Interactive (Poster) Session 1
28 October 2025 (1:30pm - 3:30pm)
28 October 2025 (1:30pm - 3:30pm)
Interactive Session A.1
Room: Interactive Area 1, Foyer of Sipadan (Level 4)
Chairs: Md Pauzi Abdullah (UTM, Malaysia), P. Susthitha Menon (Universiti Kebangsaan Malaysia, Malaysia & Institute of Microengineering and Nanoelectronics (IMEN), Malaysia)
#1 Development of BaRISTA: a Computer Vision-Based Real-Time Integrated Sorting Tool for the Assessment of Green Coffee Beans
Cristina Ysabel Almario, Kyle Aaron Coloma, John Carlo Velonza and Carlos Oppus (Ateneo de Manila University, Philippines)
Small-scale coffee bean producers remain heavily reliant on manual sorting, which can be a tedious, labor-intensive, subjective, and error-prone process. In line with this, the study aims to develop BaRISTA, an automated system that combines a custom YOLOv8 classification model with hardware components to assess and sort Robusta green coffee beans (GCBs) based on the presence of visible defects, providing small coffee producers with a more affordable yet precise and reliable alternative to current manual methods or more expensive equipment. The study entailed developing a custom dataset of Robusta GCBs, training a YOLOv8 classification model to classify the beans based on the presence of visible defects, and testing the model with batches of unseen beans. Testing showed the classification system's effectiveness as the model achieves a precision of 97.794%, an accuracy of 99.375%, a recall of 99.750%, and an F1-score of 98.762%-all of which demonstrate its strong capability in classifying GCBs.
#2 Practical 2D FSM for Stochastic Computing with Improved Hardware Efficiency and Accuracy
Jiho Kim, Gayoung Kang and Youngmin Kim (Hongik University, Korea (South))
As artificial intelligence continues to advance rapidly, traditional binary computing methods face limitations regarding power consumption and hardware complexity. Stochastic Computing, which processes data using probabilistic bitstreams, offers a computational paradigm that enables highly parallelized operations, enhances energy efficiency, and reduces hardware complexity. Despite its potential, the practical application of SC has been constrained by the trade-off between efficiency and accuracy. To address this gap, this study proposes the 4×4 two-dimensional finite state machine architecture and method for Stochastic Computing, demonstrating an optimized structure that improves both efficiency and accuracy. The architecture utilizes independent probabilistic inputs and weight-adjusted state transitions to reduce bitstream correlation and enable more flexible function implementation. This configuration achieves over 1.37× better performance in terms of accuracy, area, and power consumption compared to prior designs, while maintaining a practical and scalable design that balances hardware complexity and computational precision, making it well-suited for neuromorphic computing and other low-power, high-efficiency applications.
#3 Changes in DC Characteristics Due to Changes in Crystal Structure of AlGaN Layer Due to Voltage Stress Application in AlGaN/GaN HEMTs
Sho Nagai, Soichi Sano, Junya Takeda, Hideki Hoji and Hirohisa Taguchi (Chukyo University, Japan)
Voltage stress of 1-12 V was applied to an AlGaN/GaN high-electron-mobility transistor (HEMT) from the drain electrode, and the I-V characteristics were measured precisely. The experimental results revealed that the change in I-V characteristics caused by voltage stress resulted from changes in the crystal structure of the AlGaN layer. There was no significant change when the voltage stress value was approximately 1 or 2 V. When 3-5 V was applied, the drain current value increased, and the initial strain resulting from crystal defects remaining in the AlGaN layer was canceled by the electric field. However, under excessive stress of 6 V or higher, distortions caused by crystal defects occurred again in the AlGaN crystal structure, electron scattering increased, and the current value decreased. The current value changed greatly in the region where the drain voltage was between 1.0 and 3.2 V, but the I-V characteristics matched in the region where the drain voltage was 3.2 V or more. Therefore, it is believed that two-dimensional electron gas (2DEG) formed stably at approximately 3.2 V, and the crystal defects were not affected by the voltage stress change. Furthermore, when the measurement temperature was changed to 80°C and 100°C, an overshoot effect was observed near a drain voltage of 3.2 V. This is considered to be caused by the improvement in carrier mobility resulting from stable 2DEG formation.
