2025. 2nd Issue
Volume XVII, Number 2
Table of contents 
Full issue
PAPERS FROM OPEN CALL
Ádám Salamon, Gábor Takács and György Bognár
Validation Methodology of Wireless Brain-Computer Interface for Event-Related Potential Application
Electroencephalography (EEG) is a technique used to observe brain activity by measuring the dynamic changes of the electric field induced by neurons' activity. Brain-computer interface (BCI) systems are used in cognitive psychology examinations measuring the changes of brain activities. This paper presents a validation methodology to characterize BCI systems with wireless communication interface and the applicability on a preselected BCI system. This way, the delay, the functionality, and the frequency selectivity can be determined of the overall BCI system, taking into account the effect of the hardware, the software, and the electrodes, avoiding noise artifacts. The presented and validated BCI system proved to be successfully applied in ERP EEG measurements such as steady-state visually evoked potential, pattern-reversal visually evoked potential, and P300 event-related potential.
DOI: 10.36244/ICJ.2025.2.1
Download
Qingyu Liu, Lei Chen, Yeguo Sun, and Lei Chen
Masked Face Image Inpainting Based on Generative Adversarial Network
Face image inpainting is a critical task in computer vision due to the intricate semantic and textural features of facial structures. While existing deep learning-based methods have achieved some progress, they often produce blurred or artifactprone results when handling large occlusions, such as face masks. To address these challenges, this paper proposes a novel generative adversarial network (GAN) framework tailored for masked face inpainting. The generator adopts a U-Net architecture enhanced with a multi-scale mixed-attention residual module (MMRM), which integrates multi-branch convolutions for diverse receptive fields and combines spatial-channel attention mechanisms to prioritize semantically relevant features. The decoder further enhances feature fusion through channel attention mechanism, which selectively emphasizes meaningful patterns during feature map reconstruction. A realistic masked face dataset is synthesized using the CelebA database by dynamically adjusting mask positions, sizes, and angles based on facial landmarks, ensuring alignment with real-world scenarios. Quantitative and qualitative evaluations demonstrate that our method outperforms conventional models in both visual quality and quantitative metrics. Ablation studies further validate the effectiveness of MMRM and attention mechanisms in preserving structural coherence and reducing artifacts.
DOI: 10.36244/ICJ.2025.2.2
Download
Bogdán Asztalos, Péter Bányász
Monitoring the Semantic Change of COVID-19-related Expressions Using Dynamic Word Embeddings
In this paper, we investigate how the COVID-19 pandemic has affected the use of language in the online space through measuring the semantic changes of words during the time that includes the outbreak of the pandemic and the months of the lockdown. As a first step, we apply a recent word embedding technique on a time-labelled text corpus collected from social media which represents the semantic relation of words based on their likelihood of co-occurring next to each other. By analyzing different statistical features of the received dynamic embedding, we can identify and quantitatively describe periods where the semantic properties of a chosen word are undergoing significant changes. Since this depends on the context and the usage of these words by the users, we can infer their reaction to the COVID-19-related events and relevant news dated to these periods.
DOI: 10.36244/ICJ.2025.2.3
Download
Máté Gedeon
A Comparative Analysis of Static Word Embeddings for Hungarian
This paper presents a comprehensive analysis of various static word embeddings for the Hungarian language, including traditional models such as Word2Vec, FastText, as well as static embeddings derived from BERT-based models using different extraction methods. We evaluate these embeddings on both intrinsic and extrinsic tasks to provide a holistic view of their performance. For intrinsic evaluation, we employ a word analogy task, which assesses the embeddings’ ability to capture semantic and syntactic relationships. Our results indicate that traditional static embeddings, particularly FastText, excel in this task, achieving high accuracy and mean reciprocal rank (MRR) scores. Among the BERT-based models, the X2Static method for extracting static embeddings demonstrates superior performance compared to decontextualized and aggregate methods, approaching the effectiveness of traditional static embeddings. For extrinsic evaluation, we utilize a bidirectional LSTM model to perform Named Entity Recognition (NER) and Part-ofSpeech (POS) tagging tasks. The results reveal that embeddings derived from dynamic models, especially those extracted using the X2Static method, outperform purely static embeddings. Notably, ELMo embeddings achieve the highest accuracy in both NER and POS tagging tasks, underscoring the benefits of contextualized representations even when used in a static form. Our findings highlight the continued relevance of static word embeddings in NLP applications and the potential of advanced extraction methods to enhance the utility of BERT-based models. This piece of research contributes to the understanding of embedding performance in the Hungarian language and provides valuable insights for future developments in the field. The training scripts, evaluation codes, restricted vocabulary, and extracted embeddings will be made publicly available to support further research and reproducibility.
