Ebook: Electronics, Communications and Networks

The Conference on Electronics, Communications and Networks (CECNet) is an annual conference devoted to electronics technology, communications engineering, wireless communications, and computer engineering. CECNet 2023 returned to a face-to-face format and was held in Macau, China from 17–20 November 2023. The CECNet conference series has completed its thirteenth edition, and hopefully after the last three years in online mode, the face-to-face conference has now returned for good.

The contributions can be classified under three main categories:

– Electronics Technology and VLSI

– Internet Technology and Signal Processing

– Information Communication and Communication Networks.

The attendees enjoyed three keynote speakers: Prof. Martin Maier from the Institut National de la Recherche Scientifique (INRS), Montréal, Canada, who has a h-index of 52 and about 10,000 citations; Prof. Gi-Chul Yang from Mokpo National University, Mokpo, South Korea, who has more than 30 years experience in academia and who co-authored “A Superior Algorithm for Mutual Exclusion in Computer Networks – VISHNU”, IFIP Congress 1989: 169–174 with Vijay Kumar and Jerry Place, a work constituting an important pillar in the synchronization field of computer science and a root of the research career of these three well-known scientists; and Prof. Chi-Man Pun, Computer and Information Science, University of Macau, China who has a h-index of 40.

All papers were exhaustively reviewed by program committee members and peer-reviewers, taking into account the breadth and depth of the research topics falling under the scope of CECNet. From about 324 submissions, the 101 most promising and FAIA mainstream-relevant contributions are included in this book. These present innovative, original ideas or results of general significance supported by clear and rigorous reasoning and compelling new light in the evidence as well as the methods.

We would like to thank all the keynote speakers and authors for their efforts in preparing their contributions to this leading international conference. Moreover, we are very grateful to all those, especially the program committee members and reviewers, who devoted time to the evaluation of the papers. It is a great honor to continue with the publication of these proceedings in the prestigious series Frontiers in Artificial Intelligence and Applications (FAIA) from IOS Press. Our particular thanks also go to FAIA series editors for once again supporting this conference.

Finally, we hope you enjoy your visit to Macau.

October 2023

Antonio J. Tallón-Ballesteros

Estefanía Cortés-Ancos

Diego A. López-García

Department of Electronics, Computer Systems and Automation Engineering, University of Huelva (Spain), Huelva city, Spain

The generalized Kapitsa problem of stabilizing the upper position of a deformable pendulum under the action of small vertical oscillations of its base in a gravity field is solved. The presence of a small parameter of the problem allows us to carry out averaging and obtain approximate equations of motion of the pendulum. Two models of a pendulum are considered and compared: a flexible inextensible rod and a flexible tensile rod. The influence of each parameter of the problem on stability is studied. The limits of applicability of the model of a flexible inextensible pendulum are obtained.

Due to the necessity of transfer, the outer door of the transportation hub’s transfer hall is often kept open, resulting in significant cold air infiltration during the winter season, leading to increased heating energy consumption. This research utilized the air velocity measurement method, and air volume balance method to investigate the extent of cold air infiltration and indoor thermal comfort in a transportation hub transfer hall in Beijing during winter. The measured results indicated that during the winter test period, air permeability accounted for 57.6% of heat loss. Therefore, the cold air infiltration of the transfer hall in winter should be focused on.

The collection of nuclear power plants operating data is the basis for subsequent fault diagnosis and obtaining the operating status of nuclear power plants, but equipment failures and external interference will lead to missing operating monitoring data, which will reduce the quality of the data and thus reduce the accuracy of the subsequent analysis results. To solve this problem, this paper utilizes the fact that nuclear power plants have accumulated a large amount of operational data and researches the method of generating adversarial imputation network (GAIN)–based imputation method for missing values of nuclear power plants’ operational data. The generator in the model estimates the missing values by learning the distribution of the true values, and the discriminator in the model discriminates which values are true and which are generated with the help of a hint matrix. The hint reveals partial information about the missing original samples to the discriminator, which the discriminator uses to focus its attention on the quality of the imputation of particular data values. Finally, a training set and a test set were constructed for comparative experiments on the PCTran simulation platform by simulating the operational data of the AP1000 as an example. The experimental results demonstrate that the investigated algorithm achieves lower root mean square error (RMSE), verifying the feasibility and accuracy of the method.