#4 Development of High-Performance Multiply-Accumulate (MAC) Unit Design on FPGA for DSP Applications
Nazirul Hafiz Mohd Nazeri (Sophic Automation Sdn Bhd, Malaysia); Siti Zarina Md Naziri and Rizalafande Che Ismail (Universiti Malaysia Perlis, Malaysia)
The multiply-accumulate (MAC) unit performs multiplication and accumulation, which makes it an essential component for digital signal processing (DSP) applications. Designing high-performance MAC units is crucial for enhancing the computational efficiency of modern computer systems. However, achieving a balance between delay, power consumption, and area utilization poses a significant challenge. This study thoroughly compares and analyzes various multiplier and adder architectures, including Vedic multipliers, ripple carry adders, Brent-Kung adders and pipelined Brent-Kung adders. The research aims to design an efficient MAC unit by investigating optimal combinations of these components. The proposed design utilizes a Vedic Multiplier using Brent-Kung adder and Pipeline Brent-Kung adder which are compared with other configurations. The MAC unit is implemented in Verilog HDL and analyzed using Intel Quartus Prime and functionally verified using Intel ModelSim. The design was implemented on the Altera Cyclone V 5CSXFC6D6F31I7ES FPGA. The synthesis process reveals that the proposed design achieves a low delay of 28.506 ns, an area utilization of 360 LUT units, and a balanced power consumption of 427.600 mW, resulting in the speed of 310.850 MHz. The study concludes that combining the Vedic multiplier with Brent-Kung adder and Pipeline Brent-Kung adder significantly improves MAC unit performance.
#5 Digital Dice Design Using Pseudo Random Number Generator
Tee Hui Teo (Singapore University of Technology and Design, Singapore); Maoyang Xiang (8 Somapah Rd & Singapore University Technology and Desgin, Singapore); Junhan Li, Kai Cao and Michael Kee Han Lim (Singapore University of Technology and Design, Singapore)
The primary objective of this innovative project is to develop a digital dice that can display dice numbers in real-time. This digital dice operates by utilizing a pseudo-random number generator (PRNG) that employs the XORshift algorithm. This algorithm is coded explicitly in Verilog Hardware Description Language (HDL) and is integrated into a field-programmable gate array (FPGA). The digital dice itself comprises various components such as a tilt sensor, display screen, power management circuit, and a rechargeable battery, all neatly enclosed within a 3D-printed dice casing. When interacting with this digital dice, users can initiate the generation of a random number by simply shaking the device. This action triggers the tilt sensor within the dice, which in turn generates a seed for the PRNG to produce a random number. The numbers displayed by this digital die range from 1 to 100, effectively simulating the various sides of a traditional dice. This comprehensive kit, aptly named SUTDicey, showcases a perfect blend of cutting-edge technology and practical application in the realm of gaming and probability.
#6 LoRa-Enabled Wireless Sensor Network for Water Consumption Reading and Billing Using Optical Character Recognition
Christian P Javier, Veronniecka T Tolentino, Jasmine Joy L Atienza, Joshua Kyle G Bravo and Francis A Malabanan (FAITH Colleges, Philippines)
The growing need for precise and efficient water consumption monitoring underscores the limitations of traditional meter reading methods, which depend on manual labor and are susceptible to human error, limited accessibility, and delayed billing processes. This research presents a LoRa-based wireless sensor network that automates water meter reading and billing using Optical Character Recognition (OCR) technology. The system employs an ESP32-CAM module to capture images of water meter displays, which are then processed using OCR algorithms to extract numerical data. The extracted readings are transmitted via LoRa communication to a central gateway and stored on a cloud-based platform, which supports a web interface for real-time monitoring. The proposed solution significantly enhances data transmission range, reading accuracy, energy efficiency, and system responsiveness. By enabling continuous and remote monitoring of water usage, the system helps detect leaks early and encourages responsible consumption. This approach benefits both consumers and utility providers by improving operational efficiency, reducing costs, and promoting sustainable water management practices.
#7 From Equations to Innovation: A Bloom's Taxonomy Framework for Scaffolding Higher-Order Mathematical Thinking in Philippine Electronics Engineering Programs
Carlo N. Romero (National University Philippines, Philippines & National University - Clark Campus, Philippines)
This study investigates the integration of higher-order thinking skills (HOTS) into the mathematical competencies required in Philippine Electronics Engineering programs in response to global educational reforms and the evolving demands of Industry 4.0 and Engineering Education 5.0. Using a mixed-methods approach, the research analyzed mathematics course syllabi from ten CHED-accredited universities through Bloom's Taxonomy coding rubrics and conducted semi-structured interviews with fifteen engineering educators. Findings reveal a predominant reliance on lower order thinking skills (LOTS), particularly in memorization and procedural tasks, while under emphasizing problem-solving, critical analysis, and innovation. Key barriers include rigid curriculum structures, limited faculty training on HOTS integration, and assessment practices focused on rote learning. Based on these insights, the study proposes a framework for embedding HOTS into Electronics Engineering mathematics curricula, aligning with sustainable development goals and the competencies needed for Industry 4.0 applications. The framework and recommended curriculum redesign strategies aim to strengthen critical and innovative mathematical competencies among future Electronics Engineers in the Philippines.
