2023
Browse
67 results
Search Results
Item Multi-agent best routing in high mobility digital-twin-driven internet of vehicles(IEEE IoT Journal, Rank Q1, 2023-10) Alam, Md Zahangir; S. Khan, Komal; Jamalipour, AbbasLow-delay high-gain optimal multi-hop routing path is crucial to guarantee both the latency and reliability require- ments for infotainment services in the high mobility internet of vehicles (IoVs) subject to queue stability. The high mobility in multi-hop IoVs reduces reliability and energy efficiency, and becomes bottleneck for the optimal route solution using classical optimization methods. To a great extent, deep reinforcement learning (DRL)-based method is not applicable in IoVs envi- ronment because of the continuously changing topology and space complexity, which grows exponentially with the number of state variables as well as the relaying hops. Usually, in multi- hop scenario, network reliability and latency are affected by mobility as well as average hop count, which limit the vehicle- to-vehicle (V2V) link connectivity. To cope with this problem, in this paper, we formulate a minimum hop count delay-sensitive buffer-aided optimization problem in a dynamic complex multi- hop vehicular topology using a digital twin-enabled dynamic coordination graph (DCG). Particularly, for the first time, a DCG-based multi-agent deep deterministic policy gradient (DCG- MADDPG) decentralized algorithm is proposed that combines the advantage of DCG and MADDPG to model continuously changing topology and find the optimal routing solutions by cooperative learning in the aforementioned communications. The proposed DCG-MADDPG coordinated learning trains each agent towards highly reliable and low latency optimal decision-making path solutions while maintaining queue stability and convergence on the way to a desired state. Experimental results reveal that the proposed coordinated learning algorithm outperforms the existing learning in terms of energy consumption and latency at less computational complexity.Item A Sustainable Approach to Establish Industry-Academia Collaboration by Engaging the Rural Community for the Developing Countries(Australasian Association for Engineering Education, 2023, Rank B, 2023-09) Kabir Peya, Md Mahmudul; Puspita, Afrin Hossain; Alam, Sabrina; Mahbubul Islam, Yousuf; Shahabuddin, A. M.; Hasan, MahadySkill and knowledge both play a vital role in sustainable career development. Universities were built to nurture knowledge thus the focus was on offering a knowledge-based curriculum. On the other hand, the industry's requirement is skill. Hence, the demand focuses on employees with real-world and hands-on skills. This gap between academia and industry impacts both students and industry. Furthermore, developing countries commonly lack the infrastructure needed to work with academia to support research that solves industry-specific problems. The motivation of the study is to work in rural development using engineering knowledge from academia. The research question driving the study is, "How can universities of developing countries mitigate the gap between theoretical learning and practical industrial skills?" Due to the rural environments, developing nations rely substantially on small-scale industries related to agriculture, fisheries, forestry, etc. The goal is to develop a model where students will be exposed to rural industry driven problems throughout their academic journey and work to solve the problems. Which will also prepare them for their future professional roles. The research utilizes case study approach by using data from students at Independent University, Bangladesh, where a three-credit Live-in-Field Experience course is in place. This course is a part of the foundational coursework and involves students living in rural areas, identifying issues in the rural sector, and formulating potential solutions. The effectiveness of this approach is evaluated incrementally, with successive student groups improving upon the previously devised solutions. The study shows that incorporating the model into the academic curriculum offers various outcomes. First, the model promotes experiential learning, where students solve real-world problems, particularly in rural areas. This change in teaching practice broadens theoretical concepts and their practical applications. The model creates bridges between academia and industry. Industry personnel will be interested in working and teaching in academia, which will help students get exposure to current industry practices and gain the required skills. The analysis proves that academic curricula that integrate industry-based problem-solving have an impact on both students and industries. The proposed model connects the bridge between academia and industry. Compared to the existing knowledge, this study could redefine our understanding of effective academic-industry collaborations and the role of universities in developing nations to adapt. The novel approach highlights the necessity of a change from knowledge-based education to one that emphasizes application and problem-solving skills.Item Initial Development and Performance Evaluation of a Bengali Voice-Operated Virtual Assistant for Personal Computer Control(International Scientific Conference on Information Technology and Management