Hue University Journal of Science: Techniques and Technology

Design and manufacturing of optical multimode interference device based on a silicon photonic integrated circuit

Thanh Chuong Dang — 2025-12-31

Photonic chips will gradually replace electronic chips in several specialized fields due to their advantages, including broadband, high speed, and compatibility with current electronic chip manufacturing technologies. Researching and selecting typical essential components to build photonic chips is extremely important. The multimode interference component based on silicon material is undergoing a thorough investigation and standardization process to be added to the industrial production library. In this paper, we propose a new property of the multimode interferer MMI 4x4, which simultaneously transmits two signals in the same phase or opposite phase at the inputs to perform switching in the same direction or opposite direction at the output on the same silicon photonic integrated circuit. This result will extend to developing optical signal switchers, routers, splitters, or couplers for more complex functional circuits. We have achieved similar results between theoretical calculation, simulation, and manufacturing.

Synthesis of lemongrass essential oil–chitosan nanoemulsions via ultrasonication with sodium tripolyphosphate and evaluation of antifungal activity

Võ Văn Quốc Bảo — 2025-12-31

This study presents the synthesis of various nanoemulsion systems using lemongrass essential oil (Cymbopogon spp.) and chitosan, employing ultrasound in conjunction with sodium tripolyphosphate (TPP) to enhance stability and antifungal properties. Nanoemulsion samples, both with and without TPP supplementation, were investigated for their nanosystem properties, stability over a 90-day storage period, and antifungal efficacy. Results indicate that the sample without TPP supplementation (nano TDS-Chi3) and the samples with TPP supplementation (nano TDS-Chi-TPP3 and nano TDS-Chi-TPP4) achieved droplet sizes below 200 nm and polydispersity index (pDi) values below 0.3, reflecting high homogeneity in particle structure. Notably, the sample nano TDS-Chi-TPP₃ demonstrated exceptional antifungal activity, achieving 100% inhibition against Colletotrichum musae QB6, the pathogen causing anthracnose in post-harvest bananas. The incorporation of TPP formed a stable cross-linking network among the chitosan particles, maintaining biological activity and preventing phase separation over time; specifically, the nano TDS-Chi-TPP3 sample retained its nanoemulsion characteristics after 90 days of storage. These findings highlight the potential for developing sustainable and safe biological solutions in agriculture, aimed at improving product quality and environmental protection.

Determination of some parameters of biomass fuel pellets from water hyacinth

Công Anh Võ — 2025-12-31

This study presents the results of an investigation into the basic parameters of water hyacinth pellets, including bulk density, mechanical strength, total moisture content, ash content, volatile matter content, fixed carbon content, and calorific value. Thermogravimetric Analysis (TGA) was employed to examine the combustion characteristics, while Scanning Electron Microscopy (SEM) was used to analyze the morphological structure of the water hyacinth pellets. The influence of two important parameters-compression pressure and raw material particle size-on the bulk density of water hyacinth pellets was also evaluated. The results indicated that most of the parameters of water hyacinth pellets meet the requirements of the TCVN 13534:2022 standard for fuel pellets. An optimal compression pressure of approximately 150 MPa and raw material size smaller than 2 mm were found to be suitable for the pelletizing process. The TG and DTG analysis revealed that water hyacinth pellets have promising potential as an efficient biofuel with rapid thermal decomposition characteristics. These results provided a strong basis for the viability of using water hyacinth as a raw material for energy pellet production.

GPSR-RA: A Cross-Layer Position-Based Routing Protocol with Dynamic Bitrate Adaptation for FANETs

Cường Thọ Mai — 2025-12-31

. In FANETs, the rapid fluctuations in link quality, caused by continuous changes in UAV distances, interference, environmental conditions, transmission power, and multipath effects, significantly impact data transmission efficiency. This paper proposes the GPSR-RA protocol. This paper proposes the GPSR-RA (GPSR with BitRate Adaptation) ), an enhancement of the GPSR protocol, designed to adjust the bitrate during data transmission based on link quality. GPSR-RA dynamically adjusts the bitrate by utilizing the SNIR (Signal-to-Noise-plus-Interference Ratio) parameter to select an optimal bitrate, thereby improving packet delivery ratio, reducing latency, and maximizing bandwidth utilization. GPSR-RA retains the routing mechanism of GPSR but incorporates a dynamic bitrate adjustment mechanism based on SNIR and the adjustment thresholds for increasing or decreasing at each bitrate. GPSR-RA collects SNIR information from the physical layer via Hello packets and stores it in the neighbor table. The SNIR is then passed this information to the MAC layer immediately after the next-hop node is determined. The bitrate adjustment module (MACSnir-RA) in GPSR-RA actively increases or decreases the bitrate according to the SNIR value, enabling the system to respond promptly to changes in the link quality, and thus improve network performance. Simulation results on the OMNeT++ platform demonstrate that GPSR-RA significantly enhances packet delivery ratio, increases network throughput, and reduces end to end delay compared to GPSR.