#8 Effect of Gender and Walking Condition on Pelvic Orientation and Gait Parameters in Young Adults
Yin Qing Tan (Universiti Tunku Abdul Rahman, Malaysia); Joe Yan Tan (ViTrox, Malaysia)
Pelvic orientation plays a crucial role in gait. While gender-based anatomical differences in pelvic structure are well-documented, their influence on dynamic walking patterns remains underexplored particularly in different walking conditions such as overground and treadmill walking. This study aims to investigate the effects of gender and walking condition on pelvic orientation and spatiotemporal gait parameters among healthy undergraduate students. Fifteen healthy participants walked barefoot at their self-selected pace. Data collected using a three-axis accelerometer. The Mann-Whitney U test was used to examine gender differences while Wilcoxon test assessed differences between walking conditions. No significant gender differences were found in most parameters, except pelvic tilt range which was greater in females, likely due to anatomical differences. In contrast, significant differences were observed between walking conditions. Treadmill walking was associated with lower speed, lower cadence and reduced pelvic orientation, which may be attributed to cautious gait adaptations, limit optic flow, and reduced propulsion forces inherent to treadmill environments. Walking condition significantly influences pelvic orientation and gait parameters, whereas gender appears to have minimal effect in a homogenous, healthy young population. These findings highlight the need to consider walking environment in gait assessments, particularly in clinical and rehabilitative settings. Overground walking may better represent natural gait mechanics, while treadmill use may necessitate adaptation strategies for stability and balance.
#9 Impact Study of Faulty Sensors on Flocking-Based Cooperative Control of Nonholonomic Robots
Lamia Iftekhar (North South University, Bangladesh)
In this paper, we investigate the effects of faulty internal sensors on a group of non-holonomic robots implementing a well-established flocking algorithm. The flocking algorithm has long been proven to have robust and scalable performance for agents with nonholonomic constraints, but it traditionally assumed perfect state information for all agents. To evaluate its resilience in more realistic scenarios, we systematically introduce robots with faulty sensors that incorrectly measure their own states. These agents broadcast their erroneous states to their neighboring agents, who in turn use the compromised information to develop their control input. We conduct a comprehensive analysis across a spectrum of scenarios, varying both the density of robots with faulty sensors and the intensity of sensor errors. Qualitative insights obtained through individual simulation runs as well as aggregate quantitative results are discussed. Four carefully chosen evaluation metrics are employed to quickly assess performance degradation caused by the faulty sensors. We sought to identify critical failure points for common scenarios through two distinct sets of experiments. Our goal is to provide empirical results and discussions that can inspire the design of more robust and cost-effective controllers, specifically by targeting the primary failure mechanisms identified.
#10 Performance of Unmanned Aerial Vehicle Access Points in Cell-Free Massive Multiple-Input Multiple-Output System
Noor S Othman, Afsana Afrin and Abdulrahman Mohamed Shalaby (Universiti Tenaga Nasional, Malaysia)
Cell-free massive MIMO (CF-mMIMO) eliminates fixed cell boundaries by using distributed access points (APs) coordinated by a central unit, enabling spatial diversity and improved throughput. However, fixed ground APs lack flexibility, making them unsuitable for highly mobile users or emergency scenarios. This paper investigates the deployment of unmanned aerial vehicles (UAVs) as access points in CF-mMIMO systems, focusing on optimizing their three-dimensional (3D) placement to maximize the downlink sum rate. A comparative analysis of Particle Swarm Optimization (PSO) and Gradient Descent (GD) algorithms is conducted under varying user densities and numbers of UAVs. An existing air-to-ground path loss model, incorporating UAV altitude constraints, is employed to evaluate UAV placement in CF-mMIMO systems. Simulation results demonstrate that PSO consistently outperforms GD in both convergence speed and achievable throughput. At higher user densities, PSO achieves approximately 3 bps/Hz higher throughput than GD and up to 7.3 bps/Hz gain at the highest density considered. The results confirm the effectiveness of PSO for UAV deployment in CF-mMIMO systems, where its adaptability to dense and dynamic user distributions provides significant performance gains.
