International Journal of System of Systems Engineering (70 papers in press)

Regular Issues

• Spark Framework-Based Crop Yield Prediction Using KR-PEclat And Mish-ANFIS-GRU Technique  
  by Anupama C.G, Lakshmi C 
  Abstract: Recent advancements have made tremendous development in various fields including the agricultural sector. Existing research methodologies predicted the crop yield only based on the soil and weather conditions which in turn degraded the efficiency of the crop yield prediction. Hence, an efficient Mish-ANFIS-GRU and DI-LDA-based crop yield prediction methodology is proposed. Initially, the data obtained from the historical dataset is pre-processed and data partitioning is performed using the KR-PEclat algorithm. The partitioned data is then fed into the spark framework. Then, data balancing is done using SMOTE technique to obtain the highest matching data. Next, features are extracted from the balanced data followed by the DI-LDA-based feature reduction process. The reduced features are then fed into the Mish-ANFIS-GRU classifier. Now, when the farmer enters the condition for predicting the yield of the particular crop, feature mapping is performed to provide a better prediction of the crop yield.
  Keywords: Imputation; Gated Recurrent Unit(GRU); Linear Discriminant analysis(LDA); Adaptive Neuro Fuzzy Interference System(ANFIS); Synthetic Minority Oversampling Technique (SMOTE).
  DOI: 10.1504/IJSSE.2024.10057480
   
  
• Bidirectional LSTM and Self-Attention Mechanisms based Multi-Label Sentiment Analysis  
  by Aruna A.R 
  Abstract: This study proposes the implementation of a novel optimization-depend on deep learning algorithm for Multi-Label Sentiment Analysis (MLSA). The goal of the algorithm is to improve the accuracy of sentiment analysis, particularly in the context of e-commerce-related applications. This technique effectively categorize the text data into multiple sentiment classes, such as positive, negative, neutral, or other emotions, and to determine the overall sentiment expressed in a given text document. The challenge of MLSA on e-commerce data lies in the informal and often cryptic nature of the text, which can make sentiment analysis difficult. To address this, a novel optimization-empowered Bidirectional Long-Short Term Memory (Bi-LSTM) system with Self-Attention Mechanisms is proposed in this research work. It uses the Bi-LSTM network to capture the sequential relationships between words in the manuscript and the self-attention mechanism to dynamically weigh the importance of different words in determining the overall sentiment expressed in the text
  Keywords: Self-attention; Bi-Directional Long Short-Term Memory; Multi-Label Sentimental Analysis; Deep Learning; Sentimental Analysis; Natural Language Processing.
  DOI: 10.1504/IJSSE.2025.10059046
   
  
• Security and Data Privacy Systems Concerns in IoT Using Consensus Algorithm  
  by Maria Michael Visuwasam L, Sheetal Vishal Deshmukh, N.R.Rejin Paul, M.Arun Manicka Raja, Kanimozhi S, Anuradha Thakare 
  Abstract: The most recent technology to emerge as a result of the quick development of smart devices and related technologies from both an industry and research standpoint is the Internet of Things (IoT). IoT techniques are used in the development of apps for real-time monitoring. Smart things are vulnerable to attacks due to their insufficient processing and storage capacities and the ineffectiveness of current security and encryption techniques. The first step was an investigation and analysis of the study's system to identify any potential breaches in confidentiality or security. Second, several security options are provided by blockchain technology. The intricacies of the analysis, such as the incorporation of enabling technologies and the Internet of Things, are broken down in great detail. After that, the results of a sample experiment utilising a blockchain-driven Ethereum and an intelligent IoT system are shown to the audience.
  Keywords: IoT; smart IoT; Block Networks; Security; Privacy Systems; Cryptography Technique.
  DOI: 10.1504/IJSSE.2024.10058093
   
  
• Compact microstrip patch antenna design using Deep belief neural network for wireless application  
  by Sandhya Rani Sriram, Kumar Naik Ketavath 
  Abstract: This paper presents a deep belief neural network (DBN) to design an inset-fed E-shaped microstrip patch antenna. To design the shape of such an antenna, a DBN is proposed. This article is presented to design of a compact patch microstrip antenna with an operating frequency of 0.75 to 2.24 GHz and 3 to 3.46 GHz. The upper and lower notches maintain the same dimensions throughout the design process. Notch length and width are set for the investigation purpose. The proposed work utilizes the optimal DBNN model for the designing of the antenna in terms of area and therein significantly maximizes the bandwidth usage and is also used for simulation purposes. The outcomes are analyzed and compared with state-of-art works and show our proposed approach shows the reduced area with the maximized bandwidth usage.
  Keywords: Microstrip patch antenna; E-shaped microstrip patch antenna; Deep belief neural network; Bandwidth and Area.
  DOI: 10.1504/IJSSE.2025.10058886
   
  
• Load Balancing in Cloud Computing Systems Using Density Based Clustering Approach  
  by Pearly Charles, Vimala S 
  Abstract: Cloud computing, which uses clustering to load balance, is the current paradigm for providing ultimate services to society via the internet. This technology delivers all PAYG services. Privacy, security, reliability, and other problems offset infrastructure, platform, and software gains. Load balancing improves dispersed environments. Recent research prevents VM under- and over-loading. This research uses a density-based clustering-derived LB method. The turn around time (TAT) is much lower than K-Means. K-Means and DBSCAN cloud load balance. Clustering balances server loads. Similar queries let server clusters share the load. System performance, reaction time, and downtime improve the traditional load balancing works well. Round-robin sequence requests among servers. Cluster servers share the load. The least-connections approach sends requests to the server with the fewest active connections, ensuring each server has a similar capacity. Clustering load balances in real time. Clustering algorithms transfer groupings to other servers to balance demand. K-Means takes 269.875ms longer than anticipated.
  Keywords: Load Balancing; Turn Around Time (TAT); Virtual Machine (VM); Cloud Lets; Cloud Sim; Clustering Approach.
  DOI: 10.1504/IJSSE.2024.10058171
   
  
• Predicting Consumer's Intention of Biological Products Using E-Commerce Data  
  by Kaliraj S, Raghavendra S, Femilda Josephin J.S, Sivakumar V, Karthick K 
  Abstract: Digitalization has evolved as a boon to the e-commerce market. Biological products and organic products also target e-commerce platforms to increase their business. E-commerce has the upper hand over traditional marketing practices due to its adequate accessibility and usability. The research revolves around consumers' opinions in the form of ratings and the idea that the products sold on E-Commerce platforms correlate with the product's rating and features like brand, price, etc. This lets the practitioners predict the consumers' intention by predicting the possible rating. There are many approaches available to predict consumer intention based on e-commerce data. In this paper, we have evaluated the performance of all the machine learning classification algorithms. All of these are used in our proposed structure to predict consumer intention on a product. Here we trained machine learning algorithms using an extracted dataset for forecasting biological product ratings based on other product features. Performance of different machine learning algorithms on e-commerce data discussed using metrics.
  Keywords: Supervised machine learning; consumer behavior; classification algorithms; e-commerce biological product; deep learning.
  DOI: 10.1504/IJSSE.2025.10059048
   
  
• Cardiovascular disease prediction using Hybridization multi perception classifier in secure IoT platform
  by Safa M, Pandian A, Chakrapani K, Karpaga Selvi, Kempanna M, Arun D, Umamaheswari K.M 
  Abstract: The primary purpose of this study is to propose a hybrid fuzzy-based decision tree method for early heart attack prediction using a continuous and remote patient monitoring system. The first planned goal is to create an IoT system that detects an individual's level of stress and uses the information gathered through sensor-linked IoT to help individuals cope with stress. The sensory system detects and monitors other proposed data sets for heart disease patients involved in temperature, blood pressure, pulse oximetry, and stress. The IoT Edge intelligence device senses signals from sensors. It manages and monitors output using the MQTT protocol. The IoT Hub, in collaboration with large-scale devices, generates analytical cardiovascular predictions using streaming analysis and real-time data processing in this suggested system. Predictive models for stress analysis are designed using machine learning methods.
  Keywords: Cardiac disease prediction; big data analysis; HMPC; CHD; Stress; Sensors.
  DOI: 10.1504/IJSSE.2025.10058918
   
  
• A novel approach for Enhancing Mammographic Images  
  by Richa Sharma, Amit Kamra 
  Abstract: A severe health problem impacting many ladies worldwide is mammary calcinoma In order to reduce the risk factor connected with the disease,pre-clinical identification of breast cancer is essential,and mammography is one of the finest screening methods Mammogram quality and radiologists' ability to correctly interpret them are crucial for early cancer diagnosis In this study work,we propose a brand-new method for improving poor-quality mammography pictures Contrast Limited Adaptive Histogram Equalisation (CLAHE)and Morphological Operations(MOs)are two image processing methods we recommend using The suggested technique tries to increase the pictures' clarity and sharpness,which may help in the early identification and diagnosis of breast cancer The median filter,Low Pass Gaussian filtering,Morphological Operations(MOs),and Wavelet Decomposition are the next steps in the suggested technique after removing noise.The combination of these two approaches will increase the efficiency of mammographic image enhancement,according to our studies,which show a considerable improvement in picture quality compared to CLAHE and MOs.
  Keywords: CLAHE; Guided Image Filtering; mammograms,Morphological operations; Medical Image Processing.
  DOI: 10.1504/IJSSE.2025.10058950
   
  
• A Fuzzy Inference System Confidence Dynamic Concept Simulated Annealing Strategy for Wireless Sensor Networks  
  by Selvamani K, Kanimozhi S, N.R.Rejin Paul, M.Arun Manicka Raja, Venkatasubramanian S, Anuradha Thakare 
  Abstract: Scientific study has focused on extending the lifespan of wireless sensor networks, a cost-effective technique to collect data from a specific area. Previous studies offered a low-energy heterogeneous wireless sensor network routing technique. Few writers proposed the algorithm for finding and calculating critical node linkages. Installing more mobile nodes improved WSN's topological connection earlier. Path design was also proposed to maximise longevity and decrease connected key node effects. Some geo-cast methods used hop-to-hop neighbour data. Dynamic resource routing for wireless sensor networks is advocated using an FIS and area segmentation. Thus, correct device data flow saves energy and prolongs channel life. This work introduces geographic routing. Fuzzy logic determines neural source coordinates, and weighted centroid identification is suggested. A wireless fuzz version measures flow to determine anchor-edge device distance. It decreases localised standard errors and node placement errors. Second, boost messages to the next bounce member nodes with the latest version. Smart next-hop selection reduces node energy usage and extends network lifetime. The suggested thing outperforms existing ways in power utilisation, completion time, and location errors, according to simulations.
  Keywords: Wireless Sensor Network; Internet of Things (IoT); Segmentation; Classification; Communication; Fuzzy Logic; Fuzzy Inference System (FIS),.
  DOI: 10.1504/IJSSE.2025.10059049
   
