Smart Computing and Systems Engineering - 2021 (SCSE 2021)

Permanent URI for this collectionhttp://repository.kln.ac.lk/handle/123456789/25343

Browse

Now showing 1 - 20 of 41

A novel approach for weather prediction for agriculture in Sri Lanka using Machine Learning techniques
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Premachandra, J. S. A. N. W.; Kumara, P. P. N. V.
Climate variability in recent years has critically affected the usual aspects of human lives, where the agriculture sector can be considered as one of the most vulnerable. Sri Lanka is also facing these climate changes over the past few decades. It has resulted in rainfall pattern changes where the expected rain may not occur during the expected time and amount. The mismatch between the rainfall pattern and traditional seasonal cultivation schedule has critically affected the agricultural sustainability. Even with the current technological advancements, weather prediction is one of the most technically and scientifically challenging tasks. This paper presents a novel machine learning-based approach for predicting rainfall for precision agriculture in Sri Lanka and it can be recognized as the first attempt to validate machine learning models to predict the weather in Sri Lankan context for precision agriculture. By analyzing the nature of the weather in Sri Lanka, the relationship of weather attributes with agriculture, availability, and accessibility, seven attributes are selected including rain gauge, relative humidity, average temperature, wind speed, wind direction where solar radiation and ozone concentration are uniquely selected for Sri Lankan context. For the prediction model, cross-validated data are trained and tested with four machine learning algorithms: Multiple Linear Regression, K-Nearest Neighbors, Support Vector Machine, and Random Forest. Currently, Support Vector Machine, K-Nearest Neighbors models have achieved accuracies of 88.57%, 88.66%. Random Forest has been recognized as the best-fitted model with 89.16% accuracy. The results depict a significant accuracy in this novel approach for Sri Lankan weather prediction.
Comparison of supervised learning-based indoor localization techniques for smart building applications
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Maduraga, M. W. P.; Abeysekara, Ruvan
Smart buildings involve modern applications of the Internet of Things (IoT). Intelligent buildings could include applications based on indoor localization, such as tracking the real-time location of humans inside the building using sensors. Mobile sensor nodes can emit electromagnetic signals in an ambient sensor network, and fixed sensors in the same network can detect the Received Signal Strength (RSS) from its mobile sensor nodes. However, many works exist for RSS-based indoor localization that use deterministic algorithms. It's complicated to suggest a generated mechanism for any indoor localization application due to the fluctuation of RSSI values. This paper has investigated supervised machine learning algorithms to obtain the accurate location of an object with the aid of Received Signal Strengths Indicator (RSSI) values measured through sensors. An available RSSI data set was trained using multiple supervised learning algorithms to predict the location and their average algorithm errors were compared.
Simulation analysis of an expressway toll plaza
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Grabau, Shehara; Hewapathirana, Isuru
Since the early civilizations, transportation has played a significant role, from fulfilling basic human needs to contributing towards major economic growths all over the world. With the advancement in technology, the demand for smooth and hassle-free transportation increased and it is particularly true for road transportation in Sri Lanka as well. As a result, the expressway road network was introduced to Sri Lanka in 2011. Although a toll is payable for the use of expressways, many vehicle users prefer to utilize the expressway due to the extensive amount of time saved. Time is of utmost importance for expressway users. Hence, long queues and waiting time at toll plazas where the toll payment is made should be minimized. This study is aimed at analyzing the performance at the Peliyagoda toll plaza of the Colombo-Katunayake expressway where the formation of long queues and long waiting time in queues can be observed during peak hours. Due to the high complexity of using the analytical approach in obtaining the performance measures, a simulation approach was used with Arena Simulation Software. Few setup improvements were identified, and each of the setups were simulated to obtain the performance measures. Based on the comparison of the results, recommendations and suggestions to improve the efficiency of the operations at the Peliyagoda toll plaza have been outlined.
Vibration analysis to detect and locate engine misfires
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Jayasooriya, Prathap V.; Siriwardana, Geethal C.; Bandara, Tharaka R.