DOI: 10.36244/ICJ.2025.2.4
Download
Melinda Kosák, Gábor Lencse
IP Packet Forwarding Performance Comparison of the FD.io VPP and the Linux Kernel
There are numerous free software solutions for IPv4 or IPv6 packet forwarding. The Fast Data Project / Vector Packet Processing (FD.io VPP) is a novel and prominent solution. This paper investigates its performance and scalability compared to that of the Linux kernel. The investigation was conducted in accordance with the requirements outlined in the relevant Request for Comments (RFC) documents (RFC 2544, RFC 4814, and RFC 5180) using the siitperf measurement software. Two different test environments were used to eliminate the potential hardwarespecific side effects and to gain insight into the performance and scalability of the IPv4 and IPv6 packet forwarding capability of the two investigated solutions. It was found that FD.io VPP outperformed the Linux kernel by approximately an order of magnitude. The configuration of FD.io VPP, along with the details of the measurements, are provided, and the results are presented and analyzed in the paper.
DOI: 10.36244/ICJ.2025.2.5
Download
Balázs Kreith and Árpád Drozdy
Multipath Rate Control for Real-Time Media
In this paper, we present rate control algorithms designed for real-time media transmission over multiple paths. Our focus is on delivering media content simultaneously across multiple paths while maximizing the network traffic utilization on each path. The proposed algorithms ensure reduced network delay fluctuations, more consistent transmission rates for media content, and fairness to cross-traffic across all paths. To evaluate their performance, the solution is implemented and tested in an emulated networking environment under various test scenarios. The results demonstrate that applying our algorithms leads to reduced network delay fluctuations, improved structural similarity of the received media content, and enhanced fairness toward cross-traffic.
DOI: 10.36244/ICJ.2025.2.6
Download
David Balla, Markosz Maliosz, Csaba Simon
Completion Time Prediction of Open Source FaaS Functions
Function as a Service (FaaS) is the latest stage of application virtualization in the cloud. It enables to deploy small code pieces – functions – in the cloud. FaaS focuses on event-driven functions in response to triggers from different sources. The functions run in ephemeral virtual environments. This means that the user is charged on the basis of the time the function is busy serving the invocation requests. With the advent of Industry 4.0 the need has arisen to run applications on Edge Computing nodes. FaaS is a promising solution for serving industrial applications that require predictable latency while meeting the demands of edge computing, which operates on a limited resource base. Therefore, knowing the completion time of the invocation requests is of key importance. In this paper, we introduce a function runtime design for opensource FaaS implementations that achieves a lower deviation in request completion times compared to default runtimes by regulating the function’s access to host CPU cores. We present the implementation details of our proposed function runtime design for Python, Go and Node.js. We also introduce a simulation framework that is able to estimate the completion time distribution of the incoming invocation requests. We validate the results of our simulation framework using real measurement data.
DOI: 10.36244/ICJ.2025.2.7
Download
XY
xy
xy
DOI: 10.36244/ICJ.2025.2.8
Download
Nitin Jha, Abhishek Parakh, Mahadevan Subramaniam
Multi-photon QKD for Practical Quantum Networks
Quantum key distribution (QKD) will most likely be an integral part of any practical quantum network in the future. However, not all QKD protocols can be used in today’s networks because of the lack of single-photon emitters and noisy intermediate quantum hardware. Attenuated-photon transmission, typically used to simulate single-photon emitters, severely limits the achievable transmission distances and makes the integration of the QKD into existing classical networks, that use tens of thousands of photons per bit of transmission, difficult. Furthermore, it has been found that protocol performance varies with topology. In order to remove the reliance of QKD on single-photon emitters and increase transmission distances, it is worthwhile to explore QKD protocols that do not rely on single-photon transmissions for security, such as the 3-stage QKD protocol, which can tolerate multiple photons in each burst without information leakage. This paper compares and contrasts the 3-stage QKD protocol with conventional QKD protocols and its efficiency in different network topologies and conditions. Furthermore, we establish a mathematical relationship between achievable key rates to increase transmission distances in various topologies.