  
• Ensemble Regression Model-Based Missing Not at Random Type Missing Data Imputation on the Internet of Medical Things  
  by Iris Punitha P, J.G.R. Sathiaseelan 
  Abstract: The Internet of Medical Things (IoMT) combines IoT and health sensing technologies, which allow for the early detection of various health issues. However, the data generated from IoMT devices may contain missing values or corrupted data, particularly when the missing data is of the missing not-at-random (MNAR) type. Existing solutions for imputing missing data in IoMT have limitations such as low accuracy and high computational cost. To overcome these limitations, this paper proposes an ensemble regression model (ERM) based on MNAR-type missing data progressive imputation (MDPI). The ERM-MDPI model combines three regression models, namely Multilayer Perceptron (MLP), Support Vector Regression (SVR), and Linear Regression (LR), to improve the accuracy of imputed missing data in the cStick dataset. The experimental results demonstrate that the ERM-MDPI model-based cStick imputed dataset generated higher accuracy (93.6301%), precision (91.0385%), recall (87.0898%) and F-measure (89.0204%) than cStick missing dataset. Therefore, the proposed solution offers an efficient and accurate approach to impute MNAR-type missing data in IoMT, providing valuable insights for medical decision-making.
  Keywords: The Internet of Medical Things (IoMT); Missing not-at-random (MNAR) data; Ensemble regression model (ERM); Progressive imputation; Medical decision-making.
  DOI: 10.1504/IJSSE.2024.10058313
   
  
• Application of Deep Convolutional Neural Networks Systems in Autonomous Vehicles  
  by Souvik Ganguli, Charu Virmani, Vrince Vimal, Gunjan Chhabra, Garima Sinha, Bobur Sobirov 
  Abstract: The currently available sensor on those self-driving automobiles does a poor job of detecting the state of the road ahead of them. However, daytime and nighttime weather-related road conditions require safe driving. Deep learning study for daytime roadway identification uses data from a vehicle sensor. An overview of the use of deep convolutional neural networks in autonomous cars is given in this article. The paper starts by going through the difficulties of creating autonomous vehicles and how CNNs can be utilised to overcome these difficulties. The author thoroughly explains the basis of CNN and how it may be used for tasks like object detection, lane finding, and recognition of traffic signals. The research also examines how CNN focus techniques and transfer learning can be applied to autonomous vehicles. The authors conclude by highlighting the limits of current CNNs in this field and suggesting future research. This review article gives academic scholars and industry experts a current overview of CNNs in cars.
  Keywords: Application; Deep Convolutional Neural Networks Systems; Autonomous Vehicles; Convolutional Layers; Convolutional Neural Networks (CNNs).
  DOI: 10.1504/IJSSE.2025.10059050
   
  
• A Collaborative Approach for Testing MapReduce Access Control Issues using Agent-based Architecture  
  by Sara Hsaini, Mohammed Amine Tajioue, Salma Azzouzi, Moulay El Hassan Charaf 
  Abstract: Security mechanisms such as access control are of the utmost importance. There have been a number of applications where it has been used to control the actions performed on resources by system users. In this study, a novel approach to testing access control list (ACL) policies based on the analysis of access control responses is presented. The main objective is to verify that the policies meet the XACML language specifications. As part of addressing the issue in a practical manner, we propose an Agent-based architecture for testing access control of a MapReduce application in a multi-node Hadoop cluster. The methodology is then tested and validated through an "inverted index" case study.
  Keywords: Access Control; Distributed Testing; MapReduce; Security; XACML; Multi-Agents System.
  DOI: 10.1504/IJSSE.2025.10059051
   
  
• Detection of Node isolation attack using DNC mechanism in D2D Networks  
  by Balaji V, Selvaraj P 
  Abstract: Device to Device (D2D) network is a powerful technology that enables D2D communication without additional infrastructure support. There is no authentication procedure in the D2D routing protocol. As a result, it is a very desirable target for attackers and is exposed to several vulnerabilities, including node isolation attacks during sessions. To avoid this problem, a novel Denial of Node Contradiction (DNC) with OLSR is proposed. According to the simulation results, the proposed DNC-OLSR algorithm outperforms the alternatives in terms of the rate at which packets are forwarded, indicating that the proposed routing path is more stable than those of the Destination Sequenced Distance Vector (DSDV), Ad-hoc On-Demand Distance Vector (AODV) and Dynamic Source Routing (DSR) protocols. However, the DNC-OLSR algorithm is relatively low compared to the other three algorithmic measures of packet loss and the resulting delay. Moreover, the proposed algorithm can provide more reliability and stability for D2D communication.
  Keywords: IoT; D2D; Security Issues; OLSR; AODV; DSR; DSDV.
  DOI: 10.1504/IJSSE.2025.10059052
   
  
• Deep Convolutional Neural Networks based Cervical Cancer Detection and Classification  
  by Suguna C, Balamurugan S.P 
  Abstract: Cervical cancer (CC) is a major reason of death in cancer in women. The problem of this cancer is limited once it can be analyzed and preserved at the initial phase. With the progress of artificial intelligence (AI) technology, computer aided diagnosis (CAD) is developed most leading investigation topics of medicinal imaging during recent decades. This study develops a Computer Aided Diagnosis Model for Cervical Cancer Classification using Deep Learning (CADC3-DL) model. The presented CADC3-DL model aims to recognize the occurrence of CC on biomedical images. Initially, the CADC3-DL technique creates a mask from the input dataset. Next, the pre-processing step takes place in two levels namely Gaussian filtering (GF) based noise removal and CLAHE based contrast enhancement. Then, the CADC3-DL technique employs customized U-Net segmentation technique where the filter size in the traditional U-Net is replaced by batch normalization (BN) to accomplish enhanced classification accuracy.
  Keywords: Biomedical imaging; Cervical cancer; Deep learning; Computer aided diagnosis; Batch normalization.
  DOI: 10.1504/IJSSE.2025.10059072
   
  
• Study and analysis of Data Anonymization Techniques for Social Networks  
  by Sivasankari K, Umamaheswari K.M 
  Abstract: Many people all across the globe have been utilizing social media to exchange information. Numerous firms apply social data mining to extract numerous excit-ing insights from social data which is represented as a sophisticated network structure. However, publishing social data has a direct and indirect influence on the privacy of many of its users. Recently, numerous anonymization methods have been invented and deployed to preserve sensitive information about users and their interactions on social media. This paper presents a complete assessment of several data anonymization algorithms for social network data and evaluates their pros and downsides. It also tackles the primary research problems surround-ing the effectiveness of anonymization technologies.
  Keywords: Data Anonymization; social networks; graph modification; machine learning.
  DOI: 10.1504/IJSSE.2024.10058376
   
  
• Monitoring and Actuating Devices for Analysis of The New Security Protocols of Internet of Things  
  by A. Anandhavalli, A. Bhuvaneswari 
  Abstract: The Internet is ubiquitous and significantly influences contemporary life. A network of countless individually distinguishable, connected, and physically reachable things makes up the Internet of Things (IoT). The Internet of Things can be sensed, communicated with, and controlled by any physical object that can function as a computational device. Internet of Things, cloud computing, and pervasive computing affect corporate and software industries. Software companies must modernise their business models to compete. Data increases as devices connect. This data transformation can happen between devices, devices and humans, and between devices and businesses. Data fabrication, server and network manipulation, and serious ramifications for application platforms and networks could all result from this. The dangers and shortcomings of these technologies have increased due to their increased use in daily life. This study covers IoT basics, challenges, and security. Standardization is unattainable since so many devices are connected heterogeneously across platforms and networking protocols. Academics can now study the IoT's architectures, designs, applications, new security risks, and challenges.
  Keywords: Internet of Things (IoT); Lightweight Local Automation Protocol (LLAP); MQTT (Message Queue Telemetry Transport); Quick UDP Internet Connections (QUIC); User datagram protocol (UDP); Zigbee; Bluetooth.
  DOI: 10.1504/IJSSE.2024.10058411
   
  
• Adaptive cat optimization with Attention based bi-LSTM for automatic software bug detection  
  by Muthu Kumaran A.M.J, Umamaheswari K.M 
  Abstract: The detection of software bugs is a critical issue in the software maintenance and development process because it is connected to all software successes. Therefore, earlier software bug detection is essential for improving software's efficiency, dependability, quality, and cost. Furthermore, accurate software bug prediction is a critical and challenging task. This article thus develops the effective software bug prediction model. The pre-processing, feature selection, and bug detection phases of the proposed model are the most important ones. The input bug dataset is initially pre-processed. Remove instances of duplicate data from the dataset during pre-processing. The feature selection is carried out by the Adaptive Cat Swam Optimization Algorithm (ACS) following the pre-processing step. At last, the proposed approach uses a Bi-long transient memory (Bi-LSTM) for bug expectation. Bug prediction is done with the promise and the NASA dataset. Based on accuracy, the proposed model performs better than the simulation results.
  Keywords: bug prediction; Bi-long short-term memory (Bi-LSTM); Cat Swam Optimization; feature selection.
  DOI: 10.1504/IJSSE.2025.10059053
   
  
• Firefly Cyclic Golden Jackal Optimization Algorithm with Wavelet Artificial Neural Network for Blackmailing Attack Detection in Mobile Ad-hoc Network  
  by Parameshwar G, N.V.Koteswara Rao, Nirmala Devi L 
  Abstract: Wireless networks called mobile ad hoc networks (MANETs) have an enlarged number of peer nodes. In recent studies, the major challenges are poor false positive rate, minimum detection rate and energy efficiency with higher delay to enhance the security of MANET. To overcome the problem, in this work, the fuzzy clustering model forms the clusters in MANET. The most appropriate cluster heads are selected in the presence of the Firefly Cyclic Golden Jackal Optimization (FCGJO) algorithm thereby solving the issues of energy and mobility of nodes. Wavelet Artificial Neural Network model to detect a blackmailing attack in MANET. The NS-2 simulation tool handles the implementation works and the statistical parameters such as attack detection rate, delay, energy consumption, throughput, memory consumption and etc to compute the performance of proposed approaches. While comparing to the state-of-art studies, the statistical parametric results reveal the proposed intrusion detection performance.
  Keywords: MANET; Firefly Cyclic Golden Jackal Optimization; Fuzzy clustering model; Blackmailing attack and Wavelet Artificial Neural Network.
  DOI: 10.1504/IJSSE.2025.10058971
   