Vibration analysis is used to detect faults and anomalies in machinery and other mechanical systems that produce vibrations during operation. The study aimed to develop an algorithm that can detect and locate engine faults in automobiles by analyzing vibrational data produced during engine operation. Analysis was done on one type of engine fault – Spark Ignition Engine misfire. To detect anomalies in the vibrational pattern (waveform), analysis was carried out in both time and frequency domains. To obtain vibrational data an AVR – 32 (Arduino) based data acquisition device was built, and analysis was carried out in MATLAB using scripts and functions. The developed algorithm isolates frequency components in the waveform that corresponds to engine faults and converts them into numerical quantities that are then compared with computed ranges. The algorithm was able to identify the presence of a misfire in the engine and could locate the cylinder in which the misfire occurs with significant accuracy.
Student concentration level monitoring system based on deep convolutional neural Network
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Shamika, U. B. P.; Panduwawala, P. K. P.G.; Weerakoon, W. A. C.; Dilanka, K. A. P.
As synchronous online classrooms have grown more common in recent years, evaluating a student's attention level has become increasingly important in verifying every student's progress in an online classroom setting. This paper describes a study that used machine learning models to monitor student attentiveness to distinct gradients of engagement level. Initially, the experiments were conducted using a deep convolutional neural network of student attention and emotions exploiting Keras library. The model showed a 90% accuracy in predicting attention level of the student. This deep convolutional neural network analysis aids in identifying crucial emotions that are important in determining various levels of involvement. This study discovered that emotions such as calm, happiness, surprise, and fear are important in determining a student's attention level. These findings aided in the earlier discovery of students with poor attention levels, allowing instructors to focus their assistance and advice on the students who require it, resulting in a better online learning environment.
A tree structure-based classification of diabetic retinopathy stages using convolutional neural network
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Peiris, M. S. H.; Sotheeswaran, S.
Detection, and classification of medical images have become a trending field of study during the last few decades. There is a considerable amount of vital challenges to be overcome. Ample work has been carried out to provide proper solutions for those key challenges. This study was carried out to extend one such medical image classification process to classify the stages of Diabetic Retinopathy (DR) images from colour fundus images. The study proposes a novel Convolutional Neural Network (CNN) architecture which is considered to be one of the most trending and efficient forms of classification of DR stages. Initially, the pre-processing techniques were employed to the DR fundus images with Green channel extraction and Contrast Limited Adaptive Histogram Equalization (CLAHE). The data augmentation strategy was utilised to increase training images from the DR images. Finally, Feature extraction and classification were carried out by using the proposed CNN architecture. It consists of a 14 layered CNN model, which continues three main classifications. In this proposed classification, the images were classified into a tree structure based binary classification as No_DR and DR at the beginning, and then the DR images were again classified into two classes, namely Pre_Intermediate and Post_Intermediate. Moreover, those two classes were again separately classified into Mild, Moderate, and Proliferate_DR, Severe, respectively. The Kaggle is one of the benchmark dataset repositories which was used in this study. The proposed model was able to achieve accuracies of 81%, 96%, 84%, and 97% for the above-mentioned classifications, respectively.
A decentralized social network architecture
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Sarathchandra, Tharuka; Jayawikrama, Damith
Billions of people use social networks, and they play a significant role in people's lifestyles in the current world. At the same time, due to globalization and other factors, the use of these social platforms is expanding daily, and a variety of activities take place inside these platforms. These networks are centralized, allowing social network-owned companies to track and observe the activities of their users. Therefore, this has been challenged to the privacy of the data of users. Also, these companies tend to sell them to third parties keeping huge profits without users' permission. Since data is the most valuable asset in today's and tomorrow's world, many have pointed out this issue. Even though decentralized, community-driven applications have come to play as a solution to this problem, there is still no successful application that competes with centralized social network platforms. Therefore, this study attempted to develop a decentralized social network architecture with the basic functionalities of a social media platform to assure the privacy of the users' data.