An approach for mining fuzzy association rules based on hedge algebra

Nguyễn Công Hào — 2025-12-31

Fuzzy association rules based on fuzzy set theory have been researched by many authors and have published many significant results and flexibility in data mining with uncertain information. However, the approach using fuzzy set theory for the problem of mining fuzzy association rules still has certain limitations. In this paper, we propose a method for mining fuzzy association rules based on hedge algebra with each linguistic value represented by a their neighborhood. The hedge algebra has advantage of using the measure functions and semantic quantifier functions, the problem of mining fuzzy association rules and calculating them is quite simply and intuitively. The results obtained after extracting fuzzy association rules on the survey dataset of freshman in choosing a university are a channel providing useful information for university managers in enrollment communication.

Synthesis and evaluation of the efficiency of a fly ash-derived coagulant for industrial wastewater pre-treatment

Văn Toàn Phạm — 2025-12-31

The development of new materials for environmental treatment and the reuse of industrial waste have been receiving attention, with the aim of promoting sustainable development. In this study, fly ash from a thermal power plant was utilized to synthesize a coagulant. The synthesized coagulant was applied in the pre-treatment of centralized industrial wastewater using the coagulation-flocculation process. The optimal dosage of coagulant was determined through Jartests in the laboratory. Then, coagulation-flocculation experiments were conducted using a laboratory-scale model, comparing the fly ash-derived coagulant with commercially available polyaluminum chloride (PAC). Jartest results demonstrated that the fly ash-based coagulant achieved removal efficiencies of 85.94% for turbidity and 90.49% for suspended solids (SS) at a dosage of 7.0 mg/L and pH 8.0. Model-scale experiments showed that the fly ash-based coagulant at the selected dosage, and PAC at 500 mg/L, achieved removal efficiencies of 61.8% and 39.08% for turbidity, 52.67% and 51.42% for SS, and 57.79% and 23.53% for COD, respectively. These results indicate that fly ash from thermal power plants can be effectively reused as a coagulant, offering a promising solution for wastewater pre-treatment and contributing to the reduction of environmental impacts from coal-fired power plant waste.

Applying computer vision to train unmanned aerial vehicles for recognizing basic geometric objects

Hữu Trung Hoàng — 2025-12-31

This study proposes a generalized methodology for training unmanned aerial vehicles (UAVs) to autonomously recognize objects of simple and common geometries in real-world environments. Specifically, a Convolutional Neural Network (CNN) model is integrated into the camera system of a DJI Tello drone to detect basic geometric objects, including rectangles, triangles, circles, and regular pentagons, which are selected to evaluate the model's recognition performance. A grayscale image dataset comprising objects of varying sizes and positions is automatically generated to optimize the data collection and model training process. The proposed CNN model, designed with a lightweight architecture to ensure real-time processing capability on the drone, is trained on this dataset and achieves approximately 100% accuracy on both the training and test sets. Subsequently, the model is integrated into the drone's camera system, and experimental results confirm its ability to perform accurate real-time object detection without overfitting. These findings demonstrate the effectiveness and practical potential of the proposed method for integration into intelligent drone systems.

Fabrication of bismuth oxyiodide (BIoI) photoelectrodes via solvothermal method for photoelectrochemical applications

Nghia Van Nguyen — 2025-12-31

Bismuth oxyiodide (BiOI) thin films were successfully synthesized from Bi(NO₃)₃ and KI precursors via a solvothermal method in an ethylene glycol ((CH₂OH)₂) medium. In this study, the morphology and thickness of BiOI nanosheets were uniformly formed on conductive FTO substrates at different temperatures. X-ray diffraction (XRD) and field-emission scanning electron microscopy (FESEM) were employed to investigate the crystal structure and surface morphology of the BiOI films. UV–vis–NIR absorption spectra were measured using spectroscopy in the wavelength range of 300–700 nm to clarify the optoelectronic properties of the BiOI photoanode. The photoelectrochemical characteristics of the electrode were studied through Mott-Schottky (M-S) analysis and linear sweep voltammetry (LSV) using a three-electrode electrochemical system. The electrode prepared with a 30 mM Bi(NO₃)₃ precursor concentration exhibited a maximum photocurrent density of 0.35 mA·cm⁻², demonstrating the material's potential for photoelectrochemical applications.

