulsalam 1 2 Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Aminat ADEBAYO, Abidemi Emmanuel ADENIYI, Jumoke AJAO, Rafiu ISIAKA, Kazeem GBOLAGADE, Sulaiman ABDULSALAM 15 3 Harnessing Artificial Intelligence and Big Data to Enhance Sustainable Urban Planning in Nigeria Lukman Adegboyega ABIOYE, Omolara Oluwatoyin IDEHENRE 35 4 Adopting Artificial Intelligence in a Blended Learning Environment: Issues and Prospects in Nigeria Kennedy Arebamen Eiriemiokhale, Femi Samson Oyedepo, Abdulakeem Sodeeq Sulyman 49 5 A Review of Studies on Elliptic Curve Cryptography Schemes: The Ways Forward A.O Bajeh, A. O. Ameen, H.O. Adeleke, B.T. Adeyemi 66 6 Development of Hybridized Security Model for internet of Things Devices using Challenge Handshake Authentication Protocol and Rijndael Algorithm Ayannusi Adebowale Olufemi, Dayo Reuben Aremu 80 7 Comparative Analysis of Credit Scoring Risk in Commercial Banks Using Data Mining Techniques I.S. Olatinwo, M.J. Abdulrafiu, T.O. Adebakin, R.O. Obisesan 99 8 A Systematic Literature Review of Machine Learning and AutoML In Software Effort Estimation Shakirat Aderonke SALIHU, Khadijat Bola SALIU, Olusola Ayodele OWOYEMI 145 9 Performance Analysis of Some Machine Learning Algorithms in Prediction of Heart Disease Shakirat Aderonke SALIHU, Olusola Ayodele OWOYEMI, Khadijat Bola SALIU 169 10 Investigating the Impact of Artificial Intelligence-based Chatbots on the Critical Thinking of University of Ilorin Students Latifat Ahuoiza Ibrahim, Oluwabunmi Titilope Oladele, Ngozi Clara Akawo 182 11 Impact of Artificial Intelligence on Journalism Ethics Muhammed Shaibu ONAKPA, Fati ALIYU-OHIARE, Sharon Umola AMANA 195 Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) Copyright ©2024 v Online ISSN: 2645-2960 Print ISSN: 2141-3959 12 Digital Technologies’ Adoption by Academic Libraries for Sustainable Development in an Evolving Technological Age in Nigeria Priscilla Abike AGBETUYI, Gladys Titilayo OLORUNYOMI 211 13 The Role of Artificial Intelligence (AI) in Advertising: A Philosophical Approach Tajudeen WONUOLA, Ganiyu Olalekan AKASHORO 226 14 Adoption of AI in Nigeria for National Development: Challenges and Complexities Dalhatu Usman Abdulkadir 243 15 The Prospects and Challenges of Artificial Intelligence Utilization for Cataloguing of Information Resources in Nigerian Academic Libraries Jemilat Biogera Abubakar, Quawiyah Omoseke Aderinto, Tawakalitu Bola Abdulsalam, Babatunde Julius Odusina, Gift Titilayo Oshagbemi 251 16 Harnessing the Potentials of Computing and Communication Technologies (CCT) for National Development Dalhatu Usman Abdulkadir 268 17 Application of Wireless Communication based Portable Electro-cardiology System for Enhancing Health Care in Rural Areas Olawoyin, L.A, Oloyede A.A, Faruk N, Adeniran, Temitayo C., Oloyede M.O, Adeshina M 274 18 A Machine Learning-based Depression Diagnosis and Prediction Model G. K. Afolabi-Yusuf, H. Lawal, F. O. Uthman, A.A. Hammed 281 19 Software Defect Prediction using Optimized Deep Neural Network and Principal Component Analysis Techniques Idowu Dauda OLADIPO, Latifat Bukola ADEOYE, Abdulrauph Olarewaju BABATUNDE 303 20 Bridging the Digital Divide: Empowering Nigerian Universities through Technological Advancements in Academic Libraries Bolaji David Oladokun, Firdausi Abdulahi, Adeyinka Tella 322 21 Media Promotion in Harnessing Artificial Intelligence and Big Data to Enhance Sustainable Urban Planning in Nigeria Lukman Adegboyega ABIOYE1, Omolara Oluwatoyin IDEHENRE 333 22 Integrating Convolutional Neural Networks and Long Short-Term Memory for Enhanced Detection of Glaucoma Abidemi Emmanuel ADENIYI, Blessing Oluwatobi OLORUNFEMI, Rasheed Gbenga JIMOH, Mukaila OLAGUNJU, Joseph Bamidele AWOTUNDE 350 23 Assessment of Availability, Value, and Challenges of References Service in Academic Libraries of Universities of North West Geo-Political Zone in Northern Nigeria Balogun Tawakalitu Raufu, Abdullahi Mohammmad Ibrahim, Adewara Janet Oluwaseyi Omoniyi 361 Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) Copyright ©2024 vi Online ISSN: 2645-2960 Print ISSN: 2141-3959 Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 01-14 Online ISSN: 2645-2960 Print ISSN: 2141-3959 An Improved Machine Learning-Based Framework for Fake News Detection Oluwakemi Christaina Abikoye1, Taofeek Akintunde Abdulsalam2 Department of Computer Science, Faculty of Information and Communication Sciences, University of Ilorin, Ilorin 240003, Nigeria1 Department of Computer Science, Lens Polytechnic Offa, Kwara, Nigeria2 abikoye.o@unilorin.edu.ng1 , binarycode1995@gmail.com2 Abstract The world we know today has a large amount of information, events, data, and news circulating rapidly across the globe at an exponential rate. Due to the invention of modern telecommunication technology dissemination of fake news has increased exponentially. However, fake news has also gained high potential for spreading quickly across social media and blog posts. Researchers in the past proposed many techniques, including machine learning, and deep learning, for fake news detection. However, a review shows that previous studies have not yet established satisfactory performance, while others suffer from model overfitting on training sets with poor performance on testing data. Achieving an optimum performance with a machine learning approach is highly difficult, especially when only a machine learning algorithm is used to address NLP tasks. Hence, this research proposes a transfer learning-based framework thus, a Bidirectional Long Short Memory to Machine Learning (Bi-LSTM-2-ML) for fake news detection. The Framework is divided into two steps including, (1) The Bi-LSTM Architecture has to be pre-trained first on the fake news dataset for extracting highly contextualized features embedding, and (2) the embedding will be further used to enhance the training process of various machine learning algorithms including; SVM, LR, NB, DT, and KNN. As a result, the proposed framework (Bi-LSTM-2- ML) for the fake news detection model shows promising results and capability in enhancing the machine learning model with an average performance of 6% increase in fake news detection. Keywords: fake news, social media, machine learning, Bi-LSTM, deep learning, internet, transfer learning. 1. Introduction In recent years, social media has become the primary platform for the exchange of news and information. This transformation has been made possible by the internet, providing an enabling environment for the easy sharing of media content. (Song et al., 2021). Recent research indicates a shift in news consumption, with many users preferring digital-based news over traditional newspapers. However, the authenticity of news on social media (digital news) is a challenging and difficult process, unlike content from radio and television, which are subjected to critical review and supervision before broadcasting (Agudelo et al., 2018). Detection of fake news remains essential due to the widespread dissemination of misinformation, rumors, and fake news on platforms such as Facebook, Instagram, Twitter, and WhatsApp (Jain et al., 2019). Training intelligent systems to identify fake news on social space presents challenges due to the wide range of news forms and sources of interdisciplinary news. To effectively detect fake news, it is essential to utilize various Natural Language Tool Kits (NLTK) such as part-of-speech tagging, sentiment analyzer, and vectorization (Braşoveanu & Andonie, 2019). Additionally, other resources and tools are available to help individuals identify and spot fake news on their social media feeds, by analyzing essential facts and fake news content. The primary objective of fake news detection is to prevent the quick dissemination of false information across various platforms, including social media and messaging applications. Fake news detection systems can result in harmful consequences such as mob violence if not properly addressed Jain et al., (2019). Additionally, Hossain et mailto:abikoye.o@unilorin.edu.ng1 mailto:binarycode1995@gmail.com2 An Improved Machine Learning-Based Framework for Fake News Detection. Abikoye and Abdulsalam Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 01-14 2 al., (2022) show that fake news detection aims to identify misleading information and prevent unwarranted acts of violence, which could ultimately safeguard the societal peace. Machine learning and deep learning techniques have been adopted by many researchers in the past, for the detection of fake news, which include Support Vector Machine (SVM), Naïve Bayes (NB), K-Nearest Neighbor (KNN), Logistic Regression (LR), Decision Tree (DT), Recurrent Neural Network (RNN), Long Short Term Memory (LSTM), Bidirectional-Long Short Term Memory (Bi-LSTM), and Gated Recurrent Unit (GRU) Pandey et al., (2022). Artificial intelligence (AI) is a broad term denoting the field of study where machines are trained to perform intelligent tasks without human intervention (Zhang & Lu, 2021). AI domain is divided into subfields such as knowledge processing, pattern recognition and natural language processing, deep learning, and machine learning for building intelligent systems. Charbuty & Abdulazeez, (2021) reveal that machine learning technology greatly reduces human work, by using statistical and mathematical algorithms to achieve a task more efficiently. Natural language processing (NLP) provides the tools and technology to build communication interfaces between humans and computers using natural languages and computational algorithms (Goyal et al., 2018). NLP has wide application in fields such as medicine, language translation, sentiment analysis, and false news detection (Khurana et al., 2023). Patel & Patel, (2021) confirmed that deep learning models have been successfully deployed to address natural language processing tasks with high-performance records, commonly used algorithms include Recurrent Neural Network (RNN), Long Short Term Memory (LSTM), Gated Recurrent Unit (GRU), Convolutional Neural Network (CNN) and Bi-LSTM. The RNN architecture is the basis of all recurrent-based network architecture, which is primarily used for natural language processing tasks, and processing of sequential-based data (text data) in a feed-forward manner (Patel & Patel, 2021). Long Short-Term Memory (LSTM) architecture provides a comprehensive solution to address short-term memory limitations in a recurrent neural network. The architecture incorporates internal mechanisms known as gates, which can control the flow of information Hossain et al., (2022). LSTMs are widely adopted by researchers to address text-based classification problems (Kumar et al., 2020), and Bidirectional LSTM (Bi-LSTM) models have demonstrated high effectiveness in gathering sequential information from both forward and backward directions of the text data (Asish et al., 2022). 