  
• A Novel Architecture for Diabetic and Glaucoma Detection using Multi-Layer Convolutional Neural Network System  
  by Neha Sewal, Charu Virmani 
  Abstract: The proposed effort aims to develop a more predictive model for identifying Diabetic Retinopathy (DR) and Glaucoma, two major retinal illnesses that cause blindness in working-age individuals globally. Colour retinal photographs are challenging and time-consuming to diagnose DR and Glaucoma. A multilayer CNN model identifies DR and Glaucoma. The CNN-based network has special upgrades and features to boost diagnosis accuracy. The model can better capture retinal picture subtleties by adding data, improving diagnostic performance. 3658 retinal pictures in five categories are used for the DR problem from the Kaggle dataset. With 1103 retinal pictures and two class labels, RIGA is used for Glaucoma. Using these datasets, the proposed technique easily identifies healthy and diseased retinal pictures, reducing physician evaluations. Using two publicly available datasets, the suggested model had a prediction accuracy of 98%. These measures demonstrate the model's capacity to classify retinal pictures and detect DR and Glaucoma. Finally, data-augmented multilayer CNN models improve DR and glaucoma diagnosis. Accuracy comes from detail and large datasets. Data augmentation and multilayer CNN models improved it. The model may detect and treat severe retinal problems earlier.
  Keywords: Diabetic Retinopathy; Kaggle; Deep Learning; Convolutional Neural Network; Glaucoma; RIGA Dataset; Multi-Layer CNN; Augmentation Techniques,.
  DOI: 10.1504/IJSSE.2025.10059054
   
  
• An Optimal Insider-Threats Detection Model Based on Improved Deep Belief Network with Feature Reduction Scheme for E-Healthcare System  
  by M. Madhavi, T. Sasirooba, G.Kranthi Kumar 
  Abstract: An EHR (Electronic Health Records) dataset contains routine actions performed when accessing a patient's record,such asassessment form history,pharmacy orders, etc These actions differ depending on the medical practitioner and the access period Leakage of valuable information is a critical challenge As the Internet-of-Things(IoT)evolves,new security challenges arise in existing security architectures An organization's insider threat management is at risk because attack surfaces have expanded dramatically To solve the above challenges,In this work,an Optimized Deep belief network(DBN)is proposed to detect insider threats in EHR Significant features are generated using correlation coefficients, random forest mean reduced accuracy, and gain ratio to improve the performance of the internal threat detection model An appropriate mechanism (and function) is then used to combine the features to obtain an optimal set of features Adaptive rat optimization algorithm (AROA) optimizes DBN weight parameters to enhance performance F-measure, recall, accuracy,precision, and G-mean are calculated to measure Performance.
  Keywords: EHR; Insider threat detection; DBN; AROA and single optimized feature sets.
  DOI: 10.1504/IJSSE.2025.10059055
   
  
• Risk integrated effort estimation of software projects: a comparative analysis of machine learning techniques  
  by Prerna Singal, Prabha Sharma, Charan Kumari 
  Abstract: Accurate software project effort estimation and risk management are the pillars of delivering an on-time, within budget and quality project. In our earlier research, a formula for computing risk integrated effort estimate by adding weighted cost of risk management for each project cost factor to the cost of initial effort estimate of the project has been proposed. In this research, neural network techniques: MLP, GRNN, CCNN and RBFNN; support vector regression, and adaptive neuro fuzzy inference system to obtain the integrated effort estimate as close as possible to the actual effort spent on the project have been applied. The techniques have been tested on two datasets: Agile and Waterfall datasets. GRNN gave the best results in terms of lowest values of accuracy measures: MAE, MMRE, MBRE, and MIBRE. This research also compares performance of GRNN with evolutionary algorithms ABC, PSO and GLBPSO, and the results for GRNN are comparable.
  Keywords: Agile projects; risk management; neural networks; CoCoMo II; support vector regression; risk exposure.
  DOI: 10.1504/IJSSE.2025.10059056
   
  
• Vision Based Vehicle Tracking Network and Counting Using Deep Learning Model Systems  
  by Hemalakshmi K, A. Muthukumaravel 
  Abstract: Vehicle counting is a key component of the vehicle behaviourist approach and traffic incident detection for certain video surveillance systems. The accurate counting of vehicles in a variety of traffic conditions using deep learning algorithms and multi-object tracking systems is a popular area of research in the field of intelligent transportation. This research suggests a three-step vehicle identification, tracking, and counting process as a framework for video-based vehicle counting to estimate traffic flow. First, the Yolov3, Faster R-CNN, and SSD deep learning architectures are used to detect the vehicle, and the performances of each are compared. A modified DeepSORT algorithm tracks observed cars, and a picture shows their trajectory. In low-light and traffic conditions, a new vehicle counting system uses tracking data to count vehicles by type. Traffic figures are compared. The recommended method accurately recognises automobiles, tracks multiple objects, and detects with high precision and accuracy, according to experiments. This method meets real-time processing and vehicle counting needs. This study's method can count automobiles on difficult highways.
  Keywords: Vehicle detection; Vehicle Tracking Network; Vehicle counting; Deep learning; Traffic video; Yolov3; DeepSORT algorithm; Detection-Tracking-Counting.
  DOI: 10.1504/IJSSE.2025.10058951
   
  
• Feed Forwarded Neural Network with Learning-Based Tuna Swarm Optimization (FFNN-LBTSO) For Semen Quality Prediction Systems  
  by C. Shanthini, S.Silvia Priscila 
  Abstract: Nowadays, some new diseases have come into existence due to lifestyle diseases. The major causes of the change in semen quality are environmental and lifestyle factors. One of the key tasks to assess the fertility potential of a male partner is semen analysis. Data-mining decision support systems can help identify this influence. Some seminal quality predictions were made. This research exploited unbalanced datasets with biased majority-class performance findings. Gradient descent local training is prone to local minima. Meta-heuristic algorithm optimization permits local and global solution finding. The paper develops a neural network model to predict semen quality. This paper improves Tuna Swarm Optimization using learning-based Feed-Forward Neural Networks (FFNN) (TSO). To balance normal and atypical cases, SMOTE data balancing was used. Overflow produces minority class instances until the balance is reached. FFNN-LBTSO was tested for predictive power. Steps include data source and pre-processing, SMOTE, FFNN classification, and LBTSO for classifier weights and bias optimisation. UCI sperm prediction. Sensitivity, specificity, G-mean, and accuracy measure experimentation. Fertility-optimal semen was detected. Comparing SVM and ANN classifier results.
  Keywords: Feed Forward Neural Network (FFNN); Tuna Swarm Optimization (TSO); Synthetic Minority Oversampling Technique (SMOTE); Learning-Based Tuna Swarm Optimization (LBTSO); Artificial Neural Network (ANN),.
  DOI: 10.1504/IJSSE.2025.10058952
   
  
• Expansion of Situations Theory for Exploring Shared Awareness in Human-Intelligent Autonomous Systems  
  by Scott Humr, Anthony Canan, Mustafa Demir 
  Abstract: Intelligent Autonomous Systems (IAS) are part of a system of systems (SoS) that interact with other agents to accomplish tasks in complex environments. However, IAS-integrated SoS add additional layers of complexity based on their limited cognitive processes, specifically shared situation awareness that allows a team to respond to novel tasks. IAS's lack of shared situation awareness adversely influences team effectiveness in complex task environments, such as military command-and-control. A complementary approach of shared situation awareness, called situations theory, is beneficial for understanding the relationship between SoS's shared situation awareness and effectiveness. The current study elucidates a conceptual discussion on situations theory to investigate the development of an SoS’s shared situational awareness when humans team with IAS agents. To ground the discussion, the reviewed studies expanded situations theory within the context of SoS that result in three major conjectures that can be beneficial to the design and development of future SoSs.
  Keywords: Artificial Intelligence; Human-Machine Interaction; Intelligent Autonomous Systems; Shared Situational Awareness; Situations Theory.
  DOI: 10.1504/IJSSE.2025.10058953
   
  
• An Intelligent Neural Question Answer Generation from Text Using Seq2se2 with Attention Mechanism System  
  by Sonam Soni, Praveen Kumar, Amal Saha 
  Abstract: Utilising data to its fullest extent is becoming increasingly important due to the rapid advancement of data over the past few years. Neural Question Answering is best for this much data. Question-Answer pairings have been laborious. Self-evaluation, education, and courses require questions and answers. Other AI businesses automate customer support inquiries. Designing such a system involves curating a database of consumer enquiries and live customer support representatives responses. For a new query, the system finds the best matched response in the curated dataset. Despite lacking common sense and reasoning skills, the Question Answering System is nonetheless widely used. We propose using reading comprehension strategies to automatically generate questions from sentences. The study used several methods to find the best Question Answer Pair algorithm. To boost accuracy, the model uses BERT, ELMo, and GloVe embedding methods. The model accurately reflects semantic and syntactic characteristics of the input text using these embedding strategies. Attention mechanisms help the model focus on key inputs and generate contextual predictions. Attention and embedding improve model accuracy.
  Keywords: Neural Question Answer; Word Embedding; Encoder-Decoder; AI firms; Sequence-to-Sequence; Question Answering System,.
  DOI: 10.1504/IJSSE.2025.10058954
   
  
• SiamEEGNet: Few-Shot Learning for Electroencephalogram-based Biometric Recognition System  
  by Kriti Srivastava, Siddharth Sanghvi, Parag Vaid, Palash Rathod 
  Abstract: Authentication is verifying a user's identity when they enter a system. Due to their distinctiveness, biometric-based authentication solutions have started displacing conventional systems. This study suggests employing Electroencephalogram (EEG) or brain waves as a biometric modality since the level of uniqueness attained is higher. These noise-free ECG beats generate grayscale images using the proposed SiamEEGNet. A customised activation mechanism is also designed in this study to hasten the integration of the SiamEEGNet. The one that is suggested can extract characteristics using provided data. EEG signals are difficult to manually analyse and extract features from since they are highly dimensional and have a low signal-to-noise ratio. Because deep learning architectures have transformed end-to-end learning, this study suggests employing them. Convolutional Siamese Neural Networks are used by the suggested method, SiamEEGNet, to perform few-shot learning on a well-known and openly accessible dataset called EEG Motor Movement/Imagery Dataset (EEG- MMIDB), which consists of 106 subjects. The model is then quantitatively assessed using several criteria for person identification and authentication. SiamEEGNet competes favourably with current cutting-edge methods.
  Keywords: Biometric Recognition; Siamese Neural Networks; Convolutional Neural Networks; PhysioNet; Electroencephalogram; Spectrogram.
  DOI: 10.1504/IJSSE.2025.10058955
   