Thought identification through visual stimuli presentation from a commercially available EEG device
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Gunawardhana, M. P. A. V.; Jayatissa, C. A. N. W. K.; Seneviratne, J. A.
Thought identification has been the ultimate goal of brain-computer interface systems. However, due to the complex nature of brain signals, classification is difficult. But recent developments in deep learning have made the classification of multivariate time series data relatively easy. Studies have been carried out in the recent past to classify thoughts based on signals from medical-grade EEG devices. This study explores the possibility of thought identification using a commercially available EEG device using deep learning techniques. The crucial part of any EEG experiment is contamination-free data collection. Keeping the subject’s mind concentrated only in the decided state is important, yet challenging. To address this issue, we have developed a graphical user interface (GUI) based program that allows stimulus controlling and data recording. With the use of the low-cost commercially available EEG device, accuracies up to 89% were achieved for the classification of high contrast signals. However, tests on complex thought identification did not produce statistically significant results over the chance accuracy.
Application of AlexNet convolutional neural network architecture-based transfer learning for automated recognition of casting surface defects
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Thalagala, Shiron; Walgampay, Chamila
Automated inspection of surface defects is beneficial for casting product manufacturers in terms of inspection cost and time, which ultimately affect overall business performance. Intelligent systems that are capable of image classification are widely applied in visual inspection as a major component of modern smart manufacturing. Image classification tasks performed by Convolutional Neural Networks (CNNs) have recently shown significant performance over the conventional machine learning techniques. Particularly, AlexNet CNN architecture, which was proposed at the early stages of the development of CNN architectures, shows outstanding performance. In this paper, we investigate the application of AlexNet CNN architecture-based transfer learning for the classification of casting surface defects. We used a dataset containing casting surface defect images of a pump impeller for testing the performance. We examined four experimental schemes where the degree of the knowledge obtained from the pre-trained model is varied in each experiment. Furthermore, using a simple grid search method we explored the best overall setting for two crucial hyperparameters. Our results show that despite the simple architecture, AlexNet with transfer learning can be successfully applied for the recognition of casting surface defects of the pump impeller.
Reduce food crop wastage with hyperledger fabric-based food supply chain
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Premarathna, Dewmini
Food is the utmost important thing for every living being. The quality and safety of food has become a crucial factor in the food industry. Most of the customers tend to pay more attention to food safety and seek to get food from verifiable resources. To improve this trustworthiness Distributed Ledger Technology (DLT) - based Food Supply Chain (FSC) plays a vital role because of its traceability. There are multiple actors involved throughout the journey of FSC and with the high visibility of data in DLT, everyone can ensure trust. The transparency of data itself is a reason for some to opt-out because some of their private data can be exposed to others. Hyperledger Fabric (HF) based FSC can address that matter as it supports permissioned network solutions. Though there are a lot of solutions available in a similar kind of approach, whether the crops take their journey throughout the FSC without any wastage, is still questionable. This study focuses on reducing wastage of food crops as they take a long journey in their raw state and possible hazards are high. It discusses farmers' behavior based on the Sri Lankan context and how it accompanies food crop wastage. Further, this paper ruminates the other possible crop wastage that can take place in FSC and how to eliminate it with the proper involvement of knowledgeable and authorized parties. Then, the study explores how all the parties can collaboratively join the FSC based on HF so that everyone can benefit. Finally, it concludes on how such design is effectively contributing to reducing food crop wastage in Sri Lanka (SL).
