Emerging Technologies in Computing

Liquid crystal (LC) has proven to be a promising material for microwave (µWave) phase shifters at GHz ranges, due to their continuous and wide tunability, as well as reasonably low absorption loss. However, designing LC phase shifters that meet specific application requirements (e.g., SpaceTech) is a challenging task that entails a complex trade-off between various parameters. Physics-informed machine learning (PI-ML) combines the power of machine learning with the underlying physics to develop a more accurate and interpretable model. Leveraging PI-ML to inform LC µWave device design is a relatively new area, with tremendous opportunities for exploration and innovation. In this article, a deep learning assisted LC µWave phase shifter design and synthesis framework is proposed. By incorporating physical constraints and knowledge into deep neural networks, one can effectively balance the trade-off between different design parameters and synthesize LC phase shifter structures that meet specific performance requirements (e.g., insertion loss, insertion loss balancing, phase tuning range, tuning speed, power consumption). The framework is envisaged to allow for the efficient and effective exploration of the design space, resulting in improved accuracy and efficiency compared to traditional two-stage design methods.

Earthquake prediction remains a major challenge in the field of geophysics, with significant implications for disaster management and risk reduction. Accurate prediction depends on identifying and selecting relevant features from large and complex datasets. In this study, we present a novel feature selection method, the Chaotic Chimp based African Vulture Optimization Algorithm (CCAVO), applied to earthquake magnitude prediction. The model was trained on a dataset containing various seismic event characteristics such as latitude, longitude, depth, and other geological factors. The target variable for prediction was the magnitude of the seismic event. We conducted three stability tests on the model: Convergence Rate, Consistency Test, and Sensitivity to Parameters. Our analysis revealed that the CCAVO demonstrated good convergence behavior, with training errors reducing over successive iterations, indicating the model’s ability to learn from the data. The consistency test further showed that the model performance, as quantified by the Mean Squared Error (MSE), remained consistent across multiple runs with different random seeds, suggesting the model’s stability and robustness against randomness in initialization. Finally, a sensitivity analysis was performed to examine the model’s response to changes in its hyperparameters. The model’s performance was observed to vary with different parameter settings, indicating its sensitivity to hyperparameters. The optimal parameters found were a learning rate of 0.1 and 100 estimators, yielding 0.08 MSE from 3-fold cross-validated MSE.

Recommender Systems (RS) are information retrieval systems that can be used for serving personalized content to online users. Most industrial recommendation systems utilize a large amount of online data to generate personalized recommendations for users. The quality of the data plays an important role in the performance of the RS. The majority of the RS data is generated from event data that are stored in data lakes through multiple data pipelines. Event-based data pipelines have emerged as a popular approach to handle the massive amount of data generated by modern applications. In this paper, we explore the impact of event-based data pipelines on recommendation systems. We discuss how these pipelines enable efficient data ingestion, real-time processing, and low-latency recommendations.

This paper presents a proof-of-concept study on AI-based pre-surgery planning in plastic surgery. The study addresses the challenge of technique selection by developing an AI-driven system that utilises machine learning algorithms to analyse patient-specific data and historical outcomes. By comparing and evaluating diverse inputs, the system generates detailed results for each technique, providing surgeons with valuable insights into expected outcomes. This enhances decision-making during pre-surgery planning and improves surgical precision. The system’s development involved addressing challenges related to data availability, algorithm selection, and interpretability. Preoperative images will be processed using advanced computer vision algorithms to extract relevant features. A Convolutional Neural Network (CNN) architecture predicted technique-specific outcomes based on the extracted features. The validation included comparing predictions against ground truth data and expert evaluations. Feedback from plastic surgery practitioners will be collected to assess usability and practicality. Ethical guidelines will be strictly followed to ensure patient data protection and address potential biases. The successful implementation of the proof of concept demonstrates the potential of AI integration in pre-surgery planning for plastic surgery. By empowering surgeons with technique-specific insights, the system enhances decision-making, ultimately improving patient care and treatment outcomes. Future work involves expanding the dataset, considering additional variables, and conducting prospective clinical trials to validate the system’s real-world impact.

The increasing sophistication of generative language models and their widespread accessibility to the general public has been a cause of growing concern in academia in recent years. While these AI technologies have the potential to greatly enhance the learning experience and facilitate research, they also pose a significant threat to academic integrity.

This paper investigates the impact of using tools like chatGPT and other large language models (LLM) in higher education, discussing their potential benefits while focusing more on assessing the risks, including the possibility of plagiarism, cheating, and other types of academic misconduct. It explores how these technologies may be used to undermine established scholarly principles and practices, as well as the challenges of identifying and combating academic dishonesty. Some measures universities and academics may employ in order to mitigate such risks are proposed, and several strategies and tools for detecting AI-generated content are discussed, along with their limitations.