#11 Design and Implementation of Posit Arithmetic-Based FIR Filter on FPGA
Keerthana Menon (IIIT Kottayam, India); Jayakrushna Sahoo (Indian Institute of Information Technology Kottayam, Kerala, India); Kala S (Indian Institute of Information Technology Kottayam, India); Nalesh S (Cochin University of Science & Technology, Kochi, Kerala, India)
With increasing advancement in the field of signal processing, there is a need to develop efficient hardware that can support the digital signal processing algorithms for various applications, such as audio and video processing, image segmentation, image enhancement and other medical applications, communication systems, etc. Current systems employ the IEEE-754 format to represent numbers. However, the posit number system has emerged as a promising alternative, offering improved accuracy and dynamic range. In this work, we present a posit arithmetic-based finite impulse response (FIR) filter for signal processing applications. Here, we implement 16-bit and 32-bit posit adders and multipliers, excluding guard, round, and sticky (GRS) bits. These arithmetic modules are used to implement two 5-point FIR filters to smooth out a noisy signal. The proposed posit arithmetic-based architecture is implemented on the Zynq UltraScale+ ZCU104 FPGA device and the hardware resource utilization is reported. It is observed that the 32-bit posit FIR filter achieves a performance comparable to the 64-bit floating-point-based filter at an operating frequency of 28.57 MHz.
#12 Enhanced Local Disparity Map Estimation Using Segment-Side Window-Based Cost Aggregation
Ahmad Fauzan Kadmin (Universiti Teknikal Malaysia Melaka, Malaysia)
Accurate disparity map estimation is critical for computer vision applications such as 3D reconstruction, autonomous navigation, medical imaging, and immersive entertainment systems. Traditional local window-based methods often suffer from edge fattening and texture inconsistency, leading to inaccurate depth estimation. This paper proposes a Segment-Side Window-Based (SSB) algorithm that integrates segment-based cost aggregation to address these challenges. The SSB method combines Truncated Absolute Difference (TAD), Gradient Magnitude (GM), and Census Transform (CT) for robust matching cost computation, followed by Superpixel Linear Iterative Clustering (SLIC) segmentation and Side Window Filtering (SWF) for edge-preserving aggregation. The final stage involves postprocessing and disparity refinement to enhance the accuracy of the disparity map with left-right consistency checking (L-R Check) and median filter. Experimental results on the Middlebury dataset demonstrate that SSB achieves superior performance, reducing the average bad pixel error to 24.7% (All) and 14.6% (Nonocc), outperforming state-of-the-art methods like Bilateral Filter (BF), Guided Filter (GF), and Median Filter (MF).
#13 Distributed IMM-Based Cooperative Object Tracking Using Multiple Roadside LiDARs
Ryota Imai, Masafumi Hashimoto and Kazuhiko Takahashi (Doshisha University, Japan)
This paper proposes a tracking method of objects, such as cars, two-wheelers, and pedestrians, using multiple light detection and ranging sensors (LiDARs) set in an intersection environment. Each LiDAR unit detects objects using deep learning CenterPoint from its own LiDAR point cloud data and exchanges object information (i.e., the position, size, heading angle, and confidence score of the objects) by communicating with its neighboring LiDAR units. Then, the object information is fused. Here, each LiDAR unit estimates the object poses, and the estimates are fused by exchanging the information among the neighboring LiDAR units. A distributed interacting multimodel estimator is employed to accurately estimate the poses of objects under various motion modes, such as stopping and suddenly moving and stopping, in a distributed manner without requiring a central server. Simulation experiments using four LiDAR units with a mesh type of network topology set at signal poles in an intersection environment validate the performance of the proposed method.
#14 Pi-Droponics: an Automated Hydroponics System with Notification System Using Convolutional Neural Network
Curt Johann M Maracha, Shaun Patrick B Calumba, Joshua Shem M Barachina and Analyn Balog (National University, Philippines)
Hydroponics is a soilless farming method that utilizes technology to regulate the growth environment. A two-layer vertical hydroponic system was constructed in a greenhouse utilizing the Nutrient Film Technique, including sensors that keep track of the light intensity, air temperature, TDS, and pH levels. The objective is to develop a fully automated hydroponic system with alerts for monitoring plant growth and detecting diseases. The sensor data is sent to the NodeMCU esp8266 and saved in Firebase, while a CNN algorithm is used to create a plant disease detection model, which is stored in a Raspberry Pi 4B. The app retrieves data from Firebase and sends real-time updates on system conditions, plants' health status, and notifications when sensor values are outside the threshold range or diseases are detected. Accurate pH and TDS readings effectively monitor the greenhouse environment for the hydroponics system. Air humidity, temperature, and light intensity sensors indicate no significant difference between the greenhouse and the outside environment. A disease detection model using CNNs was developed and successfully tested on external plant leaves. The healthy state of the system's plants indirectly validates the model's effectiveness, although direct testing with the system's plants was not conducted.