ERDQ-learning: reinforcement learning model integrating hpa for scaling serverless pods in Kubernetes

Thanh Chuong Dang — 2025-12-31

In serverless cloud computing environments, maintaining performance and Quality of Service (QoS) under fluctuating traffic conditions remains a significant challenge. The traditional Horizontal Pod Autoscaler (HPA) in Kubernetes, while fundamental, relies solely on fixed CPU utilization thresholds to scale Pods, often resulting in delayed responses or inefficient orchestration in dynamic load scenarios. This paper proposes an enhanced reinforcement learning approach, ERDQ-learning, which integrates Experience Replay and Double Q-learning to optimize the scaling behavior of serverless Pods. The ERDQ-learning model dynamically adjusts HPA’s CPU activation threshold in real time, using system state parameters such as CPU utilization, current Pod count, response latency, and the existing threshold. Experimental evaluations in a simulated Kubernetes environment with varying traffic patterns demonstrate that ERDQ-learning significantly enhances system adaptability, reduces latency, and improves resource efficiency compared to the traditional HPA. These results highlight the feasibility and effectiveness of the proposed model for intelligent resource orchestration in modern serverless systems.

A One-Dimensional Convolutional Neural Network-Based Equalizer for 100 Gb/s Short-Reach Optical Communication Systems

Phước Vương Quang — 2025-12-31

Short-reach optical communication systems that use intensity modulation and direct detection (IM/DD) with PAM-4 signaling are considered as a desirable option due to its simple architecture and lower cost. However, bandwidth limitations, chromatic dispersion and device nonlinearities cause serious signal distortions, which degrade overall system performance. In this paper, we present a one-dimensional convolutional neural network (1D-CNN)-based equalizer to address the challenges of traditional feed-forward equalizers (FFE), which are fundamentally limited to linear impairment compensation. The proposed equalizer is designed with a lightweight CNN architecture that maintains structural simplicity while effectively exploiting temporal features from the received symbol sequences. Simulation results demonstrate that the 1D-CNN equalizer can improve receiver sensitivity by nearly 2.5 dB compared to the conventional FFE and by around 1 dB compared to the artificial neural network (ANN)-based equalizer, at the same BER level. Furthermore, we investigate the impact of network depth and feature map size on equalization performance, providing practical insights for real-world deployment.

Optimizing Protein-Protein Interaction Prediction from Language Representations via Multi-stage Feature Selection and Stacking Ensemble Learning

Xuân Văn Mai — 2025-12-31

Protein–protein interactions (PPIs) form the foundation of many intracellular biological processes, and predicting PPIs directly from amino acid sequences remains a core direction in computational biology. The advent of next-generation Protein Language Models (PLMs), such as ESM-2, enables the generation of sequence representations rich in evolutionary information and latent structural signals. However, these representations often possess extremely high dimensionality, contain significant noise, and exhibit high internal correlation, making it difficult for traditional machine learning models to exploit them effectively and increasing the risk of overfitting. This challenge demands an approach capable of distilling knowledge and eliminating data redundancy while preserving core biological signals. In this work, we propose E–StackPPI (Embedding-Stacking Protein-Protein Interaction prediction framework), a fully embedding-based PPI prediction framework centered on a three-stage layer-wise feature selection mechanism applied directly to embeddings aggregated from the last hidden layers of the ESM-2 650M model. Specifically, the process sequentially: (1) removes dimensions with low variance; (2) retains highly discriminative features based on LightGBM feature importance; and (3) eliminates dimensions with high Pearson correlation to reduce information redundancy. The refined feature set is fed into a stacking architecture, where two parallel LightGBM branches are integrated at the decision layer via Logistic Regression (LR). Experiments on two benchmark datasets from the Database of Interacting Proteins (DIP) [1], including DIP–Yeast and DIP–Human, show that E–StackPPI achieves favorable and stable results across key metrics, including accuracy, MCC, as well as ROC-AUC and PR-AUC indices. When benchmarked against twelve advanced methods summarized in the study by Li et al. [2], our model demonstrates competitive performance on both datasets. These findings highlight the essential role of layer-wise feature selection in mitigating noise and effectively leveraging high-dimensional PLM embeddings, thereby opening a feasible and promising sequence–only approach to PPI prediction without the need for supplementary structural data. Protein-protein interaction; Multi-stage feature selection; Protein Language Models; Stacking model.

Evaluation of knowledge-based query tree protocols with different rfid tag distributions

Đức Nhật Quang Nguyễn — 2025-12-31

Tag query in RFID systems is a key operation for identifying tags that appear in the reader’s interrogation zone. Many tag query protocols have been proposed; however, reducing the number of collision slots often comes at the cost of increasing the number of idle slots. In certain deployment scenarios, the number of tags, the tag ID space, and the tag distribution can be known in advance; this information constitutes knowledge and has been exploited in several knowledge-based query algorithms. This paper evaluates knowledge-based query tree protocols under different tag distributions: uniform and non-uniform. Implementation results show that knowledge-based query tree protocols consistently achieve better query efficiency than the traditional query tree protocol, with QTKS and QTKS_TE providing the best performance. The analytical and implementation results also indicate that the more knowledge is exploited in the protocols, the higher the achievable query efficiency.