1.1 Research Problem Pandey et al., (2022) revealed that the drastic transformation in news dissemination from traditional newspapers to electronic-based news forecasting has contributed a lot to the widespread of fake news in today’s world. This resulted from a limited or no supervision policy scheme for monitoring news been published online (Hossain et al., 2022). Many researchers in the past have proposed various techniques to address these issues but with less satisfactory performance Pandey et al., (2022); Kumar et al., (2020); Jaybhaye et al., (2023). Achieving an optimum performance in the domain of NLP is highly challenging. Hence, methods such as transfer learning and hybridization techniques are suggested by researchers Kaliyar et al., (2021); Jain et al., (2019); Kumar et al., (2020); Jaybhaye et al., (2023) to improve the performance of a real-time fake news detection model. Additionally, Pandey et al., (2022); Jaybhaye et al., (2023) adopted various machine learning algorithms for building fake news detection models including Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Logistic Regression (LR), Decision (DT), and Naïve Bayes (NB) algorithms. The result of Pandey et al., (2022) shows that the highest accuracy attained was 90.04%, which creates room for further improvement, and Jaybhaye et al., (2023) suffered from over-fitting during training with low performance on testing data. Hence, this study proposed an improved Framework (Bi-LSTM-2-ML) for machine learning fake news detection. An Improved Machine Learning-Based Framework for Fake News Detection. Abikoye and Abdulsalam Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 01-14 3 1.2 Research Goal The research goal is to design an improved fake news detection framework using Bi-directional Long Short Term Memory (Bi-LSTM) in a transfer learning approach with various machine learning algorithms. Hence, a Bi- LSTM-2-ML fake news detection framework 1.3 Contribution to Knowledge Design of an enhanced Bi-lSTM-2-Ml framework to mitigate identified gaps such as the limitation of ML to extract high contextual features and lesser performance on testing data. 2. Related Works Varshini et al., (2024), identify that fake news detection is essential for societal well-being. However, current fake news detection models prioritize performance only resulting in data overfitting and reduction in data generalization. In addressing this issue, the researchers introduced a Robust Distribution Generalization of Transformers-based Generative Adversarial Network Architecture (RDGT-GAN). This developed model is robust enough to identify, and properly classify any COVID-19 fake news datasets with varying distributions without retraining. The experimental result shows that the proposed model outperforms existing state-of-the-art models with 83%, 76%, 55%, and 84% accuracy using D1, D2, D3, and D4 datasets. Hossain et al., (2022) implemented a Bi-LSTM model with GLOVE, Fast-Text word embedding, and cross-fold validation techniques to detect Bangla fake news. A Gated Recurrent Unit (GRU) deep learning model is introduced to detect fake news. Based on the experimental analysis conducted, an accuracy of 96% is achieved using the Bi-LSTM deep learning model, and 77% accuracy using the GRU model for detecting and classifying fake news from real news. Based on the research work of Pandey et al., (2022) and Jaybhaye et al., (2023), various Machine learning techniques such as K-Nearest Neighbor, Support Vector Machine, Decision Tree, Logistic Regression classifier are introduced for accurate prediction and identification of fake news from real news. Performance comparison of Pandey et al., (2022) show that logistic regression performs best with 90.48% accuracy. However, Jaybhaye et al., (2023) encountered data overfitting during training, which resulted in poor performance on testing data. The researchers obtained 86% accuracy using the LSTM model. The deep LSTM model achieved better performance than the machine learning model. However, Kaliyar et al., (2021) revealed that adopting bidirectional encoding for fake news detection performs better than the unidirectional-based model. Hence, the researcher proposed a Bidirectional Encoder Representation from Transformer (FakeBERT) model integrated with a single-layer deep CNN network. As a result, the developed model achieved an accuracy of 98.90%, which outperformed the existing study. In the research work of Wang et al., (2023) it identify that the exponential increase of information on social media has heightened the need for swift and accurate real-time rumor detection. The traditional methods of verifying fake news using AI tools are impractical due to the large volume of information uploaded daily. Consequently, there is a growing interest in developing automated systems to detect fake news online. The researchers highlighted that fine-tuned BERT and RobertA models excel in identifying news generated using AI, with RobertA achieving an impressive accuracy score of 98% with exceptional precision. This study underscores the effectiveness of neural networks, particularly the RobertA and BERT models, in addressing misinformation, specifically AI-generated news from platforms like ChatGPT. An Improved Machine Learning-Based Framework for Fake News Detection. Abikoye and Abdulsalam Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 01-14 4 In today’s world, the social media platform is ranked as the first internet tool used in spreading news across the globe. Choudhary & Arora, (2021) claim that social media is an essential tool for disseminating information and creating awareness among the public, but has no authentication schemes for validating the origin and content of all published news. A linguistic model introduced for extracting essential syntax and semantic information describing the features of fake news from real news is proposed by the researchers to address the widespread of fake news uploaded on the internet. The developed model achieves an average accuracy of 86% for detecting and classifying fake news. The comparative analysis shows that the developed model performs efficiently better than the existing model. Vijjali et al., (2020) explore the high alarming rate of fake news occurrence during the COVID-19 pandemic, which results in people finding it difficult to differentiate false information from real COVID-19 news. Hence, the researcher proposed a state-of-the-art machine learning model based on a two-level automated pipeline for detecting COVID-19 fake news. The study reveals that the Bidirectional Encoder from Transformer (BERT) (92%), and the ALBERT (94%) models perform satisfactorily. Furthermore, fake information created on the internet today contains manipulated images, audio, text, and video data uploaded online. Singhal et al., (2019) revealed that manipulated data are easily spread via social media due to the fast-growing nature of social media users. The researcher proposed a multimodal-based fake news detection system, which internally utilized a Bidirectional Encoder from Transformer (BERT) Architecture to extract features from text, and the VGG19 pre-trained model for image feature extraction. As a result, the developed model outperformed the existing state-of-the-art model train on Twitter and Weibo datasets with an accuracy increment of 3.27% and 6.83%. 3. Research Methodology Figure 1. Proposed Framework (Bi-LSTM-2-ML) Methodology The study proposed a Transfer learning-based Bidirectional Long Short Term Memory (Bi-LSTM) to Machine learning (ML) fake news detection Framework (Bi-LSTM-2-ML). Technically, the Bi-LSTM Architecture would be used to pre-train the fake news detection model for extracting contextualized bidirectional-based feature embedding, which will be transferred to machine learning algorithms for further training. Figure 1 shows the methodological process of the proposed Bi-LSTM-2-ML Framework for Fake news detection. The methodology stages include the following; 3.1 Methodological Process An Improved Machine Learning-Based Framework for Fake News Detection. Abikoye and Abdulsalam Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 01-14 5 3.1.1 Fake News Data Collection This first stage includes data collection from the Kaggle data science repository. The dataset will be downloaded in a comma-separated-value format (CSV) into the local system drive for further preprocessing. However, the kaggle repository provides different datasets for addressing scientific problems, a development environment (Jupyter Notebook), and a virtual hardware processing unit (GPU, TPU, and CPU) for solving complex tasks. The proposed fake news dataset samples are shown in Figure 2, and the dataset link on the Kaggle repository is provided below. DatasetLink:https://github.com/GeorgeMcIntire/fake_real_news_dataset Figure 2. Fake News Dataset (CSV file) 3.1.2 Data Exploration and Visualization This stage is also known as data exploratory analysis (EDA), it involves an in-depth study of the fake news dataset for proper insight. Additionally, this step helps identify hidden patterns, empty strings, duplicate samples, stop words, and special symbols in the dataset. Data visualization denotes a graphical representation of information and exploration conducted on the dataset. This study will use various charts from Python Library (Matplotlib.pyplot) to perform visual exploration. 