A community-based hybrid blockchain architecture for the organic food supply chain
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Thanujan, Thanushya; Rajapakse, Chathura; Wickramaarachchi, Dilani
This paper presents a novel blockchain architecture to incorporate community-level trust into the organic food supply chain by hybridizing Proof of Authority (PoA) and Federated Byzantine Agreement (FBA) consensus protocols. Community-level trust is an important aspect in the organic agriculture industry. Organic farming, in most parts of the world, happens in small scale farms where the farmers represent rural and less-privileged communities. Even though third-party certification systems exist for quality assurance in organic farming, due to many socio-economic reasons, participatory guarantee systems (PGS) have become a popular alternative among organic farmers and consumers. However, such participatory guarantee systems are still prone to frauds and have limitations in scalability as well. With the recent rise of blockchain technology, there is an emerging trend to adopt blockchain technology to enhance the credibility of organic food supply chains and mitigate the risk of fraudulent transactions. However, despite the popularity of participatory guarantee systems among organic farmer communities, the blockchain researchers have paid little attention to develop blockchain architectures by adopting the community-level trust into their consensus protocols. The hybrid consensus mechanism presented in this paper addresses that gap in existing blockchain research. Apart from discussing the details of the proposed blockchain architecture and the underlying consensus protocol, this paper also presents a qualitative analysis on the proposed architecture based on expert opinions.
What makes job satisfaction in the information technology industry?
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Arambepola, Nimasha; Munasinghe, Lankeshwara
Having a rich human resource is critical for an organization to move towards success. Especially, for business organizations such as technology companies, the human resource is the driving factor of the company's growth which depends on employees' motivation, skills and quality of work. Employees often change their jobs when they are not satisfied with it. Different factors may cause a change in the level of job satisfaction of an employee. For example, the dynamic nature of the Information Technology (IT) industry is an impactful factor that determines the job satisfaction of IT professionals. Foreseeing the employees' job satisfaction makes it easy for a company to take swift actions to improve the job satisfaction of its employees. In this research, we analyzed the effectiveness of machine learning (ML) methods for predicting job satisfaction using employee job profiles. There are job-specific factors in each job domain, and those factors may influence job satisfaction levels. Therefore, this research focused on the following fundamental questions: 1) How do existing ML models perform when predicting job satisfaction of software developers? 2) Can the job satisfaction prediction models be generalized to the other job roles in the IT industry? This study compared the performance of classification models: Random Forest (RF), Logistic Regression (LR), Support Vector Machine (SVM), and Neural Network (NN) in predicting the level of job satisfaction. Our experiments used two benchmark datasets: Stack Overflow developer survey and IBM HR analytics dataset. The experimental analysis shows that both employee-related factors and company-related factors contribute similarly to predicting job satisfaction. On average, the above ML models predict the job satisfaction of software developers with an accuracy of around 79%.
An exploratory evaluation of replacing ESB with microservices in service-oriented architecture
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Weerasinghe, L. D. S. B.; Perera, Indika
With the continuous progress in technology during the past few decades, cloud computing has become a fast-growing technology in the world, making computerized systems widespread. The emergence of Cloud Computing has evolved towards microservice concepts, which are highly demanded by corporates for enterprise application level. Most enterprise applications have moved away from traditional unified models of software programs like monolithic architecture and traditional SOA architecture to microservice architecture to ensure better scalability, lesser investment in hardware, and high performance. The monolithic architecture is designed in a manner that all the components and the modules are packed together and deployed on a single binary. However, in the microservice architecture, components are developed as small services so that horizontally and vertically scaling is made easier in comparison to monolith or SOA architecture. SOA and monolithic architecture are at a disadvantage compared to Microservice architecture, as they require colossal hardware specifications to scale the software. In general terms, the system performance of these architectures can be measured considering different aspects such as system capacity, throughput, and latency. This research focuses on how scalability and performance software quality attributes behave when converting the SOA system to microservice architecture. Experimental results have shown that microservice architecture can bring more scalability with a minimum cost generation. Nevertheless, specific gaps in performance are identified in the perspective of the final user experiences due to the interservice communication in the microservice architecture in a distributed environment.