  
• Prediction of Lockdown Via Opinion Mining from Tweets Using Machine Learning System  
  by Jayalakshmi V, M. Lakshmi 
  Abstract: Social networks are connected to the internet by architecture, facilitating instantaneous digital information sharing. Twitter users can share their thoughts and opinions. During the COVID-19 pandemic, polling and data helped choose the best health intervention. The COVID-19 pandemic showed that online forums and other electronic media spread disinformation more than the disease itself, threatening the world health system. Since December 2019, the new coronavirus has expanded significantly, infecting more Indians since March 2020.The Indian authorities locked down the country to limit citizen mobility and stop the infection. Social media outlets shaped user attitudes about the severe lockdown enforcement. We analyse user perception of lockdown enforcement by compiling lockdown 1.0, 2.0, and 3.0 tweets from many timelines. A Python tool trains and evaluates the deep learning framework using user feedback. Lockdown 3.0 and the government's policies are tested using new data after creation. Python analyses the forecast's performance in the three lockout scenarios. Simulation findings show that the proposed strategy outperforms existing classification algorithms.
  Keywords: Prediction; Lockdown; Opinion Mining; Tweets; Machine Learning; Indian Government; Python; COVID-19.
  DOI: 10.1504/IJSSE.2025.10059014
   
  
• Application of Custom Ant Lion Optimization Convolutional Neural Networks for Liver Lesion Classification System  
  by A.Bathsheba Parimala, R.S. Shanmugasundaram 
  Abstract: In order to save a person's life, it's essential to categorize the lesions of liver in their early stages. The majority of scientists prefer classifying liver tumours using machine learning approaches. Recently, the use of computer-aided technology for this purpose has captured the interest of scientists. This paper classifies perceptual datasets using pre-trained network models and a lion-optimized CNN classifier. However, neural network learning can be improved, and deep learning-based neural networks and its applications are rarely studied. Additionally, the Custom Optimized Convolution Neural Network (CO-CNN) is suggested in this research as a very accurate method for classifying liver lesions. The de-noising steps in this suggested method include a median filter, the Random Forest (RF) method for extracting the liver, the Gray Level Run Length Matrix (GLRLM) method for extracting features, and the CO-CNN method for classification. This technique is tested on Python. Experimental results showed that the suggested approach exceeds existing approaches in accuracy, sensitivity, and specificity. It has 96% sensitivity and 97.77% accuracy.
  Keywords: Classification System; Liver lesion; RF; Custom Optimized Convolution Neural Network (CO-CNN); Sensitivity; Gray Level Run Length Matrix (GLRLM); Ant Lion Optimization (ALO); Lesion Classification.
  DOI: 10.1504/IJSSE.2025.10059811
   
  
• An Efficient Data Sharing Scheme Using MultiTransaction Mode Consortium Blockchain for Smart Healthcare  
  by Deepak Kumar Sharma, Adarsh Kumar 
  Abstract: Electronic health records (EHRs) face security and transparency challenges, necessitating new standards. Blockchain technology holds promise for improving EHR security in smart healthcare systems. However, privacy and scalability issues persist, particularly in off-chain transaction management. We propose a method using a multi-transaction mode consortium blockchain (MTMCB) on Redis, enhancing EHR retrieval speed via an Adaptive Balanced Merkle (AB-M) tree. This approach combines binary tree efficiency with Merkle tree robustness. We employ a lattice-based ring signature scheme for secure patient EHR storage and retrieval. Our method significantly improves upload and download times compared to existing techniques, offering a potential solution to EHR access and security challenges.
  Keywords: Blockchain; Smart Healthcare; Electronic Health Records; Data retrieval; Optimized Redis cache.
  DOI: 10.1504/IJSSE.2025.10059329
   
  
• A Secure Blockchain based food supply chain management framework using hybrid IDEA algorithm  
  by Mohammed Musthafa Sheriff I, John Aravindhar D 
  Abstract: An agri based food supply chain oriented blockchain technology might provide lot of benefits like increased transparency, accountability and traceability. But there exist some challenges because of improper education, policies, frameworks and some technical aspects. Hence there occurs some need in developing a system which is more reliable that ensures the traceability, trustworthy and proper delivery mechanism in managing the Agri food supply chain. Therefore, a modified blockchain-based food supply chain is proposed based on Hybrid IDEA(International Data Encryption Algorithm) algorithm. It uses MDCNN (Multimodal Deep Convolutional Neural Networks) along with IDEA algorithm for ensuring the quality of the food Here in this work, the farmers, food processors and the various distributors will enter their data into the blockchain for generating the incontrovertible record needed for transaction. The transaction is encrypted using HIDEA algorithm which makes the authorization between the parties for accessing the data.
  Keywords: Multimodal Deep Convolutional Neural Networks; blockchain-based food supply chain; International Data Encryption Algorithm; transparency; accountability and traceability.
  DOI: 10.1504/IJSSE.2025.10059358
   
  
• Covid-19: A Comprehensive Study of the Emergence, Impact, Diagnosis, Treatment, Challenges, and Future Perspectives  
  by Prabakaran G, Jayanthi K 
  Abstract: The COVID-19 pandemic caused by the SARS-CoV-2 virus has affected millions of people worldwide and has become a major global public health challenge. This review article provides a comprehensive overview of the current knowledge on COVID-19, including its epidemiology, pathophysiology, clinical manifestations, diagnosis, and treatment. The article focuses on the various diagnostic techniques used for COVID-19, including molecular, antigen, antibody, and imaging tests, and provides a comparison of their advantages and limitations. The review also discusses the various treatment options for COVID-19, including antiviral drugs, immunomodulators, and supportive care. Furthermore, the article examines the impact of COVID-19 on vulnerable populations, such as elderly individuals, immunocompromised patients, and individuals with underlying medical conditions. Finally, the review highlights the current challenges and future perspectives for COVID-19 research, including the development of effective vaccines and strategies for pandemic preparedness.
  Keywords: COVID-19; Epidemiology; Pathophysiology; Immunocompromised; Diagnosis; Treatment.
  DOI: 10.1504/IJSSE.2025.10060236
   
  
• An Acceleration of Blockchain Mining by Parallel Process and Proof-of-Luck with Fair Share Technique System  
  by K. Lino Fathima Chinna Rani Vincent, M.P. Anuradha 
  Abstract: Blockchain plays a vital role in several applications; specifically, cryptocurrency transactions are performed through Blockchain applications. This technique stores network data securely via a distributed ledger. The transactions are part of Bitcoin mining. Miners in the network compute and verify these values. Blockchain networks using proof-of-work (PoW) consensus methods may have scalability issues as transaction volumes climb. The PoW mining mechanism cracks cryptographic riddles to authenticate more transactions and attach them to the Blockchain. When more miners join the network, transaction throughput and efficiency decrease, increasing authentication time and energy consumption. The proof-of-luck fair share parallel mining approach addresses PoW minings scalability issue. While retaining network security, this method enhances transactional validation quality and speed. Parallel mining uses several processors and processes to speed up Solo mining. The unique method boosts solo mining to TPS. Validators can mine various test case scenarios utilising the proposed methodologies in the testing environment. Experimental results show that the proposed technique can greatly increase Blockchain TPS. Experimental results show computer power's potential.
  Keywords: Parallel Mining; Scalability; Blockchain; Proof-of-Luck; TPS; Fair Share; Miners; and Transactions System.
  DOI: 10.1504/IJSSE.2025.10060238
   
  
• Customized U-Net Model based Brain Tumor Segmentation in MRI Images and Ensemble based Tumor Classification Systems  
  by Devisivasankari P, Lavanya K 
  Abstract: Medical image processing requires autonomous brain tumour segmentation because early diagnosis can improve survival by treating brain cancers quickly. Brain tumours are manually classified by experts, which is time-consuming. Brain tumour (BT) diagnosis takes time and skill, hence radiologists must be skilled. As patient numbers have grown, so has data volume, making outdated methods expensive and inefficient. Many scholars have studied fast and accurate BT detection and classification algorithms. DL can locate BTs in medical photos using trained convolutional neural network (CNN) models. Brain tumour segmentation is easier with automatic segmentation, which is widespread. This work categorises and automates brain tumour segmentation using customised UNet model-based brain tumour segmentation (CUNet-BTS). Classification, preprocessing, segmentation, feature extraction, and fusion are modelled. Gaussian filtering enhances MRI pictures. Finally, an ensemble classification algorithm is suggested. For classification, this model combines the output scores of optimal DeepMaxout, DCNN, and RNN classifiers. The excellent training model Pelican Assisted Chimp Optimisation (PACO) Method can change classification model weights.
  Keywords: Magnetic Resonance Imaging; Convolutional Neural Network; Intersection-over-Union; Fully Convolutional Neural Networks; Recurrent Regression based Neural Network; Internet of Medical Things,.
  DOI: 10.1504/IJSSE.2025.10059671
   
  
• Implementation of Octavia;s Openstack for Futuristic Cloud Computing by Optimization of Resources and Traffics
  by R.Nathiya Senthil Kumar, S.K. Piramu Preethika 
  Abstract: The Internet provides computing resources as a service on demand using cloud computing. In order to remedy this, locals have access to a range of open-source cloud operating systems, enabling them to use the cloud for useful reasons. Amazon Web Services (AWS), Microsoft Azure, Oracle Cloud, and OpenStack offer open-source cloud operating systems. Only OpenStack is free and has a substantial user base out of all of these. In private institutions or businesses, installation and use are free. Cloud load balancing is seen as a lifeline. In order to avoid website failures and other operational problems, it’s critical to balance traffic when websites face unexpected traffic spikes. This article discusses the concept and methods for automatic load balancing in OpenStack to help control bandwidth and stresses among virtual machines. The goal is to minimise traffic and maximise resources. The OpenStack networking feature LBaaS balances load as a service. An open-source LBaaS architecture for dynamic load-balancing clouds is presented here.
  Keywords: Cloud Computing; LBaaS; Load Balancer; Network Traffic; OpenStack Octavia; Octavia; bandwidth and stresses; Software-Defined Data Centers (SDDC); Software Defined Networking (SDN),.
  DOI: 10.1504/IJSSE.2025.10060239
   