3.1.3 Data Preprocessing and Cleaning The Natural Language Tool Kit (NLTK) is essential for solving natural language processing tasks (Goyal et al., 2018). The data preprocessing and cleaning will be conducted using Python modules such as Stopwords, PotterStemmer, and Tokenizer from the NLTK library. The cleaning process is based on the exploration that will be done on the dataset. The cleaning process will include; stopwords removal, punctuation marks, special symbols, and special characters. Moreover, the preprocessing activities include stemming, tokenizing, and text case conversion. https://github.com/GeorgeMcIntire/fake_real_news_dataset An Improved Machine Learning-Based Framework for Fake News Detection. Abikoye and Abdulsalam Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 01-14 6 3.1.4 Feature Extraction (Word Embeddings) The process of converting text data into numeric vectors denotes vectorization. However, Word Embedding denotes feature vectors generated using shallow or deep neural network architecture, such as Word2Vec, FastText, and GloVe (Tai et al., 2023). This study proposed Word2Vec (CBOW) techniques for generating a 2- Dimentanal feature vector from the fake news dataset. 3.1.5 Data Splitting Data splitting is a standard approach used in dividing datasets into training and testing samples, this provides separate data for training the fake news detection model and evaluates the performance of the model using testing samples. 70% of the fake news dataset will be allocated for training and the remaining 30% will be assigned for model evaluation. 3.1.6 Bi-LSTM training The Bidirectional Long Short Term Memory (Bi-LSTM) architecture is a recurrent-based neural network architecture that processes text input data in both directions. The architecture will be used to pre-train the fake news prediction model for extracting contextualized word embedding vectors, which would be further used as input vectors for various machine learning algorithms during training in a transfer learning approach. 3.1.7 Machine Learning Algorithms (SVM, KNN, RF, & DT) Five Machine learning algorithms including Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Radom Forest (RF) Naïve Bayes (NB), and Decision Tree (DT) will be trained for detecting fake news. The pre- trained fake news weight (word embeddings from Bi-LSTM) will be used to enhance the training process of the considered machine learning algorithms for fake news detection. 3.1.7.1 Naïve Bayes (NB) Algorithm Naïve Bayes is primarily used to compute conditional probability, which originates from Bayes' theorem by expressing the likelihood of an event occurring based on the occurrence of another event (Villagracia Octaviano, 2021). Naïve Bayes functions as a classifier within the domain of supervised learning algorithms, and it performs predictions based on probability scores from different classes (Octaviano, 2021). P(A|B) = P(B|A) *P(A) (1) P (B) P (A|B): Probability of event A such that event B has already occurred. 3.1.7.2 Support Vector Machine (SVM) The SVM Algorithm is highly efficient for addressing binary classification problems Jain et al., (2019). SVM is a supervised machine learning algorithm that can be used for addressing both classification and regression problems, it is based on the idea of finding a hyper-plane that best splits the dataset into two classes. Hyper-planes are decision boundaries used by machine learning algorithms to classify data points properly Jain et al., (2019). The algorithm provided in Figure 4 is based on feature set N (0-N), and sorted using information gained in decreasing order of accuracy. The algorithm representation of SVM is illustrated in Figure 2. An Improved Machine Learning-Based Framework for Fake News Detection. Abikoye and Abdulsalam Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 01-14 7 Figure 3: SVM Algorithm Khanam et al., (2021) 3.1.7.3 K-Nearest Neighbors (KNN) The K-NN algorithm operates on the concept of similarity measures between new samples and existing samples. The algorithm groups the new sample point to the category that is most likely to be the available categories. It stores all the accessible data and classifies new data points using similarity measures. Figure 4 shows the algorithm illustration of KNN. Figure 4: K-Nearest Neighbor Algorithm Khanam et al., (2021) 3.1.7.4 Decision Tree Algorithm Khanam et al., (2021) Introduce decision tree algorithms as techniques primarily used for solving classification problems, which operate like a flow chart. Each internal node of a decision tree sets a condition or "test" on an attribute, guiding the branching based on these conditions. Ultimately, the leaf node holds a class label which is determined by evaluating all attributes. The distance from the root to a leaf node defines the classification rule. The pseudocode in Figure 5 illustrates the decision tree algorithm. An Improved Machine Learning-Based Framework for Fake News Detection. Abikoye and Abdulsalam Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 01-14 8 Figure 5: Decision Tree Algorithm Khanam et al., (2021) 3.1.7.5 Logistic Regression Logistic Regression is a common approach employed for addressing binary classification problems, it’s also specifically designed for assigning labels to observations within a distinct set of classes. In addressing binary classification tasks, Logistic Regression has proven to be a successful choice (Sangamnerkar et al., 2020). The main strength of Logistic Regression techniques lies in its sigmoid function, which generates a probability value, that is subsequently linked to a class within a discrete set for comparing two or more classes (Sangamnerkar et al., 2020). S (x) = 1 (2) 1 + e-x If S (x) > 0.5 assign label = class A, else assign label = class B, where 0 < S(x) < 1 and (A, B) is a set of the classes (Sangamnerkar et al., 2020). 3.1.8 Performance Metrics It's essential to evaluate the performance of the fake news prediction models using the test samples to measure their performance using standard performance metrics such as accuracy, precision, recall, and fi-score. 3.1.8.1 Accuracy The accuracy metric is used when the dependent variable is categorical or discrete. The accuracy denotes the fractional time a model has predicted to the total predictions made (Kadhim, 2018). Accuracy = TP + TN TP + FP + FN + TN (3) 3.1.8.2 Precision The precision of a text classification model indicates the level of exact prediction carried out by the model (Kadhim, 2018). Technically, it denotes all positive cases (the relevant classes) to the number of classes been classified correctly (Kadhim, 2018). Precision =TP / (TP + FP) (4) 3.1.8.3 Recall An Improved Machine Learning-Based Framework for Fake News Detection. Abikoye and Abdulsalam Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 01-14 9 The Recall and Precision metrics are complimentary to each other. The recall denotes how well a model can captures the recall of positive classes in a problem statement. Hence, given all the positive predictions it made and how many of these positive predictions are truly positive (Kadhim, 2018). Recall = TP / (TP + FN) (5) 3.1.8.4 F1-Score The F1-score metric merges the precision and recall metrics into a single metric, this in turn also captures the alternative forgone between the precision metrics and the recall (thus, the completeness and the exactness) (Kadhim, 2018). F1-score = 2 x Pi x Ri (Pi + Ri) (6) 3.2 System Analysis and Design Figure 6: A Conceptual Framework for The Proposed Bi-LSTM-2-ML Figure 6, illustrates the conceptual framework of how the Bidirectional Long Short Term Memory (Bi-LSTM) to Machine Learning (ML) Transfer learning-based model (Bi-LSMT-2-ML) for detecting fake news will be designed and developed. The various technology, entities, and tools that will be utilized in designing, and An Improved Machine Learning-Based Framework for Fake News Detection. Abikoye and Abdulsalam Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 01-14 10 developing the Bi-LSTM-2-ML fake News detection framework are also described in Figure 6. The fake news dataset is downloaded from the Kaggle repository in a Comma Separate Value (CSV) format, and the data scientist will use the Jupyter Notebook environment to load the downloaded fake news dataset into the IDE using the Python ‘load_csv’ function. Furthermore, the loaded fake news data will be forwarded to stage (1) or preprocessing stage, which includes; Data exploration, Data Cleaning, and Data visualization. The exploration helps to get more insights into the dataset and identify possible correlations, relationships, or features that can help during machine learning prediction. The X, and Y variables (independent, and dependent variables) of the fake news dataset will be extracted in stage 2. Furthermore, transformation (Word2Vec) will be conducted on the raw text to convert the text data into numerical features or word embeddings. After feature extraction, the next step is to split the dataset into training and testing samples, the training sample will be utilized to train the Bi- LSTM deep learning model in other to extract highly contextualized information about the fake news data, then use the pre-trained Bi-LSTM weight to train the machine learning algorithm in a transfer learning approach. the developed model will then be evaluated using various performance metrics such as Accuracy, precision, recall, and F1-score. 