#15 Design of a Mobile Navigation System for a University Pedestrian
Jan Kevin Albior Galicia (Ateneo de Manila University, Philippines)
This paper presents ADMUNAV, a mobile navigation system developed to assist pedestrian wayfinding within Ateneo de Manila University. Addressing the challenges of navigating a large, congested campus, ADMUNAV utilizes geospatial datasets and Dijkstra's algorithm to compute optimal walking routes in real time. The application features an offline-capable Realm database, intuitive Android user interface, multi-entrance building handling, and dynamic path recalculations with alternative route suggestions. Performance testing demonstrated 100% routing accuracy, an average computation time of 98.9 ms, and minimal memory usage across multiple Android devices. By offering a lightweight, infrastructure-independent navigation solution tailored for university environments, ADMUNAV contributes toward enhancing campus mobility and supports Sustainable Development Goal 9 on innovation and infrastructure.
#16 High-Sensitivity Fiber Ring Microwave Photonics Filter-Based Temperature Sensor with Coarse Sampling Introduced Vernier Effect
Xuewen Shu (Huazhong University of Science and Technology, China)
This paper focuses on the development of a high - sensitivity temperature sensor based on a fiber - ring microwave photonics filter (FR - MPF) by leveraging the Vernier effect introduced through coarse sampling. The FR - MPF is characterized by its straightforward structure, which is formed by connecting the input and output ports of an optical fiber coupler. To validate the proposed concept, a basic experimental setup is meticulously constructed, including a vector network analyzer, a broadband optical source, and other necessary components. When the sampling frequency interval approaches the free spectral range of the FR - MPF, the Vernier effect is effectively triggered. This phenomenon significantly amplifies the sensor's response to temperature variations. Experimental results demonstrate a remarkable improvement in temperature sensitivity. The original sensitivity of - 12.3 kHz/℃ is enhanced by dozens of times, reaching up to - 415.8 kHz/℃. This substantial enhancement indicates that the coarse sampling method is a highly effective approach for boosting the performance of FR - MPF - based fiber optic sensors. It also offers a promising solution for various applications requiring high - precision temperature measurement.
#17 From Triple Helix to Quadruple Helix: the Evolution of University-Government-Industry Collaboration
Yaeko Mitsumori (Osaka University, Japan)
Until the early 2000s, the prevailing assumption in most national innovation systems was that scientific discoveries and inventions would naturally drive economic development and, in turn, societal advancement. The R&D community guided research trajectories in basic, applied, and industrial research, while the public remained passive recipients of innovation. However, over the past two decades, a new approach has gained prominence. Today, research trajectories are expected to be legitimized among relevant publics, designed to create positive public impact, and shaped with public participation. A science park is considered a structure that fosters innovation. Traditionally, the key players in a science park were academia, government, and industry. However, with the advent of the SDGs era, the science park model has been shifting to the Quadruple Helix Model. In this framework, not only academia, government, and industry but also the community and citizens play a crucial role in developing a science park. This paper focuses on Tsuruoka Science Park in Yamagata and examines the transition from the Triple Helix collaboration model to the Quadruple Helix Model.
#18 Democratizing Gait Analysis: Development and Validation of a Customizable Wearable IMU System
Siow Cheng Chan (University Tunku Abdul Rahman, Malaysia); Boon Meng Gan (Universiti Tunku Abdul Rahman, Malaysia)
Gait analysis plays a crucial role in clinical settings for assessing abnormalities associated with musculoskeletal and neurological conditions. Traditional laboratory-based equipment used for gait assessment is costly and impractical for continuous monitoring in daily life. Wearable inertial sensors offer a promising alternative due to their affordability, portability, and user-friendly nature. However, their reliability and validity compared to established techniques remain underexplored. This study proposes a wearable inertial sensor system designed to address these challenges by providing a low-cost, flexible solution that enhances accessibility to gait analysis. The system's reliability was validated against the BTS G-Walk wearable inertial sensor. By employing kinematic equations and advanced filtering algorithms, including a fine-tuned Kalman filter and algorithm, the inherent uncertainties were effectively mitigated. The prototype integrates ESP32 and MPU6050 components to enhance functionality. Comparative analysis with the BTS G-Walk demonstrated high accuracy, achieving over 85% agreement and less than 16% Relative Root Mean Square Error (RRMSE) for gait speed, step count, step length, stride length, and cadence measurements. Overall, this study establishes the proposed system as a reliable and valid tool for spatiotemporal gait assessment in healthy adults, offering a practical solution for clinical applications and research.