  
• Numerical Solution of Boundary Value Problems Using Quantum Computing System  
  by Ajanta Das, Debabrata Datta, Suman Rajest George, Varun Kumar Nomula, R. Dharani, K. Danesh 
  Abstract: Boundary value problems (BVPs) arise in various scientific and engineering disciplines, including physics, finance, and biology Classical computers may take too long to solve these issues numerically Quantum computing may solve this issue using quantum parallelism to calculate quicker than classical computers Recent quantum computer research shows that Quantum Fourier Transform (QFT) can efficiently solve Schr
  Keywords: Boundary value problems; Quantum computing; HHL algorithm; Quantum phase; estimation algorithm; Linear Solver; Quantum Fourier Transform.
  DOI: 10.1504/IJSSE.2025.10059826
   
  
• Feature Interpreted Convolutional Neural Networks for Real Time Implementation of Respiratory Data  
  by Rampriya R, Suguna N, Sudhakar P 
  Abstract: Automatic detection of respiratory diseases is important to prevent any sudden death in patients. At present, respiratory diseases are detected by a physician who normally consumes more time to detect. In this work, the respiratory data from the persons are classified into either normal or abnormal using the proposed deep learning architecture. The proposed work consists of two subsequent phases namely training and testing. In training phase of the Respiratory Classification System (RCS), the respiratory data from both normal and both abnormal cases are individually data augmented in order to eliminate the overfitting issues in deep learning architecture. This data augmented respiratory data from both normal and abnormal case is used to construct Data Augmented Matrix (DAM) which is trained by the proposed Feature Interpreted Convolutional Neural Networks (FICNN) to produce the trained data. The proposed FICNN work obtained a 99.9% RDR with 0.05 ms as computational time.
  Keywords: FICNN; Over fitting; Data Augmentation; Deep Learning; Respiratory Classification System; RDR.
  DOI: 10.1504/IJSSE.2025.10060148
   
  
• Pulmonary Carcinoma cells Classification using a Novel DCNN model Integrated with Cloud Computing Environment  
  by Sudha R, Umamaheswari K.M 
  Abstract: In this work, we presented a carcinoma cells classification of Non-Small Cell Lung Cancer (NSCLC) which is a more difficult challenge in CAD detection. So, a modified CADx is being investigated to alleviate radiologists' excessive work and the need for the following interpretations. We describe an approach for identifying and verifying different types of pulmonary carcinoma. In addition, a novel deep convolutional neural network (DCNN) and data were obtained via a cloud system for classifying lung nodule cell types in this study. As an integrated approach for CT images, the presented system includes a Cloud-based Lung Carcinoma cell Classifier. The suggested Cloud Based -on LCC first applied active snake model-based segmentation. A Novel DCNN for identifying distinct malignant cells of lung nodules is designed and verified using open sources Lung images. When compared to current strategies, our suggested technique reaches an accuracy of 96%, which is higher than other models.
  Keywords: Artificial Intelligence; Cloud Computing; CT Scans; Deep Neural Networks; Pulmonary Carcinoma.
  DOI: 10.1504/IJSSE.2025.10060240
   
  
• Seeker Optimization Algorithm with Deep Learning based Feature Fusion Model for Tomato Plant Leaf Disease Detection  
  by Jayaprakash K, Balamurugan S.P 
  Abstract: The study focus on design and development of the Seeker Optimization Algorithm with Deep Learning based Feature Fusion Model for Tomato Plant Leaf Disease Detection (SOADLF-TPLDD) technique. The goal of the SOADLF-TPLDD technique is to apply DL technique for the segmentation and classification of plant disease. In the primary stage, the SOADLF-TPLDD technique uses U2Net model for background removal and UNetPP model for segmentation process. Besides, a feature fusion of two DL models takes place namely InceptionV3+EfficientNetB2. For disease detection and classification, Attention Convolutional Gated Recurrent Unit (ACGRU) model is applied. Furthermore, the SOA is used for optimal hyperparameter selection of the ACGRU model. Finally, the recommendation of pesticides for the detected plant diseases takes place using matrix factorization (MF) approach. The stimulation outcomes of the SOADLF-TPLDD method on benchmark dataset are validated and the outcomes represented the betterment of the SOADLF-TPLDD method over other existing techniques.
  Keywords: Tomato; Plant leaf diseases; Deep learning; Segmentation; Seeker Optimization Algorithm.
  DOI: 10.1504/IJSSE.2025.10060560
   
  
• Database Systems under Rayleigh Fading Channels: MIMO-NOMA Based Performance Evaluation Modeling  
  by Bharathi C, Manjunatha Reddy H. S 
  Abstract: In recent years, Non-Orthogonal Multiple Access (NOMA) has emerged as a transformative technique, revolutionizing spectral efficiency and enabling massive connectivity in next-gen wireless communication. This groundbreaking research delves into MIMO NOMA systems’ power allocation strategies, prioritizing system throughput and user fairness. Our innovative approach leverages QPSK modulation in a comprehensive system model, meticulously simulating performance across various variables, including user distance and power differentials. Across varying bandwidths, our model showcases remarkable improvements in user experience, with a jaw-dropping 12 dB/Hz increase in spectral efficiency at a mere 1 dB SNR boost, coupled with a staggering 15x10-3 and 12x10-3 reduction in outage probabilities at 0.17 dB. Even in the challenging Rayleigh fading channel, the Bit Error Rate (BER) sees substantial reductions, reaffirming the prowess of our proposed methodology.
  Keywords: Bit Error Rate (BER); Rayleigh; Spectrum Efficiency; Outage probability; Multiple Input Multiple Output (MIMO); Database Systems; Non-Orthogonal Multiple Access (NOMA); Simultaneous Multi-User.
  DOI: 10.1504/IJSSE.2025.10060561
   
  
• Improved Gray Wolf Optimisation Based Energy Efficient Spectral Sensing in Cognitive Radio Network  
  by Praveen Hipparge, Shivkumar Jawaligi 
  Abstract: In a 5G heterogeneous network, the Cognitive Radio Network (CRN) must choose amongst energy efficiency and spectrum sensing efficiency. The major goal of existing techniques is to apply convex optimization to solve the energy efficiency optimization problem in spectrum sensing. Real-time spectrum sensing, nevertheless, is a non-convex optimization issue. we propose a novel Improved Gray wolf optimization (IGWO) based approach to detect the enhanced energy usage spectrum holes to overcome the non-convex issues. The cuckoo search (CS) algorithm is used to balance the exploitation and exploration phases of GWO. The energy efficient spectrum can be detected with the factors such as power spectral density, transmission power, and sensing bandwidth. Experiments are demonstrated with the MATLAB simulator and compared the outcomes with the state-of-art works. Our proposed approach surpasses all the other works while considering the selection of energy efficient spectrum holes for the communication.
  Keywords: Spectrum; 5G network; cognitive radio network; optimization; and energy efficient.
  DOI: 10.1504/IJSSE.2025.10060744
   
  
• HSaaS: Healthcare Security Systems as a Service for Cloud-Based Electronic Health Records with Blockchain Integration  
  by Aruna Kumari B, Sreelatha M 
  Abstract: The use of cloud-based electronic health records (EHRs) has led to improvements in healthcare delivery, but concerns about data privacy and security remain Healthcare Security as a Service (HSaaS) is a promising solution to address these concerns and integrating HSaaS with blockchain technology can further enhance the security and privacy of cloud based EHRs Smart healthcare provides an opportunity to improve the security and accessibility of cloud based EHRs, and this research aims to explore the potential of HSaaS with blockchain integration for smart healthcare The research will investigate the current state of security challenges in cloud-based EHRs, evaluate the potential of HSaaS for addressing these challenges, and develop a proof-of-concept prototype that integrates HSaaS with blockchain technology into a cloud-based EHR system The performance and security of the proposed solution will be evaluated, and the benefits and limitations of using HSaaS with blockchain integration for securing cloud based EHRs will be analysed. This research aims to provide a more secure and trustworthy environment for cloudbased EHRs, contributing to improving healthcare delivery.
  Keywords: Cloud-based Electronic Health Records (EHRs); Healthcare Security as a Service (HSaaS); Blockchain; Data Security.
  DOI: 10.1504/IJSSE.2025.10060880
   
  
• Modified oversampling based Borderline Smote with Noise Reduction Techniques for IoT Smart Farming dataset  
  by Suresh M, Manju Priya 
  Abstract: Global population is anticipated to grow exponentially to ten billion in the future years. To feed the globe, agriculture must be prioritised. Agriculture is vital to human survival. Every field plant breeding, agricultural monitoring, automated maintenance systems, sensor use, and agrochemicals has evolved physiologically and technologically. Technology and analytics merge in IoT-based farm data. Machine learning algorithms analyse massive agricultural data. Predictive analytics learning algorithms built with machine learning are fast and effective. The data pipeline's pre-processing stage uses the SMOTE with Noise Reduction, an advanced oversampling technique. To evaluate its efficacy and robustness, this novel pre-processing method is heavily compared to SMOTE, ADASYN, and NRAS. This comparison analysis assesses our updated technique's precision, recall, and accuracy in class imbalance scenarios, a common machine learning difficulty. To increase synthetic sample quality and model prediction, address dataset noise and borderline occurrences. This article states that pre-processing procedures shape machine learning models,
  Keywords: Machine Learning; IoT Smart Farming; Noise Reduction; Modified Borderline Synthetic Minority Over-sampling Technique (SMOTE); Data pre-processing.
  DOI: 10.1504/IJSSE.2026.10061579
   
  
• Oppositional remora based AUV path planning with optimal energy efficient data collection in underwater wireless sensor network  
  by A. Arul Rex, Jemila Rose R, Alphiya A. 
  Abstract: In the proposed model, the data packets are sent from the autonomous underwater vehicles (AUVs) to the base station. When creating underwater wireless sensor networks, it is essential to consider the energy consumption of AUVs. Still, it can be difficult to choose an AUV's optimal path. To solve this issue, the suggested method utilizes the optimal path selection. This is addressed by the oppositional remora optimization algorithm (OROA). The performance of the proposed method is assessed in terms of residual energy, network lifetime, delivery ratio, energy consumption, and delay. The suggested solution provides average performance parameters of 95% packet delivery ratio, 12032J residual energy, 1.54s delay, and 3908s network life. The suggested model is exceptionally energy-efficient because it achieved the highest delivery ratio while consuming the least amount of energy. The proposed approach is implemented on the NS2 platform.
  Keywords: Underwater wireless sensor networks; Manhattan based k means; salp swarm; remora optimization and autonomous underwater vehicles.
  DOI: 10.1504/IJSSE.2026.10061692
   