Autism spectrum disorder diagnosis support model using InceptionV3
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Lakmini, Herath; Marasingha, M. A. J. C.; Meedeniya, Dulani; Weerasinghe, Vajira
Autism spectrum disorder (ASD) is one of the most common neurodevelopment disorders that severely affect patients in performing their day-to-day activities and social interactions. Early and accurate diagnosis can help decide the correct therapeutic adaptations for the patients to lead an almost normal life. The present practices of diagnosis of ASD are highly subjective and time-consuming. Today, as a popular solution, understanding abnormalities in brain functions using brain imagery such as functional magnetic resonance imaging (fMRI), is being performed using machine learning. This study presents a transfer learning-based approach using Inception v3 for ASD classification with fMRI data. The approach transforms the raw 4D fMRI dataset to 2D epi, stat map, and glass brain images. The classification results show higher accuracy values with pre-trained weights. Thus, the pre-trained ImageNet models with transfer learning provides a viable solution for diagnosing ASD from fMRI images.
Temporal preferential attachment: Predicting new links in temporal social networks
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Wickramarachchi, Panchani; Munasinghe, Lankeshwara
Social networks have shown an exponential growth in the recent past. It has estimated that nearly 4 billion people are currently using social networks. The growth of social networks can be explained using different models. Preferential Attachment (PA) is a widely used model, which is often used to link prediction in social networks. PA tells that the social network users prefer to get linked with popular users in the network. However, the popularity of a node depends not only on the node’s degree but also on the node's activeness which is reflected by the amount of active links the node has at present. Activeness of a link can be quantified using the timestamp of the link. The present work introduces a novel method called Temporal Preferential Attachment (TPA) which is defined on the activeness and strength of a node. Strength of a node is the sum of weights of links attached to the node. Here, the weights of the links are assigned according to their activeness. Thus, TPA captures the temporal behaviors of nodes, which is a vital factor for new link formation. The novel method uses min - max scaling to scale the time differences between current time and the timestamps of the links. Here, the min value is the earliest timestamp of the links in the given network and max value is the latest timestamp of the links. The scaled time difference of a link is considered as the temporal weight of the link, which reflects its activeness. TPA was evaluated in terms of its link prediction performance using well-known social network data sets. The results show that TPA performs well in link prediction compared to PA, and show a significant improvement in prediction accuracy.
Solution approach to incompatibility of products in a multi-product and heterogeneous vehicle routing problem: An application in the 3PL industry
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Weerakkody, H. D. W.; Niwunhella, D. H. H.; Wijayanayake, A. N.
Vehicle Routing Problem (VRP) is an extensively discussed area under supply chain literature, though it has variety of applications. Multi-product related VRP considers about optimizing the routes of vehicles distributing multiple commodities. Domestic distribution of goods of multiple clients from a third-party logistics distribution center (DC) is one example of such an application. Compatibility of products is a major factor taken into consideration when consolidating and distributing multiple products in the same vehicle. From the literature, it was identified that, though compatibility is a major consideration, it has not been considered in the literature when developing vehicle routing models. Therefore, this study has been carried out with the objective of minimizing the cost of distribution in the multi-product VRP while considering the compatibility of the products distributed, using heterogeneous vehicle types. The extended mathematical model proposed has been validated using data obtained from a leading 3PL firm in Sri Lanka which has been simulated using the Supply Chain Guru software. The numerical results showcase that cost has been reduced when consolidating shipments in a 3PL DC. The study will contribute to literature with the finding that the compatibility factor of products can be considered when developing vehicle routing models for the multi-product related VRP.
Exploiting optimum acoustic features in COVID-19 individual’s breathing sounds
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Milani, M. G. Manisha; Ramashini, Murugaiya; Murugiah, Krishani; Chamal, Lanka Geeganage Shamaan
The world is facing an extreme crisis due to the COVID-19 pandemic. The COVID-19 virus interrupts the world’s economy and social factors; thus, many countries fall into poverty. Also, they lack expertise in this field and could not make an effort to perform the necessary polymerase chain reaction (PCR) or other expensive laboratory tests. Therefore, it is important to find an alternative solution to the early prediction of COVID-19 infected persons with a low-cost method. The objective of this study is to detect COVID-19 infected individuals through their breathing sounds. To perform this task, twenty-two (22) acoustic features are extracted. The optimum features in each COVID-19 infected breathing sound is identified among these features through a feature engineering method. This proposed feature engineering method is a hybrid model that includes; statistical feature evaluation, PCA, and k-mean clustering techniques. The final results of this proposed Optimum Acoustic Feature Engineering (OAFE) model show that breathing sound signals' Kurtosis feature is more effective in distinguishing COVID-19 infected individuals from healthy individuals.