#19 EcoFruit14K: a Large-Scale Collection for Spotting Green Produce in Natural Backdrops
Nirban Roy (Institute of Engineering & Management, Kolkata, India); Swapnanil Adhikary (Institute of Engineering and Management Kolkata, India); Shreyan Kundu (Institute of Engineering and Management, India); Susovan Jana (Institute of Engineering & Management, Kolkata, India & Jadavpur University, India); Shuvam Chakraborty (Indian Institute of Technology Delhi, India)
Detecting camouflaged objects in agricultural environments-specifically green lime fruits blending into dense foliage-poses a significant challenge due to minimal color and texture contrast. To address this, we present DLP-14k, a benchmark dataset comprising 14,000 high-resolution images of lime fruits captured under realistic field conditions. The dataset encompasses varied lighting scenarios (morning shadows, midday glare, late-afternoon diffuse light), multiple occlusion levels (overlapping leaves, branches), and includes 1,000 foliage-only negative samples to reduce false positives. We augment the dataset through systematic transformations-random rotations, horizontal and vertical flips, and brightness and contrast adjustments-to enhance model generalization and mitigate overfitting. We evaluate two state-of-the-art deep learning frameworks: YOLOv8, a one-stage detector optimized for real-time applications, and Faster R-CNN, a two-stage detector known for high precision. Experimental results demonstrate that YOLOv8 achieves inference speeds exceeding forty-five frames per second with competitive mean Average Precision (mAP), but suffers from elevated false positive rates in low-contrast settings. In contrast, Faster R-CNN attains approximately a 1.5% higher mAP and lower misclassification rates under heavy occlusion, albeit with reduced inference throughput. We conduct ablation studies on augmentation strategies, anchor box configurations, and backbone network depths, quantifying their impacts on precision, recall, and mAP. We release DLP-14k alongside standardized training and evaluation protocols and baseline code to foster reproducibility. Models are trained using an 80:10:10 split for training, validation, and testing. Baseline code includes example training and inference scripts. These resources underpin reproducible research and facilitate rapid prototyping of precision agriculture solutions such as autonomous harvesting robots and drone-based monitoring systems. Future work includes extending the dataset to additional fruit species and environmental conditions, exploring hybrid attention-based architectures for improved discrimination, and integrating segmentation masks. Our contributions provide the computer vision and precision agriculture communities with a robust resource and benchmark results for advancing camouflaged object detection in natural environments.
#20 Development of Organizational Community Extension Tracking and Approval System Using Microsoft as a Low-Code Development Platform
Jan Guiller U. Vergara, Christian Aldwin D. Canlapan, Gian Mark T. Pulgar, Jenelyn J. Salimbagat, Ron Andrei E. Soriano, Steffi Gabrielle O. Dillague, John Jev C. Sablaya and Hameil A. Renabor (National University, Philippines)
Community extension focuses on faculty members' transformative role in societal needs and stakeholder collaborations; however, it relies heavily on face-to-face interaction. Ensuring continuous and comprehensive training for faculty members in higher education institutions (HEIs) as extension workers or community development facilitators is crucial. However, the absence of a precise monitoring and evaluation system challenges the faculty member's contribution to institutional attainment in community engagements. One such solution is to streamline the processing of community engagements and their approvals through automation, which uses the Low-Code Development Platforms (LCDPs). The tracking system was designed according to the developed swimlane process using Microsoft 365 and Power Automate, divided into a Recording and Approval System and a Monitoring System. The system was divided into two flows: NUArCE and NUArCE Go, which were used for both recording with approvals and monitoring. Results showed the system offers a short processing time for evaluating and crediting employee community engagement, with an average of 72.9 seconds for 10 runs for the NUArCE and 22 seconds for 3 runs for the NUArCE Go. Issues and challenges include storage, printable areas, file extension conversion, and bottlenecking due to long user action times. Nevertheless, the tracking system provided the desired output within seconds, ensuring real-time monitoring of faculty members' community engagement summary.