3.2.1 System Algorithm Algorithm 1: Fake news Bi-LSTM-2-ML transfer learning-based Model (Pseudocode) 1. Start 2. Load fake news dataset 3. perform data cleaning tokenization remove punctuation marks remove stop_words stemming: potter_stermmer() convert.lower_case() 4. Transform dataset replace fake=> 0 and real=> 1 word_embeddings => word2vector() 5. Feature extraction Tensor_Tokenization generating vocabulary: fit_on_text(train_data) convert to numeric: text_to_sequences generating equal length text: padding_sequenc() input_matrix => generating:weighted_matrix() 6. Data splitting data splitting: train_test_split(train_size=70, test_size=30) return xtrain, ytrain, xtest, ytest 7. fake news deep learning training i. build a bi-LSTM fake news sequential model: embedding_layer(input_matrix) bi-lstm_layer(128 neuron) dense_layer(32 neuron, activation=relu) dense_layer(1 neuron, activation =sigmoid) ii. compile bi-lstm model iii. train bi-lstm model: model.fit (xtrain, ytrain, epoch=10) An Improved Machine Learning-Based Framework for Fake News Detection. Abikoye and Abdulsalam Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 01-14 11 iv. evaluate bi-lstm model: mode.predict (xtest) v. print (accuracy, precision, recall, f1-score) vi. bi-lstm_model.save() 8. machineLearning.train() => (pre-trained weight) i. x_weight => bi-lstm.load_weights() ii. fakeNewModel.train() LogisticRegression.fit(x_weight, ytrain) KNeighborsClassifier.fit(x_weight, ytrain) GaussianNB.fit(x_weight, ytrain) support_vector_classifier.fit(x_weight, ytrain) DecisionTreeClassifier.fit(x_weight, ytrain) iii. fakeNewsModel.evaluate() fakeNews_prediction => ml_model.predict (x_test) print: classification_report (fakeNews_prediction, ytrain) print: confusion_matrix (fakeNews_prediction, ytrain) print: accuracy_score (fakeNews_prediction, ytrain) print: precision_score (fakeNews_prediction, ytrain) print: recall_score (fakeNews_prediction, ytrain) print: f1-score (fakeNews_prediction, ytrain) 9. End Algorithm 1, shows the step-by-step approach that can be followed in developing the proposed Bi-LSTM-2-ML fake news detection model. The algorithm is divided into 9 major steps with subtasks associated with those steps. The major steps include; data loading, data cleaning, data transformation, feature extraction, data splitting, Bi- LSTM deep learning training, machine training using transfer learning, and model evaluation. 4. Implementation 4.1 Bi-LSTM Pre-training The Pre-Trained Bi-LSTM model implementation along with the Bi-LSTM transfer learning for machine learning training will be comprehensively discussed in this sub-section. Various Python libraries such as sklearn, tensorflow, and keras will be used for training the deep/machine learning models. To achieve the Bi-LSTM training process, a sequential-based Bi-LSTM model will be built, compiled, and summarized before training the model. 4.2 Machine Learning Implementation The subsection entails the training process of the various machine learning models, which include Support Vector Machine (SVM), Decision Tree (DT), Naïve Bayes (NB), Logistic Regression (LR), and K-Nearest Neighbor (KNN). Additionally, a Transfer learning-based machine learning models will be trained using Bi-LSTM pre- trained weight for achieving better performance. 5. Result, Conclusion & Future Direction An extensive study shows that the proposed Bi-LSTM transfer learning-2-ML framework can efficiently enhance the performance of a machine learning algorithm with a 6% increase in average accuracy (Singhal et al., 2019). Furthermore, the Bi-LSTM-2-ML Framework for fake news detection shows a promising increase in performance for detecting fake news compared to the benchmark method. An Improved Machine Learning-Based Framework for Fake News Detection. Abikoye and Abdulsalam Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 01-14 12 The research study considered five machine learning models including Logistic Regression (LR), K-Nearest Neighbor (KNN), Naïve Bayes (NB), Support Vector Machine (SVM), and Decision Tree (DT). The study introduces a framework that uses a Bi-LSTM deep learning model that will enhance the performance of the mentioned machine learning model using a transfer learning approach. Pandey et al., (2022) proposed a similar machine learning model for fake news detection, but the model performance was not satisfactory, while Jaybhaye et al., (2023) suffered from model overfitting while training and poor performance on the testing samples. This problem can be addressed by adopting the proposed Bi-LSTM-2-ML framework, and also address the overfitting issue by introducing dropout in the Bi-LSTM architecture. The table shows the performance influence when the transfer Learning technique (Bi-LSTM-2-ML) is adopted, with an average increase in Accuracy performance of 6% (Singhal et al., 2019). Table 1. Result comparison with 6% Improvement Using Bi-LSTM-2-ML Framework Classifier Pandey et al., (2022) (Jaybhaye et al., 2023) Proposed Bi-LSTM-2-ML Framework Logistic Regression 90.46% 91.22% 96.46% K-Nearest Neighbor 89.98% 74.68% 95.98% Naïve Bayes 86.89% 79.41% 92.89% Support Vector Machine 73.33% ***** 79.33% 89.33% 89.0% 98.33% Furthermore, an extensive study can be done on various deep learning and transformers-based architectures in a transfer learning approach for machine learning enhancement. Architecture such as LSTM, Bi-LSTM, GPT, BERT, Albert, GRU, and the likes can be adopted. This will reveal more insights and confirm the influence of deep learning models on machine learning performance in a transfer learning approach. Reference Agudelo, G. E. R., Parra, O. J. S., & Velandia, J. B. (2018). Raising a model for fake news detection using machine learning in Python. Challenges and Opportunities in the Digital Era: 17th IFIP WG 6.11 Conference on e-Business, e-Services, and e-Society, I3E 2018, Kuwait City, Kuwait, October 30–November 1, 2018, Proceedings 17, 596–604. Asish, K. R., Gupta, A., Kumar, A., Mason, A., Enduri, M. K., & Anamalamudi, S. (2022). A tool for fake news detection using machine learning techniques. 2022 2nd International Conference on Intelligent Technologies (CONIT), 1–6. Braşoveanu, A. M., & Andonie, R. (2019). Semantic fake news detection: A machine learning perspective. Advances in Computational Intelligence: 15th International Work-Conference on Artificial Neural Networks, IWANN 2019, Gran Canaria, Spain, June 12-14, 2019, Proceedings, Part I 15, 656–667. Charbuty, B., & Abdulazeez, A. (2021). Classification Based on Decision Tree Algorithm for Machine Learning. Journal of Applied Science and Technology Trends, 2(01), 20–28. https://doi.org/10.38094/jastt20165 An Improved Machine Learning-Based Framework for Fake News Detection. Abikoye and Abdulsalam Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 01-14 13 Goyal, P., Pandey, S., & Jain, K. (2018). Deep Learning for Natural Language Processing. Apress. https://doi.org/10.1007/978-1-4842-3685-7 Haqi Al-Tai, M., Nema, B. M., & Al-Sherbaz, A. (2023). Deep Learning for Fake News Detection: Literature Review. Al-Mustansiriyah Journal of Science, 34(2), 70–81. https://doi.org/10.23851/mjs.v34i2.1292 Hossain, E., Nadim Kaysar, M., Jalal Uddin Joy, A. Z. M., Mizanur Rahman, M., & Rahman, W. (2022). A study towards Bangla fake news detection using machine learning and deep learning. Sentimental Analysis and Deep Learning: Proceedings of ICSADL 2021, 79–95. Jain, A., Shakya, A., Khatter, H., & Gupta, A. K. (2019a). A Smart System for Fake News. 2–5. Jain, A., Shakya, A., Khatter, H., & Gupta, A. K. (2019b). A smart System for Fake News Detection Using Machine Learning. 2019 International Conference on Issues and Challenges in Intelligent Computing Techniques (ICICT), 1–4. https://doi.org/10.1109/ICICT46931.2019.8977659 Jaybhaye, M. (2023). Fake News Detection using LSTM based deep learning approach. ITM Web of Conferences. Jaybhaye—2023—Fake News Detection using LSTM based deep learning.pdf. (n.d.). Retrieved December 6, 2023, from https://www.itm-conferences.org/articles/itmconf/pdf/2023/06/itmconf_icdsac2023_03005.pdf Kadhim, A. I. (2018). An Evaluation of Preprocessing Techniques for Text Classification. International Journal of Computer Science and Information Security, 16(6), 22–32. Khanam, Z., Alwasel, B. N., Sirafi, H., & Rashid, M. (2021). Fake news detection using machine learning approaches. IOP Conference Series: Materials Science and Engineering, 1099(1), 012040. Khurana, D., Koli, A., Khatter, K., & Singh, S. (2023). Natural language processing: State of the art, current trends and challenges. Multimedia Tools and Applications, 82(3), 3713–3744. https://doi.org/10.1007/s11042- 022-13428-4 Kumar, S., Asthana, R., Upadhyay, S., Upreti, N., & Akbar, M. (2020). Fake news detection using deep learning models: A novel approach. Transactions on Emerging Telecommunications Technologies, 31(2), e3767. Pandey, S., Prabhakaran, S., Subba Reddy, N. V., & Acharya, D. (2022). Fake News Detection from Online media using Machine learning Classifiers. Journal of Physics: Conference Series, 2161(1), 012027. https://doi.org/10.1088/1742-6596/2161/1/012027 Patel, R., & Patel, S. (2021). Deep Learning for Natural Language Processing (pp. 523–533). https://doi.org/10.1007/978-981-16-0882-7_45 Sangamnerkar, S., Srinivasan, R., Christhuraj, M. R., & Sukumaran, R. (2020). An ensemble technique to detect fabricated news article using machine learning and natural language processing techniques. 