Nowadays, Machine Learning (ML) plays a crucial role in improving healthcare by enabling researchers, doctors, and patients to explore, diagnose, and prevent diseases such as dengue, typhoid, jaundice, pneumonia, and other major ailments. Our research focuses on leveraging ML to detect various diseases from a patient’s speech. The patient will describe their symptoms to the machine, akin to explaining their concerns to a doctor. The machine will then identify the disease and provide primary medication recommendations along with suggesting a specialized doctor for that particular ailment. To optimize our system’s performance, we trained our machine using multiple algorithms and evaluated their results. Our evaluation revealed an accuracy of 86.59% for Naive Bayes, 83.17% for Unhyperd SVM, 98.05% for Hyperd SVM, 97.4% for Decision Tree, and the highest accuracy of 99.35% was achieved by Random Forest.

How COVID-19 has impacted digitalisation in high schools is presented. The scale of the utilisation of Information Systems (ISs) is presented as statistical study which addresses a notable gap in the published literature. The study focussed on the impact of the adoption of LMS (Learning Management System) and other related variables in digitalisation in higher education with a comparison made before and after the pandemic. The research also sought to understand the level of LMS usage by the administration during the pandemic. The research methodology employed was a triangulation technique, which combined qualitative and quantitative methods. This approach allowed for a comprehensive investigation of the research questions. The quantitative method involved the use of questionnaires to acquire data from a large number of participants, whilst the qualitative method involved the use of focus groups to gain more in-depth insights into the experiences and perceptions of the participants. The study used statistical data analyses: ANOVA; one sample t-test; crosstab comparison; Bayesian factor and estimates – to analyse the data collected. These analyses provided valuable insights into the impact of the pandemic on the digitalisation process in high schools. These being to provide recommendations for enhancing the digitalisation process in high schools such as guiding educators, school administrators and policymakers in their efforts to improve the digital learning experience for high school students in the context of the ongoing pandemic and beyond. The study concluded that the digitalisation process has improved after the pandemic.

Lightweight cryptography aims to design secure and efficient cryptographic algorithms for resource-constrained devices. Traditional cryptographic algorithms may not be readily usable in resource-constrained environments. To standardise cryptographic solutions tailored for such resource-constraint environments, the National Institute of Standards and Technology (NIST) launched the Lightweight Cryptography (LWC) project. Grain-128AEAD is a stream cipher-based finalist in the NIST LWC project. In this work, we examine the security of the initial version of Grain-128AEAD against cube attacks. We present distinguishing attacks on a reduced-round version of the cipher, assuming that the keystream can be observed immediately after the reduced-round initialisation of the pre-output generator. We obtained various cubes of sizes 25 to 45 for reduced-round Grain-128AEAD. The best cube reported in this work can distinguish the output of a 165-round initialisation of Grain-128AEAD with a cube size of 35. The complexity of the distinguishing attack is \(\mathcal {O}(2^{35})\). The results are confirmed experimentally. We conclude that even with fewer rounds of initialisation for the first version of Grain-128AEAD, the cipher still has a good security margin against cube attacks.

Malware disrupts the natural behaviour of computer systems, hinders performance, and may cause a significant loss to the computer system owner. The growth or advancement in the number of malware variants has necessitated the requirement of advanced techniques for the detection, identification, and classification of malware. The hybrid approach is predominantly employed since static and dynamic analysis methods have drawbacks and are time-consuming. Moreover, recent malware variants use obfuscation techniques and exhibit polymorphic and metamorphic behaviour. It was noticed that even though classical machine learning methods gave better performance and quicker classification, they suffered from the problem of misclassification. Newer approaches such as image processing techniques and deep learning architectures are thus employed. The paper focuses on the survey of various detection, identification, and classification methods of malware and is an effort to put forward the best approach.

DES has been serving as the official federal standard for data encryption since the 70s, until it was replaced in 2001 by the Advanced Encryption Standard (AES). This paper thoroughly examines the encryption and decryption procedure, describing in detail and illustrating each step of the process. Additionally, we showcase the practical implementation of DES using the CrypTool platform, addressing the lack of a comprehensive and user-friendly tutorial on DES within this context. The paper endeavors to address this gap, enhancing the readers’ understanding of the DES operation by delving into its internal mechanism and the transformations the plaintext undergoes to become ciphertext. In addition, we engage in a critical examination of CrypTool's implementation of DES and the platform’s utility as an educational resource, highlighting both its strengths and shortcomings. The review of CrypTool as a learning tool not only provides insights into the practical implementation of DES but also emphasizes the significance of hands-on learning in modern cryptography. Moreover, we present a historical analysis of attempts at breaking the DES cipher, from its approval as a standard until more recent developments, and assess its relevance in today's cryptographic landscape. Ultimately, this paper aspires to serve as a valuable teaching resource for cryptography students and educators, bridging the gap between theoretical knowledge and practical application.