#21 CrOptimize: Integrated Agri-Tech Solution for Optimized Farm Lot Utilization and Resource Management
Hannah Aizel C Garcia, Gilbert P Mendoza, Jake Harold B Roxas and Vianca Dhenise D Vergara (FAITH Colleges, Philippines); Adonis Santos (First Asia Institute of Technology and Humanities, Philippines)
CrOptimize is an integrated agri-tech solution designed to optimize farm lot utilization and resource management by combining geo-tagging, drone surveillance, and soil nutrient monitoring. The system aims to improve agricultural efficiency and sustainability, especially in regions like the Philippines, where food security is a growing concern. Geo-tagging enables accurate measurement of farm areas, ensuring precise seed allocation and minimizing surplus or shortages. Soil nutrient sensors analyze soil composition and guide farmers in selecting suitable crops and planting strategies, enhancing yield potential. Drone surveillance validates seed distribution and continuously monitors crop growth, promoting transparency and accountability in farming practices. This data-driven approach reduces inefficiencies and resource wastage while supporting informed decision-making among farmers and agricultural suppliers. Suppliers also benefit from enhanced supply chain accuracy, enabling better distribution planning and inventory management. Although challenges such as GPS signal limitations, sensor calibration, and drone maintenance may impact deployment, CrOptimize offers a scalable and adaptable framework for modernizing agriculture. By integrating advanced technologies, the system empowers stakeholders to improve productivity and align with global efforts toward sustainable and resilient food systems.
#22 WELLNEST: Fostering Student Well-Being in Academic Settings
Francine Alanysse Amigo, Anjelie Carpio and Vernon Jhon Quilang (National University, Philippines); Alyssa C Vicente (National University Philippines, Philippines); Charlyn A. Malimata (National University, Philippines)
The "WellNest" system is a web and mobile-based platform designed to enhance mental health services in academic settings. It streamlines the core operations of the Guidance Services Office (GSO), automating tasks such as appointment scheduling, user management, and counseling form processing. Developed using an agile approach and an Input-Process-Output (IPO) framework, "WellNest" offers students and employees flexible, confidential access to mental health support. The platform provides tailored features for different user roles, ensuring that administrators, counselors, and users can efficiently access the services and resources relevant to their needs. By integrating digital tools, the system reduces administrative burdens, improves service accessibility, and fosters a more supportive environment for addressing mental health challenges within the university community. Additionally, "WellNest" takes a holistic approach to mental health management, empowering academic institutions to proactively support the well-being of their members and cultivate a healthier, more responsive mental health ecosystem in the academic setting.
#23 Anti-Poaching System Using Wireless Communications and Image Processing for Wildlife Sanctuaries
Irvin Johnson Zari, John Christopher Delos Santos, John Philip Alcala, Alain Bernard Rañola, Bernadeth B Zari, Melannie B. Mendoza and Edwin L. Astorga (Adamson University, Philippines)
Poaching in the Philippines constitutes a critical problem that endangers species survival and inflicts detrimental environmental consequences. Despite the government's efforts to tackle the issue through enhanced legislation and enforcement, the challenge remains. Biak-na-Bato National Park is a protected area in the country impacted by poaching. The objective of the study is to develop a device that combines PIR sensors and cameras to detect human intrusions and alert local authorities. The collected data will be transmitted wirelessly through internet connectivity for distant observation and analysis. A web server from the ESP32 CAM Module was employed to surveil the remote area, and its incorporation into the PyCharm programming environment facilitated image analysis for the identification of unauthorized individuals. Consequently, real-time alerts and alarms were generated simultaneously, allowing users to make informed decisions regarding the need for on-site involvement. These applications served as the primary interface for assessing the security of the remote area safeguarded by the prototypes and the data gathered from the system, utilizing Arduino and Python programming languages. The system's accuracy was assessed utilizing three cameras and ten trials across six variables, yielding results ranging from 83.33 percent to 100 percent. The data transfer rate attained around 9.6 Mbps, with a minor fluctuation of 3.03 percent in device uptime. The technology exhibited absolute reliability in identifying individuals. Moreover, all real-time testing data was systematically archived within the system, meeting the design specifications and providing a feasible solution for combating poaching activities. The results validate the system's dependability in incursion detection, data transfer, and remote monitoring, offering a cost-efficient and scalable alternative to bolster conservation initiatives in protected regions. Their testing confirmed the system's effectiveness in detecting human intrusion, providing data rapidly, notifying authorities through audible alarms, and employing many integrated devices for reliable data transmission.