Estimation of the incubation period of COVID-19 using boosted random forest algorithm
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Rathnayake, P. P. P. M. T. D.; Senanayake, Janaka; Wickramaarachchi, Dilani
Coronavirus disease was first discovered in December 2019. As of July 2021, within nineteen months since this infectious disease started, more than one hundred and eighty million cases have been reported. The incubation period of the virus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), can be defined as the period between exposure to the virus and symptom onset. Most of the affected cases are asymptomatic during this period, but they can transmit the virus to others. The incubation period is an important factor in deciding quarantine or isolation periods. According to current studies, the incubation period of SARS-CoV-2 ranges from2 to 14 days. Since there is a range, it is difficult to identify a specific incubation period for suspected cases. Therefore, all suspected cases should undergo an isolation period of 14 days, and it may lead to unnecessarily allocation of resources. The main objective of this research is to develop a classification model to classify the incubation period using machine learning techniques after identifying the factors affecting the incubation period. Patient records within the age group 5-80 years were used in this study. The dataset consists of 500 patient records from various countries such as China, Japan, South Korea and the USA. This study identified that the patients' age, immunocompetent state, gender, direct/indirect contact with the affected patients and the residing location affect the incubation period. Several supervised learning classification algorithms were compared in this study to find the best performing algorithm to classify the incubation classes. The weighted average of each incubation class was used to evaluate the overall model performance. The random forest algorithm outperformed other algorithms achieving 0.78 precision, 0.84 recall, and 0.80 F1-score in classifying the incubation classes. To fine-tune the model AdaBoost algorithm was used.
Architectural framework for an interactive learning toolkit
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Jayasiriwardene, Shakyani; Meedeniya, Dulani
At present, a significant demand has emerged for online educational tools that can be used as replacement for classroom education. Due to the ease of access, the preference of many users is focused on m-learning applications. This paper presents an architectural framework for an interactive mobile learning toolkit. This study explores different software design patterns and presents the implementation details of the prototype. As a case study, the application is applied for the primary education sector in Sri Lanka, as there is a lack of adaptive learning mobile toolkits that allow teachers and students to interact effectively. The study is concluded to be user-friendly, understandable, useful, and efficient through a System Usability Study.
Smart technologies in tourism: a study using systematic review and grounded theory
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2021) Nafrees, Abdul Cader Mohamed; Shibly, F. H. A.
Tourism that uses smart technology and practices to boost resource management and sustainability while growing their businesses' overall competitiveness is known as smart tourism. Information and communication technologies (ICTs) have had a profound impact on the tourism industry, and they continue to be the key drivers of tourism innovation. ICTs have fundamentally changed the way tourism products are developed, presented, and offered, according to the literature. Any empirical studies or experiments must be focused on accepted or formed hypotheses. In this regard, grounded theory measures were used for interpretation, while a systematic review was performed to assess the research scope from current studies and works. The main goal of the study is to investigate and propose long-lasting and stable smart technologies for implementing smart tourism. Grounded theory is a concept that uses methodical rules to gather and dissect data in order to construct an unbiased theory. Fewer studies on smart technology in tourism have been conducted, with a majority of them concentrating on IoT, virtual and augmented reality, big data, cloud computing, and mobile applications. In either case, there is space for further investigation into this important field of study. As a result, this paper is a vital first step toward a clearer understanding of how smart technology can be applied to the tourism industry. The number of available research work on smart technologies in tourism were fewer from the selected journals and conference proceedings, which led to the accessibility of lesser data for analysis.