The aim of this paper is to investigate and discuss issues on the design and performance evaluation of Android applications concerning alarming systems that would serve users in emergency or dangerous situations to be able to send text messages in real time giving information about their location and about their needs in searching for help. The user, for instance, could send a message to their chosen contacts just by shaking the phone a predefined number of times or following any other pattern of action involving the smartphone. The application designed has, also, some other features like sounding an alarm for 10 s, automatically opening the hospital map searching for hospitals in the area and automatically opening the police map searching for police stations nearby. While opening each of the maps mentioned, the user could check their location as well. Many other similar features could be added in the designed application. But the main goal of the paper is to investigate design issues in Android applications regarding real time performance as well as to evaluate performance in such time critical applications. Therefore, the contribution of this paper lies on comparing real time performance of different implementations of such an alarming application as well as on presenting step by step design, architectural and implementation issues of it.

Air pollution poses a significant issue in numerous cities worldwide, impacting public health and the environment. We study three significant cities under the Dhaka division, including Kuril Bishow Road, Uttara, and Tongi. Traditional air quality monitoring methods often need more coverage and accuracy. Leveraging Internet of Things (IoT) technology as well as machine learning (ML) algorithms, this study deploys an IoT-based sensor network using Arduino boards and various devices, including MQ135, DHT22, PM2.5, MQ9, and dust sensors to gather real-time values on air pollutants. The gathered data, including sulfur dioxide, ozone, particulate matter 2.5, nitrogen dioxide, particulate matter 10, as well as carbon monoxide, provides a comprehensive view of city pollution levels. ML models such as linear regression, decision trees, K-Nearest Neighbors (KNN), Naive Bayes (NB), Gradient Boosting (GB), and Random Forest classifiers are applied to predict pollution levels using environmental parameters. The Random Forest classifier achieves an impressive prediction accuracy of \(97.2\%\). Evaluation metrics, including precision, recall, F1 score, Kappa score, mean square error (MSE), root mean square error(RMSE), along mean absolute error (MAE), are used to assess the performance of the models. This study demonstrates the potential of IoT technology along with ML algorithms in accurately predicting air pollution levels, aiding in environmental management and public health efforts in urban areas.

Although blockchain technology was first introduced in 2008 and materialised in 2009, the early usage of blockchain were mainly limited to financial technologies, particularly cryptocurrencies. Later, blockchain became a widespread emerging technology, utilised in multifaceted sectors and applications. In fact, various new and innovative application of blockchain and distributed ledger technologies are still continuously being researched and explored. On the other hand, smart-contracts were first introduced in 1990s, however, it did not gain enough popularity until being integrated with blockchain technologies lately. The duo lately been seen as the key to many innovations in various industries and sectors. So, we took data from 1445 blockchain-related patent documents and tried to map out the historical and current trends in patenting activities in the blockchain field. This helps us get a better grasp of how blockchain technologies are evolving and being tracked. In addition to serving as an indicator of science and technology growth, patents are also used to judge the research potential and development of a particular technology.

Social media has revolutionized the way people communicate, but it also comes with risks such as the proliferation of misinformation and cyberbullying and violations of community guidelines. To address these concerns, we propose the development of a social networking platform, called SocialEcho, which leverages a combination of advanced filtering techniques and third-party API services, to ensure that all user-generated content complies with established community guidelines. Our platform built using modern tools to create a regulated and safe space for online communication that promotes responsible and respectful interactions among users. In addition, we have implemented an NLP-based feature that automatically categorizes user posts into different channels, enabling users to easily find and engage with content that aligns with their interests. Through these features, SocialEcho enhances the user experience and promotes a more organized and structured community.

Nowadays fields are seen everywhere around us like in schools, colleges, universities, parks, stadiums, and many more places. But the maintenance cost is very expensive and they also consume many resources like electricity, water, fuel, manpower, etc. Our main target is to use those resources efficiently and create an automation system for the field. The traditional maintenance system of a field requires a lot of workers and they need to handle everything manually. They water randomly in the field without worrying about which part of the field will need water. This process wastes a lot of water and electricity. Also, we want to implement a system that measures grass growth in that field. Workers do not use the grass trimmer efficiently because they don’t know where the grass is actually grown. So, a lot of fuel or electricity is wasted during that process. We want to overcome those problems. Our proposed idea is that we will create an automation system that will detect field moisture, grass growth and it will report the field condition to a device.