  
• Develop a Convolutional Neural Network Architecture to Accurately Detect and Track Moving Objects in Video Sequence Systems  
  by P. Nagaraju, Manchala Sadanandam 
  Abstract: Object motion detection constitutes the initial crucial step in collecting data about moving objects. The research offered a precise video sequence system motion detection method using a novel object tracking and recognition method through faster region convolutional neural network (R-CNN) that enhances object detection accuracy. To associate things, looks and improved motion are used. The RoI pooling layer uses max pooling to create a compact feature map with a given spatial extent from all admissible region of interest features. The assessment findings demonstrate that the performance of existing work has improved by minimising identification transitions and segmentation. Visual examination, accuracy testing, and comparison with other methods were used to examine the suggested techniques detection outcomes. The proposed Project is implemented using Python software. The FRCNN architecture outperforms other conventional techniques, such as the R-CNN, convolutional neural network (CNN), and deep neural network (DNN), with an accuracy rate of 97.31%, demonstrating a greater effectiveness.
  Keywords: Convolutional Neural Network; faster R-CNN; Network Architecture; Object Motion Detection; Track Moving Objects; Video Sequence Systems.
  DOI: 10.1504/IJSSE.2026.10061732
   
  
• Radio Fingerprint-based UAV Detection and Identification using Discrete Wavelet Feature Extraction and Deep Learning approaches  
  by Khush Attarde, Sameer Sayyad, Satish Kumar, Arunkumar Bongale 
  Abstract: The increasing use of drones in various industries presents security detection challenges due to their high altitude and remotely controlled capabilities. Researchers have developed methods for identifying UAVs, including camera, audio, radar, and thermal-based techniques. Radio fingerprinting is the effective method for detecting drones at high altitudes. This research used Discrete Wavelet Transform (DWT), Machine Learning (ML) and Deep Learning (DL) models for UAV identification and detection. The Random Forest feature selection method improved classification models’ accuracy and reduced computational time. The LSTM model demonstrated promising results, achieving classification accuracy of 95.27%, 87.54%, 82.21%, and 96.7% for identifying UAVs, controller signals, UAV model specifications, and mode of operation. It also demonstrated accuracy of 86.42% and 88.79% in non-line of sight. This research offers valuable insights into practical methods for identifying and detecting UAVs, with significant commercial implications.
  Keywords: Deep Learning; Discrete Wavelet Transform; Radio Fingerprinting signals; Time Domain Features; UAV detection and characteristics identification.
  DOI: 10.1504/IJSSE.2026.10061822
   
  
• An Optimized Deep Auto Encoder with Enhanced Extreme Learning Machine Model for Heart Disease Prediction and Classification  
  by Duraisamy M, Balamurugan S.P 
  Abstract: This research proposes a comprehensive and innovative approach that includes an optimised deep auto encoder with an enhanced extreme learning machine (ODAE-EELM) model. The model combines a novel deep encoder with red deer optimisation (RDO) for feature selection, and an extreme learning machine (ELM) with Stochastic gradient descent (SGD) optimisation for classification. The presented ODAE-EELM model employs pre-processing to convert the actual data into a usable format. Next, the integration of RDO in the encoder optimises the feature selection process by mimicking the foraging behaviour of red deer to enhance exploration and exploitation, thereby yielding more robust and discriminative features. The extreme learning machine is employed for the classification stage due to its simplicity, efficiency, and ability to handle high-dimensional data effectively. The ELM model is optimised using Stochastic Gradient Descent, which ensures faster convergence and efficient utilisation of computational resources. Experimental results shows, our proposed method attained the maximum accuracy of 99%.
  Keywords: Red Deer Optimization; Deep Auto Encoder; Feature Selection; Extreme Learning Machine; Stochastic Gradient Descent; Heart Disease Classification.
  DOI: 10.1504/IJSSE.2026.10062277
   
  
• A NewMethod for Minimizing PAPR in LFDMA Systems  
  by Lekshmi R. Nair, Sakuntala S. Pillai, Shiny G 
  Abstract: Single carrier frequency division multiple access (SC-FDMA) is a standardised technology for high data rate uplink communications in prospective cellular networks that can meet the increasing needs of users. When compared to orthogonal frequency division multiple access (OFDMA), SC-FDMA strategically blends the low peak to average power ratio (PAPR) of single-carrier systems, which demonstrates higher performance for a high data rate, with bit error rate (BER). This proposed model examines the impact of subcarrier mapping on PAPR using the selected mapping approach for localised single carrier frequency division multiple access (LFDMA), a sort of SC-FDMA uplink system. The primary objective of this paper is to attain reduced PAPR and better BER performance by using an efficient modified Chaotic Whale Optimisation Algorithm (MCWOA). The suggested method improves PAPR by 5 dB using QPSK and by 4.3 dB utilising 16-QAM and 64QAM modulation techniques, respectively.
  Keywords: BER; CWOA; LFDMA; PAPR; SC-FDMA; SLM.
  DOI: 10.1504/IJSSE.2026.10062776
   
  
• Blockchain and Cloud-Powered Architecture for Enhancing Efficiency and Security and in Smart Home Environments  
  by Priya Iyer K.B, Ramesh Swaminathan, Sinduja B, Dixy F.L 
  Abstract: However, the widespread adoption of IoT devices has raised significant concerns regarding data privacy and security. This paper presents a novel approach to enhance the security of IoT-based home automation systems through the implementation of blockchain technology. The primary objective of this research is to enhance the smart home network's efficiency, scalability, and security. In this paper, we introduce a novel architectural framework for a sustainable smart home service that incorporates both blockchain and cloud computing technologies. Since the blockchain does not depend on a centralized authority, it efficiently transacts data among the IoT devices. The four main segments of the proposed architecture are the home segment, cloud computing segment, blockchain segment, and services segment and the renderers. These results underscore the efficacy of blockchain technology as a robust platform for constructing a smart home environment.
  Keywords: IoT; blockchain; smart home; cloud; efficiency; scalability; security.
  DOI: 10.1504/IJSSE.2026.10062777
   
  
• Combined Enhanced Caesar Cipher Algorithm and steganography for Biometricsecurity  
  by Gurumurthy S. B, Ajit Danti 
  Abstract: When biometrics data is transferred in the network, it is exposed to others, and if a hacker obtains the data on the network, the biometrics system's security is compromised. As a result, it is critical to maintain the data security for biometrics while it is transmitted via an untrustworthy channel. Therefore, we present a steganography technique for biometric data security in this study. First, the fingerprint image (secret image) is converted to generate the encrypted image. To encrypt fingerprint images, an enhanced Caesar Cipher Algorithm (EC2A) has been created. The dual-tree complex wavelet(DTCWT)algorithm is then used to transform the cover image. The high-frequency wavelet transform is selected for data embedding. During the embedding process, the cover image has the encrypted biometric data embedded into it. Following the embedding process,a stego-image is obtained. The PSNR, SSIM, NC, MAE, BER, and NAE are used to evaluate the performance of the technique.
  Keywords: Embedding; biometric; EC2A; DTCWT; steganography; fingerprint image.
  DOI: 10.1504/IJSSE.2026.10062810
   
  
• Unifying Advanced Meta-Systems for AI-Driven Bipedal Robot Locomotion  
  by Sivakani R, Mahasree M, Sunilkumar Kn, Puneet Mittal, S.Chitra Selvi, Sukhwinder Singh Sran 
  Abstract: In the age of digitization, the integration of complex meta-systems has become increasingly crucial. This paper explores the convergence of artificial intelligence (AI) and robotics, two fields that exemplify such integration by combining elements from computer science, mechanical engineering, and electronics engineering. Here, we focus on the development and evaluation of a self-adaptive, two-legged mobile robot. This robot not only mimics human activities but also learns and makes decisions based on its environment, thus serving as a humanoid meta-system. The study employs advanced machine learning algorithms namely, Na
  Keywords: Mobile robot; Locomotion; Humanoid; Naïve bayes; Support vector machine; Modelling adaptive control; Bipedal robots; A single inflexible body; Artificial intelligence.
  DOI: 10.1504/IJSSE.2026.10062816
   
  
• XLM-R-BiLSTM-CNN: Ensemble Deep Learning Algorithm for Code-Mixed Sentiment Systems Analysis  
  by C. Kumaresan, P. Thangaraju 
  Abstract: This research addresses the challenge of analysing sentiment in text that combines different languages, known as code-mixed sentiment analysis. The study presents a specialised deep-learning algorithm designed for code-mixed sentiment analysis in English and Tamil. The approach combines language identification and sentiment analysis models to improve accuracy in classifying sentiments in code-mixed text. The process involves pre-processing and tokenising the code-mixed data and using pre-trained word embeddings specific to each language. The XLM-R-BiLSTM-CNN architecture is used to create separate sentiment analysis models for English and Tamil. The ensemble method utilises language identification to select the appropriate sentiment analysis model for each token. The sentiment predictions from both models are combined, considering language-specific weights, to produce the final sentiment prediction. Experimental assessments on a code-mixed dataset demonstrate that our ensemble approach outperforms baseline models in terms of accuracy, precision, recall, and F1 score. This proposed technique significantly enhances the accuracy of of sentiment analysis in multilingual code-mixed data, making it a valuable tool for understanding sentiments in various language contexts.
  Keywords: Code-mixed sentiment analysis; ensemble deep learning; language identification; BiLSTM; CNN; XLM-R; Mixed Sentiment Systems Analysis.
  DOI: 10.1504/IJSSE.2026.10062830
   
  
• A Systematic Literature Review of Secure and Advanced Software Defect Prediction  
  by Ayushmaan Pandey, Jagdeep Kaur 
  Abstract: Software defect prediction is a critical component of software development. It enables developers to find and address potential issues before they turn into significant ones. In this paper different problems have been discussed that arise when attempting to predict software defects, including the requirement for reliable and effective data balancing and feature selection techniques to deal with the complex datasets frequently used in this field. We examine how data security, class imbalance, and feature selection issues are related to the performance of software defect prediction models. We have emphasized the significance of taking security concerns into account when predicting software defects and talking about potential solutions to these problems. This survey paper offers insights into these primary issues of data security and reliability of software defect prediction models and their potential solutions in addition to a thorough analysis of the current advancements made in software defect prediction today. After a detailed study of SDP techniques developed in the past few years, we have provided some future challenges and recommendations that may further enhance the performance of the current models.
  Keywords: class imbalance; homomorphic encryption; differential privacy; feature selection; software defect prediction; federated learning.
  DOI: 10.1504/IJSSE.2026.10062929
   