Aviation connectors play a vital role in complex electromechanical products. It is essential to have access to contact recognition within the connector model to facilitate the registration and guidance of intelligent aviation connector insertion operations. However, lightweight models often lack contact recognition, leading users to manually label the contacts, reducing efficiency and introducing errors easily. Therefore, this paper proposes lightweight model-based intelligent recognition of aviation connector’s contacts for AR guidance. The objective is to automate the contact recognition. Firstly, the hough circle transform is utilized for contact coarse positioning, followed by deep-learning using the positioned contacts as a dataset. Subsequently, the central contact’s centroid is determined, and contacts are categorized into concentric layers based on their distance from the central contact. Finally, the main dowel is positioned through corner detection, enabling sequential sorting of contacts within each layer based on the main dowel. Experimental validation demonstrates a 100% accuracy rate in contact recognition when utilizing this method, affirming the effectiveness of lightweight model-based intelligent recognition of aviation connector’s contacts.

The application of augmented reality (AR) technology in the assembly process of small cabins is currently a development trend. However, in the assembly guidance of the cabin cable network, the flexible characteristics of the cables make it a significant challenge for registration, and current technology makes it difficult to accurately overlay virtual cables into real scenarios. To address these issues, this article proposes an AR-oriented precise registration method for cabin cables. Firstly, precise registration of the model is achieved through coarse registration to determine the initial pose and subsequent pose correction. Subsequently, by analyzing the composition of flexible cables, the precise registration of cables is achieved through the assembly relationship between the cabin model and the flexible cables and the transformation of coordinate systems. Through this strategy, we have successfully achieved precise stacking of virtual, flexible cables in real scenarios. Finally, an experimental verification was conducted using a small missile cabin cable registration as a case study, and this method achieved accurate cable registration with a registration error of within 1.5 mm.

The outlet recognition for large-scale cable laying has problems such as difficulty obtaining prior information, similarity in local recognition information, and low recognition accuracy, making it difficult for most outlet recognition based on real prior information for large-scale cable laying to proceed smoothly. In response to the above issues, this paper proposes an outlet recognition method based on deep learning and sensors for large-scale cable laying. A system of outlet recognition based on deep learning and sensors for large-scale cable laying is formed based on the self-labeling of outlet images, data augmentation of outlet features, and relative pose constraints. The feasibility and effectiveness of this technology have been verified by recognizing outlets at different locations in a large-scale cable laying scene. This technology can recognize and locate outlets, effectively solving the problem of misrecognition at different positions.

A special binary representation/coding of an element in a free partially commutative monoid initially introduced in 1980 has been successfully used as a basis for the effective/polynomial solution of the following long standing open problems: functional equivalence of program schemata with non-degenerate operators, equivalence of deterministic multitape finite automata (MFA), equivalence of deterministic multidimensional multitape finite automata (MMFA), regular expressions for MFA, systems of equations for MFA regular sets. In addition, the consideration of the coding leads to an alternative characterization of commutation classes in free partially commutative monoids, which, in comparison with the already known characterization implying from the projection lemma, brings to a better efficiency when checking the equality of traces with lengths longer than a certain number or with alphabets containing more than two symbols. Regular expressions for languages of MFA and MMFA are also defined based on the mentioned coding. A brief overview of relevant AI applications concludes considerations.

The spring-mass system is a simple physics problem that has been widely discussed due to its importance and wide applications in different problems, particularly the waist-legs-feet system of a person could be represented by a spring-mass model, and the leap of a person can also be represented by the compression-release phenomena of the spring-mass system previously mentioned, this phenomenon is of great importance in different areas such as human health, and robotics. Thus, a model of damped spring-mass with a deformable load cell platform was used for getting a characteristic Green’s Function of a vertical jump of a person.

Maintenance plays a significant role in semiconductor manufacturing as plant yield, factory downtime and operation cost are all closely related to maintenance efficiency. Accordingly, maintenance strategies in semiconductor manufacturing industries are increasingly shifting from traditional preventive maintenance (PM) to more efficient predictive maintenance (PdM). PdM uses manufacturing process data to develop predictive models for remaining useful life (RUL) estimation of key equipment components. Traditional approaches to building predictive models for RUL estimation involve manual selection of features from manufacturing process data. This paper proposes to use deep convolutional neural networks (CNN) for the task of estimating RUL of lenses for an ion beam etch tool in semiconductor manufacturing. The proposed approach has the advantage of automatic feature extraction through the use of convolution and pool filters along the temporal dimension of the optical emission spectroscopy (OES) data from the endpoint detection system. Simulation studies demonstrate the feasibility and the effectiveness of the proposed approach.