Science of Riga Technical University (ITMS), 2023, IEEE, 2023-10) Mark Quiah, Raven; Akter, Soma; Ahmed, Shad; Masudur Rahman, Mohammad; Alam, Sanzar Adnan; Islam, AshrafulThis paper presents the preliminary development and performance evaluation of a personal computer assistant designed for voice-operated interaction in Bengali. A bespoke phonetic grammar has been devised to map Bengali phonemes onto English representations, and an algorithm has been developed to mitigate the confusion of the machine in recognizing the same type of phonetics in the language, which helps to enhance the precision in comprehending Bengali commands. The system integrates the Microsoft speech synthesizer to articulate speech in Bengali, which facilitates the expansive human-computer interaction capabilities of the system, particularly for visually impaired individuals who face challenges in text-based computer interactions. The applications of this system extend among Bengali speakers and hold potential utility within rural regions of Bangladesh. The system also aims to replace the traditional mouse and keyboard for users with their own voice commands in their native language. Upon successful implementation of the system, it achieved 92.5% accuracy in the noise-free environment and 76.4% accuracy in the noisy environment with the help of 10 Bangladeshi individuals, including both males and females.Item A Comparative Overview of Local Mobile Financial Services Smartphone Apps Available in Bangladesh(International Scientific Conference on Information Technology and Management Science of Riga Technical University (ITMS), 2023, IEEE, 2023-10) Smaron, J,M, Sadik-Ul Islam; Tabassum, Yousra; Simoon, M.M.; Rahman, Zara; Rafid, Lishan; Islam, AshrafulThe rapid growth of mobile financial services (MFS) contributes toward the revolution of the financial landscape in developing economies, particularly Bangladesh. As a key player in promoting financial inclusion and the rising numbers of smartphone users, MFS smartphone apps have become increasingly popular in the country, catering to the needs of millions of unbanked and underbanked individuals. This paper aims to conduct a comprehensive comparative analysis of the various official MFS apps (n=13) available in Bangladesh, evaluating their features, functionalities, operational aspects, security measures, and overall facilities. The comprehensive exploration identified 18 distinct elements spanning four primary themes that stand out in the functionality of these apps: (1) Money Transfers and Transactions, (2) Financial Services and Bill Payments, (3) Service-related Charges, and (4) Consumer Finance. Each theme brings forth critical insights into the capabilities of these MFS apps, their user-friendliness, and their potential to effectively serve the unbanked and underbanked populations of Bangladesh. This comparative analysis is anticipated to provide valuable insights that can serve as a foundation for future advancements in MFS, fostering financial inclusion and promoting a more efficient and secure digital financial ecosystem in Bangladesh.Item Relational Agents for Type-2 Diabetes Management(The International Diabetes Federation (IDF), 2023, 2023-10) Islam, Ashraful; Chaudhry, Beenish; Islam, AminulRelational agents (RAs), often in the form of virtual humans, are designed to establish a long-term, social-emotional relationship with the user [1]. The technology has shown promise in areas like mental health, wellness, and chronic disease management [2]. RAs may be able to support Type-2 Diabetes (T2D) patients with management of their disease that often requires several lifestyle modifications and medication adherence [3]. However, RAs must be designed according to patients' needs and their success would depend on a patient's willingness to use and adopt this technology. To design a smartphone-based RA that will act as a social companion and a coach to help T2D patients self-manage their disease. The proposed RA is being developed using an iterative user-centered design process that will involve multiple user studies and a comprehensive literature review. The resulting RA will be capable of providing disease-specific recommendations to T2D patients that are personalized to their needs. RA’s logical operation can be simplified as: input -> process -> analyze -> output (Figure 1). Users can provide data to the RA using multiple modalities such as voice, text, touch, etc. Connected devices such as smartwatches or fitness bands can also feed data, such as vital sign information, into the RA, for continuous monitoring. The user data and T2D-specific information is stored in secure databases that are used during input and output data/interaction processing. The RA will analyze user’s queries and health conditions to provide guidance regarding medication personalized advice, reminders, and support for disease management tasks. It will also prompt the user to ensure they are able to implement these recommendations and supports. Integrating RAs into T2D management holds immense promise for revolutionizing the way individuals interact with their conditions and healthcare. In the future, further refinements of this RA will be crucial in evaluating its large-scale effectiveness and adoption.Item Assessing Early Stage Design of a mHealth App