2020 International Conference for Emerging Technology, INCET 2020, 1–7. https://doi.org/10.1109/INCET49848.2020.9154053 Song, C., Ning, N., Zhang, Y., & Wu, B. (2021). A multimodal fake news detection model based on crossmodal attention residual and multichannel convolutional neural networks. Information Processing & Management, 58(1), 102437. An Improved Machine Learning-Based Framework for Fake News Detection. Abikoye and Abdulsalam Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 01-14 14 Varshini, U. S. S., Sree, R. P., Srinivas, M., & Subramanyam, R. B. V. (2024). RDGT-GAN: Robust Distribution Generalization of Transformers for COVID-19 Fake News Detection. IEEE Transactions on Computational Social Systems, 1–15. https://doi.org/10.1109/TCSS.2023.3269595 Villagracia Octaviano, M. (2021). Fake News Detection Using Machine Learning. ACM International Conference Proceeding Series, 177–180. https://doi.org/10.1145/3485768.3485774 Wang, Z., Cheng, J., Cui, C., & Yu, C. (n.d.). Implementing BERT and fine-tuned RobertA to detect AI generated news by ChatGPT. Zhang, C., & Lu, Y. (2021). Study on artificial intelligence: The state of the art and future prospects. Journal of Industrial Information Integration, 23, 100224. https://doi.org/10.1016/j.jii.2021.10022 Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 Online ISSN: 2645-2960 Print ISSN: 2141-3959 Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Aminat ADEBAYO1, Abidemi Emmanuel ADENIYI2, Jumoke AJAO3, Rafiu ISIAKA4, Kazeem GBOLAGADE5, Sulaiman ABDULSALAM6 1Department of Computer Science, Al-Hikmah University, Ilorin, Nigeria; ayakub@alhikmah.edu.ng 2Department of Computer Science, College of Computing and Communication Studies, Bowen University, Iwo, Nigeria. abidemi.adeniyi@bowen.edu.ng 3,4,5,6Department of Computer Science, Kwara State University, Malete, Nigeria. jumoke.ajao@kwasu.edu.ng, abdulrafiu.isiaka@kwasu.edu.ng, kazeem.gbolagade@kwasu.edu.ng, sulaiman.abdulsalam@kwasu.edu.ng Abstract Data security is critical in ensuring the privacy of data including sensitive material that ought to only be known by a few people. Every society needs secured data to maintain the integrity and authentication of the data. Data Encryption Standard (DES) is a block cipher algorithm that has been used to secure data or information over the years. Despite the tremendous efforts made by researchers on DES algorithm and efforts to reduce its computational complexity, DES is still susceptible to brute force attack. The need to increase the degree of security of DES algorithm led to the introduction of Residue Number System (RNS) to the DES algorithm as proposed in this study. The method for DES-RNS multilevel encryption uses a 64bits plaintext message which was encrypted using the DES technique. The 64bits plaintext message was divided into two equal halves; 32bit left plaintext (LPT) and 32bit right plaintext (RPT). The RPT was encrypted using 48bit sub-keys and the result was XORed with LPT. The transformation of RPT and LPT was performed for sixteen (16) rounds to produce encrypted text of 64bits. The encrypted text of DES was converted to American Standard Code for Information Interchange (ASCII) and passed through RNS forward conversion. The RNS made use of the moduli set 〖m_1=2 〗^n+1,m_2= 2^n and m_3=2^n-1. The decryption was performed using Chinese Remainder Theorem (CRT). The result was evaluated when it comes to cryptographic time, encryption/decryption memory, encryption/decryption throughput and security on three varying text sizes (256, 800 and 1472 bit) for DES only and DES-RNS multilevel cyrptosystem. Using DES, time and throughput shows a better performance for 256bit, 800bit and 1472bit message size but lesser performance in memory and security for 256bit, 800bit and 1472bit message size. On the other hand, using DES-RNS, time and throughput gives a lesser performance for 256bit, 800bit and 1472bit message size. Therefore, DES-RNS multilevel encryption model outperformed the conventional DES model in regard to storage utilization and safety, thereby achieving the aim of this research. Consequently, this DES-RNS model can be employed where security and memory conservation is of utmost concern. Keywords: Cryptography, Block Cipher, Data Encryption Standard (DES), Residue Number System (RNS). mailto:ayakub@alhikmah.edu.ng mailto:abidemi.adeniyi@bowen.edu.ng mailto:jumoke.ajao@kwasu.edu.ng mailto:abdulrafiu.isiaka@kwasu.edu.ng mailto:kazeem.gbolagade@kwasu.edu.ng mailto:sulaiman.abdulsalam@kwasu.edu.ng Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 16 1. Introduction Data security is a collection of techniques designed to protect data from unwanted access or change. Data protection is critical in ensuring the privacy of sensitive content that ought to only be accessed by specified parties (Logunleko et al, 2020). Every society needs secured data to maintain the integrity and authentication of the data. Cryptography is a sub-field under data security which deals with safeguarding sensitive data. It's a way of maintaining and disseminating information in a manner that only those who are authorized can access it supposed to read and process it can do so (Maurya, 2021). Cryptography was introduced to keep our information secure. Two categories of cryptographic methods which are used to secure data are Asymmetrical cryptography and symmetric cryptographic encryption are two types of encryption. There are many cryptographic algorithms that are developed and widely used for information security such as Data Encryption Standard (DES), Advanced Encryption Standard (AES), Tripple Data Encryption Standard (TDES), Blowfish etc. Some of these cryptographic algorithms were developed using different techniques and structures. Among these techniques are caesar cipher, monoalphabetic cipher, and feistel cipher model. DES is one of the algorithm that uses the structure of feistel cipher, which is one of the major structure used for the design of block cipher. DES is a block cipher, which means that rather than one bit at a time, an encryption key and procedure are deployed to a block of data at the same time (Jawahar and Nagesh, 2018). DES divides an original message into 64-bit blocks to encrypt it. Using permutations and replacement, each block is enciphered into a 64-bit ciphertext using the private key. The method consists of 16 rounds, with the final encryption block being reliant on all preceding blocks. Decryption is just the inverse of encryption, with the identical stages but with the keys applied in the opposite order. The most fundamental form of attack for any cipher is brute force, which includes attempting each key until you locate the correct one. The length of the key influences the number of potential keys and thus the attack's viability. DES employs a 64-bit key, however eight of those bits are dedicated to parity tests, thereby reducing the key to 56 bits. As a result, it would take a maximum of 256 attempts, or 72,057,594,037,927,936, to locate the proper key. DES is weak against brute force attack, susceptible to linear and differential cryptanalysis attack (Mahajan and Sachdeva, 2013). Due to this brute force attack, it is imperative to enhance the security level of DES. In ameliorating this problem, this research work employed Residue Number System (RNS) to further enhance the security of DES algorithm using a multi-level approach. RNS is a numeric scheme that quickens numerical calculations by dividing numbers into smaller units that are independent of one another (Babatunde et al, 2016). RNS designs are usually made up of three major components: a binary to residual converter, residue mathematical units, and a residue to binary converter. The Chinese Remainder Theorem (CRT) and Mixed Radix Conversion are two of the most extensively utilized reverse conversion procedures (MRC). This research work employed forward and reverse conversion technique. The motivation of this study was as a result security challenges of Data encryption standard. In today's environment, encryption is critical for protecting confidential material from eavesdropping, intruders, and unauthorized users. There have been several assaults in which unknown individuals attempted to disrupt the interaction between two approved devices. DES can be conclusively said to have security and space complexity issues. Therefore, this study focus on the improving the security flaws of DES algorithm by enhancing its Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 17 structure in terms of security. In addition enhance the performance of the algorithm through the use of RNS, thus, deliver an efficient algorithm in terms of security and memory consumption. The remaining part of this study is divided into five sections. Section two discussed the review of related works with the summary of literature. The material and methods used in this study was discussed in section three. The detailed of the method including the DES and DES+RNS algorithm was given in details. Section four gives the results generated in the study with detailed discussion. The performance and metric used was display with tables and figures in this section. The conclusion of this study with future research work was given in section five. 2. Related Works To examine and improve the performance of different cryptography algorithms, a lot of investigation and the symmetric encryption scheme has been improved. This literature provides an overview of various improvements and analysis performed on the DES algorithm and other cryptographic algorithms. Reyad et al, (2021) worked on Key-Based Enhancement of DES for Text Security. New key distribution function (KE-DES) was created for the key and data for more security. The security is limited to the key and the input data is textual. Kumar et al, (2021) present a hybrid approach for cloud security using DES and RSA cryptographic algorithms. It was found that when compared to the present system, the model raises data security to the highest level possible while also taking less time to upload and retrieve text files of DES alone. Maurya, (2021) carried out a comparative study of security algorithm for data transfer. The cyrptographic algorithms compared are DES, AES and RSA. The experimental results are implemented using the Visual studio Net packages. Seven factors were used as the performances metrics, which include key length, cipher type, block size, designed, crypto algorithms, safety, possibility key, ACSII readable letter keys, and time necessary to verify all possible keys are also compared for different file size. Based on the security factor, AES was reported to be excellent secured, DES was said to be less secured and RSA was reported to be very less secured. Logunleko et al, (2020) compared the symmetric cryptographic mechanisms DES, AES, and EB64 for data security. This study focused on cryptography as a means of achieving zero tolerance for data in transit security. Performance of those algorithm is evaluated by considering the parameters such as simulation time taken, encryption time and decryption. The result showed that AES is excellent secured, DES is not secured enough and EB64 is adequately secured. Kumar and Saxena, (2020) conducted a research on System Development File Transmission Safety Using AES and DES Encryption Smartphone. Three cryptographic algorithms were combined i.e AES and DES. It was discovered that the hybrid method to secure the transmission of information from the sender to the recipient via cloud service is more dependable than traditional asymmetrical or symmetrical encryption and decryption of information. Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 18 Alhassan et al, (2020) carried out a research on Enhancment of RNS and k-shuffle are used to secure images during communication. The k-shuffling approach was used to jumble picture pixels and then dissect them using RNS. The result shows that images are secured, no loss of information and reduction in images size as well as transmission speed was achieved. Odunayo (2020) worked on improving AES with RNS. Instead of transmitting just one private key straight to the receiver, the proposed approach sends two private keys to the recipient decoding of the two cipher text is converted returned to original message. At the same time, there is a speed issue. The results demonstrated the improved AES's effectiveness on text data. As a result, while the traditional AES is slightly faster than the implemented algorithm, the proposed algorithm's performance can still be improved. Mahmoud and Alqumboz, (2019) performed an EBMSR traditional database cryptography relies on multilayer security. The cryptographic algorithms combined are AES-256, DES and RC4. It was concluded that cryptography centered on a layered technique improved database security and created challenges for intruders to reveal the data. Anbazhagan et al, (2019) carried out a research on enhanced privacy preserving using multilevel encryption technique. The technique combines blowfish with AES and it was reported that a multilevel data encryption technique reduces the theft of data. Alhag and Mohamed, (2018) presented an enhancement of DES algorithm (KE-DES). The research focuses on increasing the key length to (1024 bits) thereby making it better than the conventional DES which is limited to 64 bits. This research is limited to key length expansion. Kaur et al, (2017) carried out a study of multilevel cryptography algorithm: Multi-Prime RSA and DES. In this work, the Multi-Prime RSA was used to encrypt the shared DES key thereby making it more secured. The result showed that the data is more protected with the dual protections of DES and Multi-Prime RSA. Table 1. Summary of Reviewed studies. Author Method Result Gap Kaur et al. (2017) Multi-Prime RSA and DES Both algorithms are secured Computational cost Alhag and Mohamed (2018) Enhanced DES (KE-DES) Perform better than classical DES Key length expansion limitation Anbazhagan et al, (2019) Combined blowfish and AES Reduces the theft of data Still vulnerable to attacks Mahmoud and Alqumboz (2019) AES-256, DES and RC4 Enhanced security Time complexity Odunnayo (2020) AES and RNS Traditional AES is faster than the proposed algorithm Issue with speed and memory complexity Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 19 Alhassan et al. (2020) RSA and K-shuffle No loss of information during encryption of image file Vulnerability attacks Kumar and Szxena (2020) AES and DES It is more reliable for cloud service security Computational cost Logunleko et al, (2020) DES, AES and EB64 AES is excellent secured while DES is not Challenges with DES security Maurya (2021) DES, AES and RSA AES outperform other two algorithms DES is vulnerable to attacks. Kumar et al. (2021) Hybrid DES and RSA The hybrid approach increase the data security Complexity issues Reyad et al. (2021) Key-Based Enhancement of DES Improve in the data security of textual data The security is limited to the key input on text data. 2.1 Inference from the Literature From all the reviewed research works, DES can be conclusively said to have security and space complexity issues. In addition, none of the multilevel approaches reviewed used RNS as a means of enhancement. The security problem of DES as well as memory space consumption can be improved upon through the proposed multilevel DES-RNS scheme employed in this research. 3. Material and Methods The intended work's goal is to create a model to enhance the security vulnerability of DES using the RNS. The stage involved in this research are: i. Design of the DES-RNS model ii. Implementation of the DES-RNS model iii. Evaluation of the implemented model in step 2. Design of the DES-RNS Model The multilevel DES-RNS cryptosystem model is shown in Figure 1. The model takes input plaintext of 64bits and generate 64bits key for encryption. Out of this 64bits key, 48bits was used to encrypt with the RPT of 32bits and the result was XORed with LPT. This was done for 16 rounds to generate the DES ciphertext. The 64bits ciphertext was converted to ASCII which was later passed to RNS forward conversion to produce DES-RNS ciphertext. Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 20 Figure 1: The flow Diagram for the Designed model DES Encryption Process Message Input Random messages of varying bits was utilized to analyze and assess the model's performance. Three sizes of message types were chosen for experimental purposes. The 256 message bit size, 800 message bit size and 1472 message bit size. Encryption Phase The encryption phase considered using the DES Algorithm as stated in Algorithm 1. Algorithm 1 : DES Encryption Step 1: Input Message Step 2: Generate 64 bit long key Step 3: Encrypt Message Step 4: Return DES cipher-text Decryption Phase The decryption phase considered using the DES Algorithm as stated in Algorithm 2 Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 21 Algorithm 2 : DES Decryption Step 1: Input ciphertext Step 2: Enter Equivalent 64 bit long key generated during encryption Step 3: Decrypt Message Step 4: Return plaintext DES-RNS Encryption Process Message Input Random messages of varying bits was utilized to analyze and assess the model's performance. Three sizes of message types were chosen for experimental purposes. The 256 message bit size, 800 message bit size and 1472 message bit size. Encryption Phase The encryption phase considered using the DES+RNS as stated in Algorithm 3. Algorithm 3 : DES-RNS Encryption Step 1: Input Message Step 2: Generate 64 bit long key Step 3: Encrypt Message Step 4: Return DES ciphertext Step 5: Convert Cipher Text to ASCII values Step 6: Convert ASCII values to Residues (Encrypted Output) using forward conversion. Decryption Phase The decryption phase considered using the DES with RNS enhancement. The sequential integration is given in Algorithm 4. Algorithm 4 : DES-RNS Decryption Step 1: Input DES-RNS ciphertext Step 2: Perform reverse conversion using CRT Step 3: Return DES ciphertext Step 4: Enter Equivalent 64 bit long key generated during encryption Step 5: Decrypt message Step 6: Return plaintext 64 bit plaintext was encrypted using the DES encryption algorithm. Initial Permutation was performed on the plaintext after which it was divided into two equal halves; the 32 bit left plaintext (LPT) and the 32 bit right plaintext (RPT). DES uses a 64 bit key and this by eliminating every eighth bit of the original key, a 64-bit key is turned into a 56-bit key which was then used as parity check bit. As a result, a 56-bit key is accessible for every cycle. For each round, a distinct 48-bit Sub Key was produced from this 56-bit key using a procedure in key transformation called compression permutation. The 56-bit key is compressed to 48 bits during the key translation procedure. The extension permutations procedure then increases the 32-bit RPT to 48 bits. The 48-bit key has been utilized to encode the 48-bit RPT as well as the result was XORed with the 48 bit of the LPT, the output of which served as the input to the next round of the DES encryption process. This process was repeated for 16 rounds. ASCII Conversion The ASCII conversion is being used to convert the cipher text of the DES which are resultant combination of string output of characters and numbers. The ASCII was triggered in the MATLAB using the double conversion technique. Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 22 RNS Forward Conversion (Encryption) RNS is defined by a collection of k integers m 1, m 2, m 3, ldots, m k, known as the moduli, which are commonly assumed to be pairwise coprime (that is, any two of them have a greatest common divisor equal to one). RNS have been defined for non-coprime moduli, however they are rarely utilized due to their poor characteristics. The moduli set {𝑚1 = 2𝑛 + 1, 𝑚2 = 2𝑛 𝑎𝑛𝑑 𝑚3 = 2𝑛 − 1 }, was used for the implementation of the proposed scheme. The RNS forward conversion helped to generate a secured cipher text that further secured the DES algorithm. The forward conversion method is used for the encryption process of the RNS thereby forming the residues. Forward conversion is the process of converting from decimal to RNS notation. The most direct technique to change from a standard notation to a residue illustration is to divide by each of the specified moduli and then capture the remnants. The residues can be defined as xi = X mod mi. (1) Where xi = Residues of X with respect to modulo mi, X = Decimal Number Mod = Modulus operation mi = Moduli sets. RNS Backward Conversion (Decryption) The resultant decrypted text by the DES-RNS technique which produces residues are being passed to the RNS backward conversion with the supply of equivalent moduli set {𝑚1 = 2𝑛 + 1, 𝑚2 = 2𝑛 𝑎𝑛𝑑 𝑚3 = 2𝑛 − 1 } that was used during the encryption stage. The Backward conversion process utilizes the CRT to reverse the residues back to the DES cipher text. M n 1 i m 1- ii i MM X  = = ix (2) where M =  = n i im 1 Mi =M/M i and M-1 is the multiplicative inverse of M with respect to mi Implementation of the DES-RNS Model The DES-RNS model was performed done and created using Matlab programming (MATLAB 2016A). For user engagement and reactivity, many functionalities were built and linked to a visual user interface. To build and output the results of the produced work, the developed systems used several component environments in Matlab. 