#24 Design of Dual Band Antenna Using CMA for IEEE 802.11a and Wi-Fi Applications
Sai Debasisa Patra and Sambhudutta Nanda (VIT AP University, India)
This paper presents a dual-narrow-band antenna design at 5.06/5.48GHz is proposed. It has a circular patch at the top layer that is loaded with an SRR-shaped slot. Dual modes are excited to achieve the dual-band, and the antenna's modal behavior is examined using characteristics mode analysis. Based on the modal behavior of the proposed antenna using a characteristic mode analysis (CMA), an antenna operating at dual resonant modes at 4.98GHz and 5.95GHz, respectively is designed for the dual-band operation. The proposed antenna is printed on a low cost FR4 substrate with a size of 25×33×1.6mm3. The full-wave simulation method is used to excite dual modes using a 50Ω coaxial feedline. The simulated reflection coefficient S11≤ 10dB and the maximum gain achieved with peak gain 6.17dBi and 4.13dBi by improving the antenna settings. The bandwidth covers IEEE 802.11a 5G sub-6GHz communication, which corresponds to 5.04GHz and 5.48GHz bands, as well as Wi-Fi (indoor Wi-Fi).
#25 Integrating DWVD Dimensionality Reduction and Bio-Inspired Feature Selection for Accurate Lung Cancer Prediction
Karthika M S (Vellore Institute of Technology, India); Harikumar Rajaguru (Bannari Amman Institute of Technology & ECE, India); Ajin R Nair (Manipal Institute of Technology Bengaluru, MAHE, India)
Early and accurate detection of lung cancer is crucial to designing effective treatments and contributes strongly to increasing survival rates among patients. The study uses a mixed-methods approach for the classification of lung cancer using microarray gene expression data. The issue of high-dimensional data is handled by Discrete Wigner-Ville Distribution (DWVD)-based dimensionality reduction without losing important spectral and temporal gene features. To further optimise the dataset, Optimal Feature Selection is carried out using the Harmonic Search (HS) and Cuckoo Search (CS) algorithms. The reduced and cleaned dataset obtained is then classified using a variety of machine learning algorithms such as Non-Linear Regression (NLR), Softmax Discriminant Classifier (SDC), Gaussian Mixture Model (GMM), Random Forest (RF), and Support Vector Machine with Radial Basis Function kernel (SVM-RBF). Performance is strictly analyzed with different metrics including accuracy, F1 Score, Jaccard Index, Matthews Correlation Coefficient (MCC), and Cohen's Kappa. Out of all the classifiers, the SVM-RBF model produced the highest potential results with accuracy of 98.34%, F1 Score of 98.99%, Jaccard Index of 98.01%, MCC of 0.9425, and Kappa of 0.9423 after HS-based feature selection. These results demonstrate the power of integrating novel signal processing, bio-inspired optimization, and strong classifiers for accurate lung cancer prediction from gene expression data
#26 TerraMori: IoT-Based Autonomous Terrarium for Optimized Growth of Moringa Oleifera
Gregory Hans B. Abundabar, Gen-Ichie Balatbat, Ma. Sopia Alyanna D. Garcia, Mark Angelo C Purio, Joan D. Sta Ana and Melannie B. Mendoza (Adamson University, Philippines)
This paper presents TerraMori, an IoT-enabled sealed terrarium designed for automated cultivation of Moringa oleifera. The system integrates embedded sensors for temperature, humidity, light, CO₂, O₂, and soil moisture, and utilizes ESP32-based control with the Blynk IoT platform for real-time monitoring and environmental regulation. Over a four-week evaluation, TerraMori consistently maintained environmental parameters within target ranges, supporting steady plant development. Experimental results are compared to growth and leaf color between TerraMori and outdoor plants. Although Welch's t-test results indicated no statistically significant differences, large effect sizes (Cohen's d = 1.046 for Plant 2 height; d > 0.99 for leaf color) were observed, suggesting potentially meaningful differences favoring the controlled environment. Confidence intervals included zero, warranting cautious interpretation. Nevertheless, the system operated with 91.67% uptime and demonstrated reliable responsiveness to environmental changes. TerraMori shows potential as a sustainable cultivation solution for urban and resource-limited environments and merits further validation through expanded sample sizes, longer deployment, and objective instrumentation such as SPAD meters.