The globalized world requires organizations to become increasingly competitive. To remain in the market, they are constantly seeking to reduce waste and enhance human capital. The goal of this study is to propose a implementation of a platform the helps in organizational knowledge searching, organization and availability to generate competitive advantage. The Moodle platform will be used as the methodology to capture the knowledge from servers, lectures, workshops, casual conversation classes and awards for the best ideas. The expected results will be the new ideas development, assertive decision-making and rework reduction by means of existing organization knowledge reuse.

Fresh produce supermarkets play a vital role in modern cities, but their management is challenging due to the perishable nature of vegetables. This research proposes and implements an automated pricing and replenishment strategy based on hybrid ML models and massive historical sales data. The combination of Seasonal ARIMA, Linear Regression, and Gradient Boosting Decision Tree (GBDT) results in an average R² for different categories of vegetable products of 0.993, indicating high model accuracy and fitting, which could guide pricing and replenishment strategies for the merchant who sells time-sensitive products.

The tremendous growth in technology is also the cause of global warming, which is harmful greenhouse gas Emissions. The information and communication technology (ICT) sector is one of the fastest-growing industries; it has the most significant impact on almost all other technologies. Energy efficiency and reduction of global warming is now a wish and realization by all key players related to this technology. The big data is one of the modern tools of ICT. It not only has a range of energy efficiency, but it can also help other departments to become intelligent energy efficient. This research article proposes several ways to achieve the mentioned goal, like optimization of battery size, overcoming uncertainties in energy generation, and overcoming storage and consumption. In a green data warehouse, the energy supply from renewable energy sources and the user traffic load are dynamic. The big data industry is also aware of the potential benefits of renewable energy to make the future more environmentally friendly and sustainable. The research progress of big data storage solutions is reviewed in this paper.

In recent years, cybersecurity threats have become more sophisticated and they have the tendency to be hard to detect and prevent. This has led to a growing interest in the use of Artificial Intelligence (AI) in cybersecurity. However, the adoption of AI in cybersecurity also raises concerns about the risks and ethical implications associated with its use. The aim of the paper is to reveal the potential benefits and challenges of AI in cybersecurity, and also to propose an educational course in AI in Cybersecurity for managers, who should be aware of current developments in this sensitive field.

Generative Artificial Intelligence (AI) offers tremendous potential in various domains but also raises critical ethical concerns. Key issues include the potential for misuse in spreading misinformation, the perpetuation of biases present in training data, the implications for privacy and data rights. It is crucial for developers, regulators and the public to collaborate in establishing robust ethical frameworks. These frameworks should guide the development and deployment of generative models, ensuring responsible use that maximizes benefits while mitigating risks. In this context the transparency, accountability, and public engagement emerge as foundational principles for navigating the ethical landscape of Generative AI. The aim of the paper is to propose a conceptual framework for solving ethical issues in Generative AI, after exploring issues, challenges, and implications for finding a more accountable integration of the technology into society.

The current predominant method for partial discharge (PD) localization necessitates manual measurements of the distances between sensors and time-of-arrival differences by maintenance personnel. To enhance the intelligence of automatic localization techniques, this paper conducted research on PD localization within Gas-Insulated Switchgear (GIS) using digital twin technology. We propose an automatic PD localization technique for GIS based on digital twins. Initially, we establish a digital twin model of a typical GIS, compute the shortest distances between any two points within the model, thus creating a “time delay fingerprint database.” Using this database in conjunction with the time-of-arrival difference method allows for the automatic localization of partial discharges. The effectiveness of this algorithm has been verified through simulation experiments.

This paper discusses the data-driven regression modelling using first-order linear ordinary differential equation (ODE). First, we consider a set of actual data and calculate the numerical derivative. Then, a general equation for the first-order linear ODE is introduced. There are two parameters, namely the regression parameters, in the equation, and their value will be determined in regression modelling. After this, a loss function is defined as the sum of squared errors to minimize the differences between estimated and actual data. A set of necessary conditions is derived, and the regression parameters are analytically determined. Based on these optimal parameter estimates, the solution of the first-order linear ODE, which matches the actual data trend, shall be obtained. Finally, two financial examples, the sales volume of Proton cars and the housing index, are illustrated. Simulation results show that an appropriate first-order ODE model for these examples can be suggested. From our study, the practicality of using the first-order linear ODE for regression modelling is significantly demonstrated.