for Gestational Diabetes Mellitus Management in Bangladeshi Women(The International Diabetes Federation (IDF), 2023, 2023-10) Islam, Ashraful; AHMED, ESHTIAK; Zaman, Marzia; Rangon, Fairy Hasan; Amin, M Ashraful; Islam, RakibulThere is significant concern over the rising incidence of Gestational Diabetes Mellitus (GDM) among expectant mothers in Bangladesh [1]. Limited healthcare facilities in rural areas hinder prompt diagnosis and efficient management of GDM in Bangladesh. Despite mHealth's benefits, there is a lack of GDM management apps in Bangla, the native language of Bangladeshi citizens. To assess the viability of the first GDM management mHealth app in Bangla, users were asked about its early designs and functionalities. The app features a blood glucose tracker, food diary, medication reminder, educational resources, activity tracker, and personalized recommendations. 30 women with pre-existing GDM who were visiting a clinic in Dhaka, Bangladesh, were freely recruited during July 2023, and participation was anonymized. Participants ranged in age from 24 to 43 years (mean 33.43, SD 5.4). Following a briefing on the app's features and functionalities, participants were shown early sketches of the app. Later, they were prompted with a series of questions to provide feedback on the initial design and features. The majority (n=24) participants exhibited a positive response towards the app, expressing a wish that they had such a tool during their experience with GDM. However, 2 participants viewed the app as an impractical tool, while 3 were uncertain, expressing concerns about the accuracy of the information and guidance provided by the app. Beyond a textual interface, 1 participant suggested the inclusion of voice-based interaction to accommodate users who are illiterate, unfamiliar with using apps or having visual impairments. All participants appreciated the interface's use of the Bangla language and its cultural tailoring. 9 participants specifically highlighted the culturally tailored dietary recommendation feature as particularly praiseworthy. The app can play a crucial role in managing GDM in Bangladesh based on the early-stage evaluation feedback. However, further research is warranted to evaluate the real-world effectiveness and feasibility of it with a high-fidelity prototype for in-situ evaluation with the target users.Item Using an Ensemble Machine Learning Model with Explainable AI (XAI) to Diagnose Gestational Diabetes Mellitus(The International Diabetes Federation (IDF), 2023, 2023-10) Pasha, Syed Tangim; Islam, Ashraful; Sikder, Sanker; Habib, Md Tarek; Alam, Md Zahangir; Amin, M AshrafulThe emergence of gestational diabetes mellitus (GDM) in pregnant women is a serious health concern and an alarming issue. According to the most recent data from the International Diabetes Federation (IDF), in 2021, 16.7 percent of pregnant women had GDM, affecting 21.1 million live births [1]. Predictive models using an Explainable AI technique to detect GDM are currently unavailable, even though early detection can significantly reduce risks to human life. To develop an ensemble machine learning model with the XAI approach for diagnosing GDM. Our study used a 1,012 GDM patient records dataset with 7 attributes sourced from [1]. Due to their unsuitability for our studies, attributes like ‘Age’ and ‘Pregnancy No.’ were omitted from the dataset. The ‘Height’ attribute was also eliminated because of its negative correlation with the ‘BMI’ feature. We used the Synthetic Minority Oversampling Technique (SMOTE) to address imbalanced class issues in the target attributes after performing feature scaling on the remaining attributes. Our strategy required developing a Stacking Ensemble model that integrated other models, including the Decision Tree, Random Forest, XGBoost, MLP, and Logistic Regression. We employed metrics, e.g., Receiver Operating Characteristic (ROC) curve, Area under the ROC Curve (AUC), and SHapley Additive exPlanations (SHAP) values to evaluate the model's effectiveness. 70% of the dataset was used for training and 30% for testing. We achieved 85% accuracy with an AUC score of 0.91 in the experiment, and the ROC curve is shown as the performance curve in Fig. 1(a). The feature plot in Fig. 1(b) shows that the ‘Heredity’ feature is more important than the ‘Weight’ and ‘BMI’ features, whereas the summary plot in Fig. 1(b) combines feature effects and importance. Findings show that 'Heredity' has a high and positive impact on predicting GDM in this dataset whereas 'Weight' and 'BMI' have a positive impact but are lower than 'Heredity'. We developed an XAI approach-based ensemble machine learning model to diagnose GDM.Item A Federated Learning Approach for Type-2 Diabetes Detection Using a Naive Bayes Classifier(, The International Diabetes Federation (IDF), 2023, 2023-10) Rahman, M. M.; Islam, Ashraful; Pasha, Syed Tangim; Islam, M. Usama; Alam, Md ZahangirFederated learning (FL) is a new way of training machine learning models across decentralized devices without exchanging the raw data. This approach preserves privacy and promotes the development of more personalized models by exploiting the heterogeneity of data. Common phenomena of FL is to employ deep learning models. Nonetheless, simple machine learning models such as Naive Bayes