DOI: 10.36244/ICJ.2025.2.9
Download
Ahmed S. Mohamed, Eszter Udvary
Mode Selection in Mode Division Multiple Access System for In Building Solution in Mobile Networks
This paper introduces the application of Mode Division Multiple Access (MDMA) in the context of In-Building Solutions (IBS) for mobile networks. The study showcases the successful generation and selection of light modes which are then efficiently multiplexed and demultiplexed at the Remote Radio Unit (RRU) end. Despite the proven operational capabilities, the findings reveal a decline in signal quality as the distance increases, thus limiting the use of MDMA for longdistance fronthaul applications. The proposed system also simplifies the RRU by centralizing key functionalities at the Central Office (CO), potentially reducing costs and the operational expenses (OPEX/CAPEX) associated with inbuilding solutions and other mobile network deployments. This work extends previous research and paves the way for future studies, particularly in the application of the Power over Fiber (PWoF) approach to reduce RRU complexities further.
DOI: 10.36244/ICJ.2025.2.10
Download
Ali Ismael Anwer, Zainab S. Muqdad, Taha A. Elwi, Zaid Asaad Abdul Hassain
A New Compact Metamaterial Microwave Sensor Design for Nondestructive Biomedical Measurements
In this work, serval sensor designs are proposed based on developing a basic sensor structure based on a lowcost, highly-sensitive microwave sensor for identifying different liquid samples through monitoring the variation in S21 magnitude. At beginning, this is attempted by applying the use of an interdigital capacitor (IDC) in series connection with a circular spiral inductor (CSI) and connected directly to a photoresistor (LDR). To enhance the sensor insertion losses with a high-quality factor, the proposed sensor is introduced to a Hilbert fractal open stub that is coupled to an interdigital capacitor to operate at 1.22 GHz. The proposed sensor insertion matching is enhanced by using two fractal stubs as matching circuits. The accuracy of the proposed sensor is significantly improved using a back loop trace. Therefore, with such technique, the nonlinear effects due to the multi-layers diffractions are eliminated. Such novel process linearized the sensor response variation with respect to the material under test introduction. An analytical model based on circuit theory is suggested to realize the proposed sensor operation. It is found an observable influence of varying the LDR value on the sensor insertion losses. Such an observation motivated the authors to realize the proposed sensor prototype. The proposed sensor is manufactured and tested experimentally before and after the samples introduction. Therefore, during the measurements, a human urine is mounted on the LDR patch through a glass slicer platform to measure the effects varying the urine properties with fixing the light intensity. Finally, the authors found a readable variation that could be an excellent candidate for new generation of urine microwave sensors.
DOI: 10.36244/ICJ.2025.2.11
Download
Muhammad Ahsan Shaikh, Tayyab Ahmed Shaikh, Sadiq ur Rehman, and Halar Mustafa
Constrained LS Channel Estimation for Massive MIMO Communication Systems
In recent years, the manufacturing of mobile and IoT devices has increased dramatically. For the service provider, the requirement for high throughput and extensive connectivity became a major obstacle. In B5G and 6G, different advanced technologies have been introduced to cater demands of users effectively. One of the most important technologies of nextgeneration networks is massive MIMO systems. In multiuser communication systems, transmission and reception of signals occur simultaneously which creates multiuser interference (MUI). The presence of MUI in the system is the major challenge for the effective operation of massive MIMO receivers. The influence of MUI must be minimized using a channel estimation technique in order to fully utilize the capabilities of a massive MIMO system. This work proposes the constrained least square (LS) channel estimate technique to improve the massive MIMO downlink system's overall performance. The Mean Square Error shows that the unconstrained LS performance is poor as compared to the constrained LS channel estimation. Additionally, the effectiveness of the proposed constraint LS channel estimate is assessed in communication systems using varying transmission antennas at the base station and number of users.
DOI: 10.36244/ICJ.2025.2.12
Download