  
• Weapon Detection Technologies for Surveillance under Different YoloV8 Models on Primary Data: Smart City Based Approach for Safe Society  
  by Rohit Rastogi, Yati Varshney, Jagrati Sharma 
  Abstract: This comparison between the yolov8s.pt and yolov8x.pt YOLOv8 models is very important for real-time applications, particularly for object recognition and surveillance. Based on the results, the 95% precision and recall of the yolov8s.pt model, together with its 96% mean average precision, demonstrate the models usefulness in situations requiring precise and quick object recognition. This model has potential applications in a variety of security systems, supporting security protocols in high-risk areas such as airports, public areas, and high-security enterprises by assisting in the quick identification of possible threats in real-time surveillance data. Conversely, the yolov8x.pt models better performance which includes an astounding 98% precision and 99% mean average precision highlights its effectiveness in demanding real-time applications that need exacting accuracy. Because of its complex capabilities, the model is a great fit for use in cutting-edge applications that require quick and accurate object recognition, such as autonomous driving technologies and sophisticated surveillance systems.
  Keywords: Convolutional Neural Network (CNN); down-sampling; Optimization; Weapons; detection; Surveillance; Object Detection; Thermal Imaging; Wave Scanning; Security Infrastructure.
  DOI: 10.1504/IJSSE.2026.10063444
   
  
• Harnessing Core-Accelerated Machine Learning for Enhanced Knowledge Acquisition in System Integration  
  by Biswaranjan Senapati, Swati Vashisht, Basant Kumar, Mohammad Sajid Mohammadi, Kamalakannan V, Karuna Markam 
  Abstract: In machine learning, the integrity of labelled datasets is paramount for achieving high model accuracy. This study introduces an innovative method leveraging accelerated machine learning (AML) to rectify mislabelled data autonomously, a common issue that detrimentally affects model efficacy. AMLs fast training features allow iterative label modifications and enhancements, enhancing model accuracy dynamically. Initial results show a considerable boost in model precision, especially in datasets with significant initial mislabelling errors. This breakthrough underscores our methodologys capacity to tackle a fundamental obstacle in machine learning: maintaining the quality of training data. Implementing our AML-based label correction strategy also diminishes the necessity for labour-intensive and prone-to-error manual label checks. Efficiency simplifies data preparation and model training. This research increases machine learning by merging AMLs sophisticated features with automatic label correction. It boosts accuracy, efficiency, and reliability in data-intensive companies. This new phase in machine learning improves model performance with fewer resources, demonstrating the benefits of integrating AML with automated data quality processes.
  Keywords: System Integration; Accelerated Machine Learning (AML); Knowledge Acquisition; Dataset Integrity; Model Accuracy; Iterative Refinement; Autonomous Labeling; Machine Learning Efficiency,.
  DOI: 10.1504/IJSSE.2026.10064490
   
  
• Severity Assessment of Code Smells using Multi-Criteria Approach  
  by Abhishilpa Nandini, Randeep Singh, Amit Rathee 
  Abstract: Context. Code smells belonging to the source code of software indicate that the source code is flawed and has design/ structural problems. In the case of large and complex software, handling all these code smells is generally not considered the best solution due to factors such as time and maintenance budget constraints. Objective. This paper targets exploring new factors that help in improving severity classification accuracy from the developer's perception. Method. We introduce a new multi-criteria based severity classification approach that combines four key factors that reflect the developer's perception. Results. When comparing the proposed approach results with the developer's perception of maintenance effort, the proposed approach is found to return justifiable accuracy that ranges between 78%-93%. Conclusions. Based on the obtained results after empirical evaluation, the proposed approach is found to be useful in estimating severity and obtaining potential critical candidates for refactoring.
  Keywords: code smell; prioritization; ranking; severity classification; modular importance; co-occurrence; code smell activeness.
  DOI: 10.1504/IJSSE.2026.10064957
   
  
• Comparative Model Study for Driver Drowsiness Detection System: A Road Safety Mechanism for Smart Cities 5.0 in Developing Countries of South-Asian Continent  
  by Rohit Rastogi, Manvi Tyagi, Pranav K, Pranjal Vaidywan 
  Abstract: In this research work, the author team implemented a driver drowsiness detection system using three different models: VGG16, InceptionV3, and Xception. Transfer learning was employed, leveraging the prior knowledge encoded in these pre-trained models to enhance the efficiency of our system. The author team utilised Haar Cascade for face and eyes detection from the video stream. After detecting eyes, the trained model predicted whether the eyes were open or closed. If the persons eyes remained closed for an extended period, an alarm would go off, waking up the driver. Live testing on the laptop itself demonstrated that the Xception model provided tremendous results, reinforcing the decision to use transfer learning with the Xception model over VGG16 and InceptionV3.While our comparison included three models, its essential to acknowledge that transfer learning offers a powerful approach to leverage pre-existing knowledge and enhance the performance of machine learning models for specific tasks, such as driver drowsiness detection.
  Keywords: Computer Vision; Datasets; Deep Learning; Drowsiness; Haar Cascade; InceptionV3; Neural Networks; OpenCv; Transfer Learning; VGG16; Video Stream; Xception.
  DOI: 10.1504/IJSSE.2026.10064960
   
  
• A Multi Parametric Partial Computation Offloading Technique in Mobile Cloud Computing  
  by Priyajit Sen, Rajat Pandit, Debabrata Sarddar 
  Abstract: Offloading parameters refer to energy consumption, network delay, jitter, bandwidth, user mobility, physical distance from cloudlet to mobile device and others that should be considered while offloading a task to a nearby cloudlet. A task in mobile device should consider all the above parameters for offloading. Again, a task can be offloaded to the cloudlet either partially or completely. When the entire task in offloaded to the cloudlet then it requires additional time, storage, and energy requirement but a partial offloading can take a portion of the application to be offloaded to the cloudlet and that will save the additional computing resources as required in full offloading. In this paper, we have introduced a multi-parametric partial computation offloading technique that will consider multiple parameters of offloading as well as instead of the entire task to be offloaded to the nearby cloudlet, part of the computation will be considered for offloading.
  Keywords: Mobile Cloud Computing; Delay; Power Consumption; Deadline; Bandwidth; Jitter; Mobility; Multi-parametric Offloading; DCIM; Partial Offloading.
  DOI: 10.1504/IJSSE.2026.10064962
   
  
• Enhanced Computer Network Performance with Decision Tree-Based Congestion Control Systems  
  by Hemali Mohit Moradiya, Kalpesh Popat 
  Abstract: This study examines internet speed and congestion control to optimise data traffic management. Tahoe, Reno, NewReno, Westwood, Westwood+, and Vegas TCP versions are tested for data transmission issues. Westwood+ substantially increases data transfer. We also examined intelligent congestion control with machine learning. Two decision tree-based supervised learning methods were suggested. The recommended decision tree uses numerical and categorical predictors. Category predictors were less accurate than numerical forecasters in decision tree sim-ulations. NS3 simulations of bus, ring, and dumbbell network topologies' practical effects. Data show Westwood+ works effectively with decision trees. Finally, this research boosts internet speed. It recommends greater research and praises accomplishments. In conclusion, the paper suggests improvements and research. Simulations em-ployed NS3 with Bus, Ring, and Dumbell topologies. This study uses NS3 to emulate Westwood+s decision tree performance with and without supervised learning. Westwood+ scored higher with decision trees. Our study uses multiple random networks to test its efficacy.
  Keywords: Bus; Decision tree; Dumbell; NewReno; NS3; Ring; Throughput; Vegas; Westwood+. Computer Networks; Conges-tion Control Systems; Network Performance; Data Traffic; Bandwidth Optimization,.
  DOI: 10.1504/IJSSE.2026.10065354
   
  
• Delay-dependent H filtering for 2D continuous state-delayed systems  
  by Khalid Badie, Laila Dami, Zakaria Chalh 
  Abstract: This paper discuss the problem of H filtering for two dimensional (2D) continuous systems with delays. First, delay-dependent stability conditions are proposed, where a delay-partitioning method and a free-weighting matrices technique are used. These proposed stability conditions are less conservative than the existing results. Second, the obtained results are extended to the H performance analysis. Then, the same method is employed to design a suitable H filter of the Luenberger observer type, such that the filtering error is asymptotically stable and satisfies a given H performance level. All the given results are formulated in the form of linear matrix in equalities(LMIs).Lastly, three numerical examples are developed to show the benefits of the given theoretical results.
  Keywords: Delays; Two dimensional systems; Roesser model; Delay partitioning; Delay-dependent stability; $H_{\infty}$ performance; Filter design.
  DOI: 10.1504/IJSSE.2026.10065584
   
  
• Deep Learning Algorithm Based Pregnant Women Heart Disease Prediction using Heart Beats  
  by Magesh S, Rajakumar P.S 
  Abstract: The main objective of this study is to acquaint participants with the procedure for diagnosing maternal cardiac disease and the function of a multidisciplinary team in its management. Heart failure, arrhythmias, and a worsening of the heart condition can all complicate pregnancy and raise the risk of morbidity and mortality for both the mother and the foetus. The doctor may now provide both normal and specialised care during pregnancy thanks to the availability of specialised guidelines on pregnancy and heart disease in the last ten years. Rely only on Deep Learning (DL) for screening pregnant women in the second trimester for congenital heart disease (CHD) and foetal ectasia. This groundbreaking research not only targets FE in the first trimester but also develops a revolutionary information system platform with a number of tools for interpreting common ultrasound data, making it innovative and ground-breaking. This gives the project its uniqueness and ingenuity.
  Keywords: Deep learning; foetal cardiology; echocardiography; decision support systems.
  DOI: 10.1504/IJSSE.2026.10065636
   
  
• LSS-AKE: Designing Lightweight, Secure and Scalable Authentication and Key Establishment Scheme for IoT-enabled Smart Grid Communication  
  by Akhil Chaurasia, Udai Pratap Rao, Alok Kumar 
  Abstract: The Internet of Things(IoT)advancement has notably enhanced power quality and reliability management, especially in smart grid (SG) systems. Due to the nature of the bi-directional communication of the SG system, a critical issue that occurs in the SG systems is to secure neighbourhood area network (NAN) communication between smart meters (SMs) and gateways (GWs) via potentially insecure channels. To address this issue, we introduce a lightweight, anonymous Schnorr digital signature-based authentication scheme utilising elliptic curve cryptography (ECC) based certificates for mutual authentication between SMs and GWs. The schemes security has undergone comprehensive evaluation, including testing its session key (SK) security under the real-or-random (ROR) model. Additionally, the LSS-AKE has been rigorously tested against various known attacks and formally verified using the AVISPA tool, confirming its robustness. Performance analysis using MIRACL demonstrates the LSS-AKE schemes superiority in reducing computational overhead and communication costs compared to existing solutions.
  Keywords: Smart Grid (SG); Elliptic Curve Cryptography (ECC); Smart meter (SM); Authentication.
  DOI: 10.1504/IJSSE.2026.10065724
   