have promising potentials for detecting diabetes mellitus in a FL environment. To explore the practical prospects of building a privacy-preserving model for identifying patients with diabetes mellitus, utilizing their individual data. A cohort of 103 persons are enrolled in this study. Each participant was sent a questionnaire to answer with their own personal data about their age, Body Mass Index (BMI), insulin level, glucose concentration, skin thickness of an individual. Subsequently, an initial model, built using the Pima Indian Diabetes dataset, was sent to their mobile devices [1]. The participants utilized the initial model to train with their own data. Following this, the model parameters are updated and sent to the server. The server aggregated the parameters and averaged them to make a global model. This completes a single iteration of federated learning life cycle. Participants are diversed in gender: male (60.2%) and female (39.8%); in age groups: 20-35 (14.6%), 36-50 (46.6%), 51-65 (38.8%). The work shows an accuracy of 89.32% and a precision of 88.89% for those having diabetes while 90.32% precision in detecting patients not having a diabetes mellitus. The number of communication rounds was 50 where in each round at least two participants participants in building federated model updates. Since one of the key reasons for using FL is to improve data privacy, quantifying the level of privacy is critical. An network intruder could decoded the model updates by examining the changes in the global model over time. However, a membership inference attack (MIA) is measured in various differential privacy (DP) budgets. DP aims to prevent this kind of inference by adding noise to the data (model updates). For instance, if the model update would normally be a weight change of +0.5, a noise from a Laplacian distribution with mean 0 is added. Hence, the resulting noisy update might then be +0.52. A naive bayes based federated learning system is built to detect diabetes mellitus (Type-2) preserving the privacy of user data at the first place.Item Low-cost relay selection in multihop cooperative networks(Journal of King Saud University - Computer and Information Sciences, Q1, 2023-09) Rahman, Suryaia; Alam, Md Zahangir; Islam, Ashraful; Habib, Md. Tarek; AHMED, ESHTIAK; Hasan, Mahady; Ahmed, TaremA best relay selection algorithm for a cooperative multi-hop cross-layer single-input single-output (SISO) amplify-and-forward (AF) wireless relay network is analyzed in this work, with the application where exact channel state information (CSI) is known. We present algorithmic strategies to simplify a multihop parallel SISO relay network into a series multi-hop network by finding the best path having the maximum received signal-to-noise ratio (SNR). The best relay selection by using dynamic programming search entails high computations and large memory requirements, as well as involves the full CSI information, making this approach impractical for large-scale networks. The goal of the proposed low-cost near-optimal routing strategy in this work is to provide close to optimal performance with much less complexity compared to traditional routing. Next, we also propose a low-cost power allocation to further improve system performance over traditional power allocation found in the literature. Computer simulations show excellent performance improvement of our proposed methods in terms of bit error rate (BER) as well as outage probability over the traditional relaying and power allocation algorithms.Item Repercussion of Image Compression on Satellite Image Classification using Deep Learning Models(Independent University, Bangladesh, 2023-08) Hossain, Md. Junayed; Barkatullah, Mohammad; Monir, Md. Fahad; Ahmed, TaremSatellite image classification using Deep Learning (DL) algorithms is crucial for various applications such as Environmental Monitoring, Remote Sensing, and Urban Planning. The high-resolution nature of satellite images leads to large data volumes, resulting in computing challenges when classifying and transmitting the data over the internet. In this paper, we address this challenge by reducing pixel size using Bicubic and Bilinear Interpolation. These techniques are known as image compression methods, which allow us to retain a significant portion of the original image information while reducing its size. Furthermore, adjusting the pixel size enables us to implement various levels of image compression supported by different DL classifiers, catering to diverse applications in NextG wireless networks or O-RAN. This pixel size reduction optimizes data transfer and speeds up satellite image transmission for Crop Monitoring and Land Cover Classification applications. After reducing the pixel, we employ different DL models such as CNN, ResNet, and stacked ensemble model, built by stacking three other EfficientNetB0 models to classify the low pixel images and compare the results using accuracy, precision, recall, and F1-Score. Staked EfficientNetB0 is the best performer among these, with an accuracy of 89.34% using a pixel size of 512x512. Our results show that we can reach up to 88.25% accuracy while maintaining a pixel size of 64x64, as our target is to reduce the image size with maintaining a tolerable accuracy so that network load and cost can be minimized.