4. Result and Discussion The developed system DES with RNS was experimented upon with samples of messages with varying bit sizes so as to determine the performance and efficiency of the implemented system. The experiment considered two case studies which are to determine the performance of the RNS on the DES Algorithm, the first case considered encryption and decryption of text messages with only the DES algorithm while the second phase considered the combination of the DES and RNS. Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 23 Performance Evaluation of DES and DES+RNS The implemented system was evaluated on three varying text sizes. The results are shown in the section below. 256 Bit Message Size A 256-bit size of sample was provided into the built software module, which contained the two studied models (DES alone and DES+RNS). Table 1 displays the results collected in these four separate modules during this approach. Encryption Process (ET), Decoding Time (DT), Encryption Memory (EM), Decryption Memory (DM), Encryption Throughput (ETP), and Decryption Throughput (DT) results are detailed (DTP). Table 2. 256 Bit Message Results 800 Bit Message Size The two examined systems were fed 800 bit data into the built programming module (DES only and DES+RNS), which were taken into consideration, the experimentation results was measured with varying evaluation measures the Table 2 gives the results evaluation. The results detailed the Encryption Time (ET), Decoding Time (DT), Encrypted Memory (EM), Decoding Memory (DM), Encrypted Throughput (ETP), and Decoding Throughput (DT) are the several types of encryption times (DTP). Table 3. 800 Bit Message Results Evaluation Measures DES (Seconds) DES+RNS (Seconds) Encryption Time 0.334354 1.85846 Decryption Time 0.120378 0.169121 Evaluation Measures DES (Byte) DES+RNS (Byte) Encryption Memory 1.38183E09 1.38106E09 Decryption Memory 1.38106E09 1.38106E09 Evaluation Measures DES (Kilobyte/Seconds) DES+RNS (Kilobyte/Seconds) Encryption Throughput 765.66472 137.74846 Decryption Throughput 2126.59910 1513.71807 Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 24 1472 Bit Message Size The two examined models were fed 1472 bit data size into the produced software application (DES only and DES+RNS), which were taken into consideration, the experimentation results was measured with varying evaluation measures the Table 3 gives the results evaluation. The results detailed the Encryption Time (ET), Decoding Time (DT), Encrypted Memory (EM), Decoding Memory (DM), Encrypted Throughput (ETP), and Decoding Throughput (DT) are the several types of encryption times (DTP). Table 4. 1472 Bit Message Results Graphical Illustration Evaluation Measures DES (Seconds) DES+RNS (Seconds) Encryption Time 0.26136 1.18305 Decryption Time 0.10733 0.45262 Evaluation Measures DES (Byte) DES+RNS (Byte) Encryption Memory 1.37526E09 1.36131E09 Decryption Memory 1.37472E09 1.36271E09 Evaluation Measures DES (Kilobyte/Seconds) DES+RNS (Kilobyte/Seconds) Encryption Throughput 3060.91215 676.21825 Decryption Throughput 7453.64763 1767.48708 Evaluation Measures DES (Seconds) DES+RNS (Seconds) Encryption Time 0.269351 3.86896 Decryption Time 0.154326 0.92687 Evaluation Measures DES (Byte) DES+RNS (Byte) Encryption Memory 1.38794E09 1.38056E09 Decryption Memory 1.38056E09 1.32954E09 Evaluation Measures DES (Kilobyte/Seconds) DES+RNS (Kilobyte/Seconds) Encryption Throughput 5464.98806 380.46400 Decryption Throughput 9538.25019 1588.14073 Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 25 The section shows the graphical illustration of the obtained results based on the varying sizes of the messages used. Performance Chart of the 256 Bit Message Size The encryption time shows the execution time for each algorithm to complete the cycle of their operation or processing for encrypting an input message while the decryption time shows the execution time for each algorithm to complete the cycle of their processing for conversion of encrypted message into its original message Encryption and Decryption Time (256 Bit Message Size) The encryption and decryption time shows that the DES model outperformed the DES+RNS model because the later uses a multi-level approach and will definitely take more time to run compared to DES only. In this work, time complexity is being traded-off for security. Figure 2. Encryption and Decryption Time for 256 Bit Message Encryption and Decryption Memory (256 Bit Message Size) The encryption and decryption memory shows the space complexity used for each of the algorithm to process their total algorithms operations. The encryption and decryption memory shows a better efficiency with the DES+RNS model as compared to the DES algorithm. 0.269351 3.86896 0.154326 0.92687 0 1 2 3 4 5 DES (Seconds) RNS+DES (Seconds) Encryption and Decryption TIme for 256 Bit Message Size Encryption Time Decryption Time 1.38E+09 1.38E+091.38E+09 1.38E+09 1.38E+09 1.38E+09 1.38E+09 1.38E+09 1.38E+09 1.38E+09 1.38E+09 1.38E+09 DES (Byte) RNS+DES (Byte) Encryption and Decryption Memory at 256 Bit Size Message Encryption Memory Decryption Memory Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 26 Figure 3. Encryption and Decryption Memory at 256 Bit Message Encryption and Decryption Throughput (256 Bit Message Size) The throughput analysis based on the encryption time is measured by the size of the input text per seconds. (Kilobyte/Secs) while the throughput analysis based on the decryption time is measured by the size of the input text per seconds. (Kilobyte/Secs). The higher the encryption or decryption throughput validates the better model, as for the implemented models, the DES model outperformed the DES+RNS model at 256 Bit Message. Though, time complexity is being traded-off for security in the DES+RNS model. Figure 4. Encryption and Decryption Throughput at 256 Bit Message. Performance Chart of the 800 Bit Message Size Encryption and Decryption Time (800 Bit Message Size) The encryption and decryption time still shows that the DES model outperformed the RNS+DES model because the later uses a multi-level approach and will definitely take more time to run compared to DES only. In this work, time complexity is being traded-off for security as security is of more concern. Figure 5. Encryption and Decryption Time for 800 Bit Message 765.66472 137.74846 2126.5991 1513.71807 0 500 1000 1500 2000 2500 DES (Kilobyte/Seconds) RNS+DES (Kilobyte/Seconds) Encryption and Decryption Throughput at 256 Bit Size Message Encryption Throughput Decryption Throughput 0.261356 1.18305 0.107325 0.452618 0 0.2 0.4 0.6 0.8 1 1.2 1.4 DES (Seconds) RNS+DES (Seconds) Encryption and Decryption Time at 800 Bit Size Message Encryption Time Decryption Time Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 27 Encryption and Decryption Memory (800 Bit Message Size) The encryption and decryption memory shows the space complexity used for each of the algorithm to process their total algorithms operations. The encryption and decryption memory shows a better efficiency with the RNS+DES model as compared to the DES algorithm at 800 bit message size. Figure 6. Encryption and Decryption Memory at 800 Bit Message Encryption and Decryption Throughput (800 Bit Message Size) The throughput analysis based on the encryption time is measured by the size of the input text per seconds. (Kilobyte/Secs) while the throughput analysis based on the decryption time is measured by the size of the input text per seconds. (Kilobyte/Secs). The higher the encryption or decryption throughput validates the better model, as for the implemented models, the DES model outperformed the DES+RNS model at 800 Bit Message. In this work, time complexity is being traded-off for security. Figure 7. Encryption and Decryption Throughput at 800 Bit Message. Performance Chart of the 1472 Bit Message Size Encryption and Decryption Time (1472 Bit Message Size) 1.38E+09 1.36E+09 1.37E+09 1.36E+09 1.35E+09 1.36E+09 1.36E+09 1.37E+09 1.37E+09 1.38E+09 1.38E+09 DES (Byte) RNS+DES (Byte) Encryption and Decryption memory at 800 Bit Message Size Encryption Memory Decryption Memory 3060.91215 676.21825 7453.64763 1767.48708 0 1000 2000 3000 4000 5000 6000 7000 8000 DES (Kilobyte/Seconds) RNS+DES (Kilobyte/Seconds) Encryption and Decryption Throughput at 800 bit message size Encryption Throughput Decryption Throughput Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 28 The encryption and decryption time still shows that the DES model outperformed the RNS+DES model because the later uses a multi-level approach and will definitely take more time to run compared to DES only. In this work, time complexity is being traded-off for security. Figure 8. Encryption and Decryption Time for 1472 Bit Message Encryption and Decryption Memory (1472 Bit Message Size) The encryption and decryption memory shows the space complexity used for each of the algorithm to process their total algorithms operations. The encryption and decryption memory shows a better efficiency with the RNS+DES model as compared to the DES algorithm even at 1472 bit message size. Figure 9. Encryption and Decryption Memory at 1472 Bit Message Encryption and Decryption Throughput (1472 Bit Message Size) The throughput analysis based on the encryption time is measured by the size of the input image per seconds. (Kilobyte/Secs) while the throughput analysis based on the decryption time is measured by the size of the input image per seconds. (Kilobyte/Secs). The higher the encryption or decryption throughput validates the better 1.39E+09 1.38E+091.38E+09 1.33E+09 1.30E+09 1.32E+09 1.34E+09 1.36E+09 1.38E+09 1.40E+09 DES (Byte) RNS+DES (Byte) Encryption and Decryption memory at 1472 Bit Message Size Encryption Memory Decryption Memory Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 29 model, as for the implemented models, the DES model outperformed the DES+RNS model even at 1472 Bit Message. Though, Security is the concern of this work; thereby trading-off time complexity for security. Figure 10. Encryption and Decryption Throughput at 1472 Bit Message. Results Comparison of the proposed system with other related works As shown in the Table 4 Maurya (2020) reported that DES is less secured, Logunleko et al (2020) concluded that DES is not secured enough, Chourasia and Singh(2016) in their result discussed that DES is not secured but runs faster than RSA and Hybrid. Prerna and Abhishek also made it known that DES is not secure and is vulnerable to Linear and differential cryptanalysis. Adebayo et al in this research has been able to show that though DES takes a lesser time to run as compared to DES+RNS cryptosystem, but the later is excellently secured. Table 4. Result comparison of the proposed system with other related works Author Security/Vulnerability Of DES Time(DES) DES+RNS Maurya (2021) Less Secured NIL NIL Logunleko et al (2020) Not Secured Enough NIL NIL Chourasia and Singh (2016) Not Secured DES Outperforms RSA and Hybrid NIL Prerna and Abhishek (2013) Not Secured Enough/Brute Forced, Linear and differential cryptanalysis attack NIL NIL Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 30 5. Conclusion The DES+RNS model was developed in this thesis to cater for the security vulnerability of DES algorithm. The DES algorithm is susceptible to brute force, linear and differential attacks, Hence the need to work on its security. The designed model (i.e DES+RNS) is more secured as compared to conventional DES encryption. It is a multi- level encryption using DES with RNS. Apart from the security aspect worked on by the proposed technique, it also maximize the memory usage though time complexity serves as a trade-off. From the results, the DES+RNS model outperformed the DES model based on the aim of this thesis. Future study could focus on improving this model by increasing the key size. Another area of future research could be to build a fault-tolerant architecture that allows for redundant data to be added. The DES-RNS design can also be improved by incorporating the RNS inside the DES algorithm itself to address the time complexity issue. Reference Aldahdooh, R. M. (2018). Parallel Implementation and Analysis of Encryption Algorithms (Doctoral dissertation). Alhag, N. M. M., & Mohamed, Y. A. (2018, August). An Enhancement of Data Encryption Standards Algorithm (DES). In 2018 International Conference on Computer, Control, Electrical, and Electronics Engineering (ICCCEEE) (pp. 1-6). IEEE. Alhassan, I. Z., Ansong, E. D., Abdul-Salaam, G., & Alhassan, S. (2020). Enhancing Image Security during Transmission using Residue Number System and k-shuffle. Earthline Journal of Mathematical Sciences, 4(2), 399–424. https://doi.org/10.34198/ejms.4220.399424 Proposed Method Not Secured and Vulnerable to attacks Lesser time compared to DES + RNS Excellent Secured Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 31 Alhassan, S., & Gbolagade, K. (2013). Enhancement of the Security of a Digital Image using the Moduli Set. International Journal of Advanced Research in Computer Engineering & Technology (IJARCET), 2(7), 2223- 2229. Ali, A. H., Duhis, A. H., Alzurfi, N. A. L., & Mnati, M. J. (2019). Smart monitoring system for pressure regulator based on IOT. International Journal of Electrical and Computer Engineering, 9(5), 3450. Anbazhagan, P., Suruthisudar, M., Madhumitha, V., & Ezhilarasan, M. (2019). ENHANCED PRIVACY PRESERVING USING MULTILEVEL ENCRYPTION TECHNIQUE. Artiles, J. A., Chaves, D. P., & Pimentel, C. (2019). Image encryption using block cipher and chaotic sequences. Signal processing: image communication, 79, 24-31. Babatunde, A. N., Jimoh, R. G., & Gbolagade, K. A. (2016). An algorithm for a residue number system based video encryption system. Computer Science Series Journal, 14(2), 136-147. Behrouz, A. and Forouzan, D. (2012). “Cryptography and Network Security”, 2nd Edition, Tata McGraw Hill. Chourasia, S., & Singh, K. N. (2016). An efficient hybrid encryption technique based on DES and RSA for textual data. In Information Systems Design and Intelligent Applications (pp. 73-80). Springer, New Delhi. Ebrahim, M., Khan, S., & Khalid, U. B. (2014). Symmetric algorithm survey: a comparative analysis. arXiv preprint arXiv:1405.0398. Gbolagade, K. A., & Cotofana, S. D. (2009, May). An O (n) residue number system to mixed radix conversion technique. In 2009 IEEE International Symposium on Circuits and Systems (pp. 521-524). IEEE. Gupta, R., & Kapoor, R. (2016). A Review Paper of Data Security in Cloud Computing. Imperial Journal of Interdisciplinary Research (IJIR), 2(8), 501-504. Hasan, L. F., Ibrahim, A. A. E. A., & Syed, S. Z. B. (2021). Variable Rounds Block Cipher Algorithm Design. International Journal of Engineering and Artificial Intelligence, 2(2), 1-17. Heeks, R., & Wall, P. J. (2018). Critical realism and ICT4D research. The Electronic Journal of Information Systems in Developing Countries, 84(6), e12051. Hema, V., & Durga, M. G. (2014). Data integrity checking based on residue number system and chinese remainder theorem in cloud. International Journal of Innovative Research in Science Engineering and Technology, 3(3), 2584–2588. Hossain, M. J., Hossain, M. B., & Morshed, K. M. (2012, December). Reconfigurable encryption system: Encrypt digital data. In 2012 15th International Conference on Computer and Inf IBRAHIM, M. B., & GBOLAGADE, K. A. (2019). PERFORMANCE COMPARISON OF HUFFMAN CODING AND LEMPEL-ZIV-WELCH TEXT COMPRESSION ALGORITHMS WITH CHINESE REMAINDER THEOREM. University of Pitesti Scientific Bulletin Series: Electronics and Computer Science, 19(2), 7-12. Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 32 Jawahar, T. and Nagesh K. (2018) “DES, AES and Blowfish: Symmetric Key Cryptography Algorithms Simulation Based Performance Analysis,” in International Journal of Emerging Technology and Advanced Engineering (ISSN 2250-2459, Volume 1, Issue 2, December 2011), pp.6-12. Jincharadze, E. (2017). Critical analysis of some cryptography algorithms. Scientific and Practical Cyber Security Journal, 1(2), 60-68. Kaur, S., Bharadwaj, P., & Mankotia, S. (2017). Study of multi-level cryptography algorithm: multi-prime RSA and DES. International Journal of Computer Network and Information Security, 11(9), 22. Kumar, J., & Saxena, V. (2020). Hybridization of Cryptography for Security of Cloud Data. International Journal of Future Generation Communication and Networking, 13(4), 4007-4014. Kumar, S., Karnani, G., Gaur, M. S., & Mishra, A. (2021, April). Cloud security using hybrid cryptography algorithms. In 2021 2nd International conference on intelligent engineering and management (ICIEM) (pp. 599-604). IEEE. Kuppuswamy, N., Alspach, A., Uttamchandani, A., Creasey, S., Ikeda, T., & Tedrake, R. (2020, April). Soft- bubble grippers for robust and perceptive manipulation. In 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) (pp. 9917-9924). IEEE. Logunleko, K. B., Adeniji, O. D., & Logunleko, A. M. (2020). A comparative study of symmetric cryptography mechanism on DES AES and EB64 for information security. Int. J. Sci. Res. in Computer Science and Engineering, 8(1). Luen, Y. W. (1974). Knowledge of Mathematics and Science in Ching-hua-yüan. Oriens Extremus, 21(2), 217- 236. Mahajan, P., & Sachdeva, A. (2013). A study of encryption algorithms AES, DES and RSA for security. Global Journal of Computer Science and Technology. Mahmoud, A. Y., & Alqumboz, M. N. A. (2019, October). Encryption based on multilevel security for relational database EBMSR. In 2019 International Conference on Promising Electronic Technologies (ICPET) (pp. 130- 135). IEEE. Mahyar, H. (2012). Reliable and high-speed KASUMI block cipher by residue number system code. World Applied Sciences Journal, 17(9), 1149–1158. Maurya, S. (2021). Comparative Study of Security algorithm for Data Transfer. Muhamin, A., Bariu, K., Alagbe, K., Kumar, R. H. S., Aithal, G., Kumar, P. H. R., Sudeepa, K. B., Mahyar, H., Kar, A., Sur, K., Godara, S., Basak, S., Mukherjee, D., Sukla, A. S., Das, R., Choudhury, R., Alhassan, A., Saeed, I., Agbedemnab, P. A., … Alhassan, S. (2020). A New Cryptographic Model based on Residue Number System and Ribonucleic Acid Properties. Communications on Applied Electronics, 4(9), 1211–1219. https://doi.org/10.34198/ejms.4220.399424 Development of a Multi-Level Data Encryption Standard with Residue Number System for Data Security Adebayo et al. Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria Proceedings of the 3rd International Conference on ICT for National Development and Its Sustainability, May 21st – 24th, 2024. (ICT4NDS2024) pp. 15-34 33 Muhammed, K. J., & Gbolagade, K. A. (2019, December). Enhanced MORE algorithm for fully homomorphic encryption based on secret information moduli set. In 2019 IEEE International Conference on Industrial Engineering and Engineering Management (IEEM) (pp. 469-473). IEEE. Murali Krishna, B., Khan, H., Madhumati, G. L., Praveen Kumar, K., Tejaswini, G., Srikanth, M., & Ravali, P. (2017). FPGA implementation of des algorithm using DNA cryptography. Journal of Theoretical and Applied Information Technology, 95(10), 2147–2158. Musliyana, Z., Arif, T. Y., & Munadi, R. (2015). Security enhancement of Advanced Encryption Standard (AES) using time-based dynamic key generation. ARPN Journal of Engineering and Applied Sciences, 10(18), 8347– 8350. Nie, T., & Zhang, T. (2009, January). A study of DES and Blowfish encryption algorithm. In Tencon 2009-2009 IEEE Region 10 Conference (pp. 1-4). IEEE. Odunayo, T. K. (2020). Improving Advanced Encryption Standard Performance with Residue Number System (Doctoral dissertation, Kwara State University (Nigeria)). Pedamkar, P. (n.d.). DES Algorithm. EDUCBA. Retrieved June 18, 2022, from https://www.educba.com/des- algorithm Rahman, M. T., Pinandito, A., & Pramukantoro, E. S. (2017). Perbandingan Performansi Algoritme Kriptografi Advanced Encryption Standard (AES) dan Blowfish pada Text di Platform Android. Jurnal Pengembangan Teknologi Informasi dan Ilmu Komputer e-ISSN, 2548, 964X. Reyad, O., Mansour, H. M., Heshmat, M., & Zanaty, E. A. (2021). Key-Based Enhancement of Data Encryption Standard for Text Security. Proceedings - 2021 IEEE 4th National Computing Colleges Conference, NCCC 2021. https://doi.org/10.1109/NCCC49330.2021.9428818 Sa'adah, N., Astawa, I. G. P.,