  
• Authentication Method Automated Internet System Searches using Machine Learning  
  by R. Ramyadevi, Varun Kumar Nomula, Kiran Peddireddy, Sandeep Rangineni, Srinivas Venkata, Yousef Jaber Alawneh 
  Abstract: As internet information grows, searching and retrieving relevant content has become increasingly important. Traditional search engines use static algorithms and keyword matching, often failing to provide the most relevant or updated results. This study introduces a new internet searching method using ML and AI to enhance precision, relevance, and adaptability. Our automatic internet search engine, AISE, employs deep learning models to understand user intent, context, and history. AISEs algorithms refine search queries, recognise patterns, and adapt to user needs, offering a personalised and dynamic search experience. NLP allows the system to comprehend complex questions in natural language, eliminating the need for keyword inputs. AISE uses reinforcement learning to optimise search results based on user interactions continuously. Early tests show that AISE delivers more relevant results faster than traditional search engines. This paper details AISEs technical foundations, design concepts, and potential applications, revolutionising internet search paradigms.
  Keywords: Deep-learning; Internet Search; Machine Learning; Artificial Intelligence; AI-Enabled Searches; Natural Language Processing; Contextual Search; Personalized Search.
  DOI: 10.1504/IJSSE.2026.10065951
   
  
• NVIDIA Powered Smart Animal Detection for Agriculture  
  by Anushka Londhe, Sai Sreeprada T, Pandian A 
  Abstract: The integration of advanced technologies, such as sensor networks, machine learning, and computer vision, exhibits a vital role in the success of effective animal detection systems for crop fields. These systems are used in agricultural areas and promptly alert farmers through alarm mechanisms. Early detection of animals can prevent crop damage, minimize economic losses, and ensure sustainable food production. Ongoing research aims to enhance the reliability of these systems through the integration of multi-sensor platforms, improved machine learning algorithms for accurate animal recognition, and the deployment of autonomous drones for comprehensive monitoring. By leveraging these technologies, farmers can implement proactive measures, such as deterrents or protective barriers, to safeguard their crops from animal intrusions. This review draws upon a corpus of approximately 15 research papers exploring topics like sensor network optimization, environmental monitoring, alarm system integration, and the use of drones for data collection and surveillance.
  Keywords: Animal Detection; Agricultural Surveillance; life Monitoring; Computer Vision; Deep Learning; MobileNet SSD; Auditory Deterrents; Farmer Notification; Sustainable Farming,.
  DOI: 10.1504/IJSSE.2026.10066184
   
  
• Music Recommendation System Using Emotion Detection  
  by Priyam Chakraborty, Mehraan Zahoor Wani, Sadiq A. K, Umamaheswari K.M 
  Abstract: The proposed system uses machine learning algorithms to analyse the user's facial expressions and vocal tone to detect emotions such as happiness, sadness, anger, and excitement. These emotions are then mapped to a set of music features such as tempo, rhythm, and key, which are used to generate a playlist that matches the user's current emotional state. The system was evaluated using a data set of user-generated music playlists and achieved an accuracy of over 40 percent in predicting the user's emotional state. Furthermore, user feedback on the recommended playlists was generally positive, with users reporting that the playlists accurately reflected their current emotional state and introduced them to new music they enjoyed. Overall, the proposed music recommendation engine offers a more personalized and accurate approach to music discovery, using emotion detection to provide recommendations that match the user's current mood and emotional state.
  Keywords: Machine Learning Algorithms; biometric data; facial expressions; emotion detection algorithms.
  
  
• Evaluation of the Accepting LMS in Malaysia Universities using a Quantitative Approach  
  by Fouad Salman, Zuriana Abu Bakar 
  Abstract: A learning management system (LMS) combines information technology (IT) and academia and has developed as a fruitful tool in e-learning. Thus, evaluation of LMSs is essential for ensuring effective outcomes, practical use, and tangible achievement for learners. In this paper, the authors investigate factors influencing student usage of LMS in higher education institutes. Based on the investigation, a model is proposed to assess the effectiveness of LMS in Malaysian universities. The model is built with three external variables by extending the existing technology acceptance model (TAM). Using quantitative techniques, the model is validated based on fitting it to data from a survey involving 300 students engaged with LMS in Malaysian universities. The survey was distributed online to degree students enrolled in various academic programs at three universities in Malaysia. The results provide empirical evidence indicating that the highlighted factors significantly increase the LMS quality.
  Keywords: Learning Management System; LMS; E-learning; Evaluation model of LMS.
  DOI: 10.1504/IJSSE.2026.10066559
   
  
• Crop Yield Prediction Using Machine Learning Hybridization-Based Crop Yield Prediction (MLH-CYP) System  
  by M. Revathi, Kanya V, Victo Sudha George 
  Abstract: With the rise of extensive data, machine learning techniques have become more prevalent in recent years. Most of the worlds soil characteristics, weather, temperature, pesticide, fertiliser, and other related variables are part of agricultural data. Estimating crop yields is crucial in agriculture for several reasons, including making better decisions and increasing productivity. Predicting crop output is a challenging endeavour that relies on past data to estimate a years potential yield. Pre-harvest yield forecasts are essential for agricultural logistics, market planning, and food delivery. For long-term yield estimates, climatic and phenological factors must be tracked to understand crop growths complex interactions. The research uses hybrid machine learning (ML) to predict and improve agricultural production estimates. 'Machine learning hybridisation-based crop yield prediction (MLH-CYP) predicts agricultural production using XGBoost and Random Forest (RF) regression via voting. This unique strategy combines various strategies to increase forecast accuracy and durability. Random Forest is best for complex interactions and overfitting. However, XGBoost has the best prediction performance and computing economy.
  Keywords: Agriculture and yield; Randomized search CV; Yield Prediction System; Machine Learning (ML); Random Forest (RF); Machine Learning Hybridization-Based Crop Yield Prediction (MLH-CYP).
  DOI: 10.1504/IJSSE.2026.10067295
   
  
• An Empirical Analysis of Data Mining Technique to Predict Learning Environment in Higher Education Systems  
  by Shilpa K, Krishna Prasad K. 
  Abstract: Predicting and assessing learning situations improves higher education. This is where data mining (DM) can help derive inferences from enormous datasets. Multiple data mining methods uncover trends and patterns in educational datasets. These databases contain feedback, student performance, stakeholder participation, and demographics. ML algorithms predict student performance and identify classroom challenges. Course materials, learning analytics, academic performance, and student demographics are variables. The model analyses past data patterns and interactions in real-time to reveal complex correlations between multiple learning environment parameters. This enables customised support plans and proactive interventions for teachers and students. Teachers can create a more personalised and supportive learning environment by identifying at-risk kids and providing targeted support. The student success prediction model is assessed using accuracy, precision, recall, F-measure, and sensitivity variables. High questionnaire reliability, student satisfaction and academic achievement were also linearly related in the ANOVA study of stakeholders satisfaction with higher education courses.
  Keywords: Higher Education Systems; Data Mining (DM); Machine Learning (ML); Predictive Models and T-test; Complex Meta-Systems; Data collection and preprocessing; Information Technology,.
  DOI: 10.1504/IJSSE.2026.10067771
   
  
• Functionality Oriented Hybrid Software Remodularisation Using Structural and Dynamic Features of Software  
  by Naveen Sharma, Randeep Singh, Amit Rathee 
  Abstract: Context: Modern software systems complex structure makes maintenance difficult due to design negligence. To improve software evolution, software remodularisation is necessary. This process uses clustering techniques and features to improve the systems modular structure. Genetic algorithms (GAs) are widely used for clustering and recovering modular software structures. Objective: This paper investigates the feasibility of software remodularisation, focusing on structural and dynamic information. The goal is to create a functionality-oriented, well-modularised system by customising GA clustering algorithms effectiveness. Method: This paper models coupling at the class level using structural and dynamic dependency information. A weighted scheme integrates these dependency information. The Jaya algorithm customises the basic GA, resulting in a hybrid GA. The remodularised solution is evaluated using MQ and MoJoFM quality metrics. Results: This paper evaluates proposed approach, addressing four research questio
  Keywords: Software modularity; Remodularisation; Software Maintenance.
  DOI: 10.1504/IJSSE.2026.10067774
   
  
• Robust Industrial Monitoring System using NI-DAQ  
  by Venkatanarayana Moram, Kishore Putha, Sonu Kumar 
  Abstract: Industrial monitoring system improves operational efficiency, promotes proactive decision-making, and aids in meeting production and sustainability goals. The suggested system aims to efficiently and robustly control industrial accident causing parameters, minimising their impact on the plant and natural environment. The suggested methodology uses LabVIEW to collect and track real-time data acquisition from multiple sensors, ensuring accurate monitoring of plant external conditions. This system measures gas emissions, plant temperature, light detection and flame detection, assisting for proactive decision making. Users benefit from an intuitive graphical interface for data visualisation, historical trend analysis, and alert notifications for threshold deviations. The system also offers remote monitoring capabilities, data logging, and reporting, enhancing its usability and accessibility. This integrated solution addresses the critical need for precise environmental monitoring, making it applicable across industries such as manufacturing, agriculture, and climate control, ultimately improving efficiency and decision making in these domains.
  Keywords: LabVIEW software; myDAQ; Flame Sensor; Gas Sensor; ThingSpeak.
  DOI: 10.1504/IJSSE.2026.10067930
   
  
• A modelling and simulation-based safety evaluation of an advanced driver assistance system (ADAS) using V-model process  
  by Shivaji Thorat, Ramesh Pawase 
  Abstract: The automotive world is approaching vehicle driving autonomy and has reached level 3 as per Society of Automotive Engineers (SAE) automation levels. The design and validation of such advanced driver assistance systems (ADAS) and automated driving systems (ADS) are challenging and complex considering their actual behaviour on roads from a safety and comfort perspective. A multifaceted ADAS development process is needed to make it a reliable and safe product before its usage on roads. A few solution developers are supporting automotive manufacturers to achieve it in time. But it is not happening at that pace of development. The V-cycle system engineering agile process can accelerate the ADAS development process and launch them on time in the market. This paper proposes a modelling and simulation-based design framework and guidance on validation flow using the system engineering V-cycle process, which is trending in the automobile industry. It can offer reliable ADAS vehicles to customers.
  Keywords: Advanced driver assistance systems; automated driving; system engineering; safety evaluation; modelling and simulation.
  DOI: 10.1504/IJSSE.2026.10068091