This project addresses in reducing pneumonia-related mortality by involving a development of an automatic screening system that analyzes X-ray images and distributed image data storage. This objective is achieved through a collaborative approach, employing Federated Learning to establish a platform that fosters cooperation of the hospitals. Each of them contributes to the advancement of diagnostics by constructing local models from their respective data. These local models are then transmitted to a central server that aggregate models to create a comprehensive global model. This process ensures that the resulting detection model remains unbiased. Importantly, patient data security is upheld, as the central server stores only global models but not sensitive patient information. Project also introduces an innovative system for archiving medical image data for a multifaceted purpose: it archives, anonymizes, and secures image, while also curating a dataset necessary for training a Computer Assisted Diagnosis systems. We are underlining this work to pushing the boundaries of machine learning especially in term of healthcare domain by a strong emphasis on patient privacy and anonymity.

The safe operation of ro-ro passenger transport has become the focus of attention in recent years, and reasonable ship stowage is the basic guarantee for the safe navigation of ro-ro passenger ships. In view of the vehicle stowage problem, the Intelligent Stowage Expert Decision-Making System for Ro-Ro Passenger Ships is proposed. Based on the real-time loading situation of the ship and combined with the ship attitude monitoring information, the expert decision-making system analyzes the balance status of the hull and guides the dynamic stowage of vehicles. It aims to achieve scientific stowage of ro-ro passenger ships through the decision-making system and obtain a reasonable stowage plan, thereby improving ship stowage efficiency and transportation safety.

Highway wind-blown sand reduces vehicle visibility and is likely to cause vehicle rollover and other accidents. In this paper, k-means algorithm is used to extract the annual maximum wind speed data of 10 stations in the central and western regions of Inner Mongolia and analyze the wind speed in different seasons. Through the results, it can be clearly seen that all stations in Inner Mongolia operate from April to June and from October to December. Wind speed shows strong characteristics. The strong performance of the wind speed is also affected by the topography of the central and western Inner Mongolia, which further increases the unsafe factors of road traffic. In addition, it can also be seen from the results that the highway environment around the meteorological station in Wengeng Town of Bayannur City is more susceptible to the influence of wind-blown sand, sandstorms, sand accumulation and other phenomena. Through the cluster analysis of this paper, it can provide scientific data support for the highway maintenance department to effectively resist the disaster caused by wind-blown sand.

Compact range is one of the most availability techniques to measure RCS indoor. Typical compact range has only one feed and one QZ. But offset and rotated feed can form another QZ. This paper is focus on this condition and simulated a novel model of double QZ compact range. A RCS measurement in multiple QZ is addressed and tested in a microwave chamber. It shows frequency division and accurate time-gate help to reduce the clutter form feed, the other QZ and chamber.

A new type of electrodynamic exciter was proposed based on resonant effects in this paper. A PID frequency tracking method was studied to keep the driving frequency consistent with the resonant frequency. The circuit system model of the electrodynamic vibration exciter was built and its impedance characteristics were obtained. It was established that the relationship of between the phase difference and different frequency by simulation. The influence of the output time and stability of the PID parameters were considered, and the parameters were optimal selected, which improved the frequency tracking accuracy of the exciter and speed up the output response speed by 10%.

The main load-bearing structure determines the stability of the space optical remote sensor and is a key structure of the system. It has the characteristics of large structural size, multiple interfaces, high stability and light weight requirements. A main bearing structure composed of aluminum honeycomb composite sandwich panels was designed to meet the high stability requirements of space optical remote sensor. The space optical remote sensor is in an extremely complicated space orbit thermal environment when it works on orbit. In order to solve the problem about the thermal deformation measurement of the main bearing structure of optical remote sensor, a digital photogrammetry system is proposed. The test system is built and the test conditions are designed for test verification. The research results indicate that the stability of the main bearing structure meets the requirements. The accuracy of digital photogrammetry system can reach 0.05mm, which provides a technical basis for the reasonable prediction of optical remote sensor optical structure on-orbit thermal deformation.

Alzheimer’s disease will lead to the atrophy of the Hippocampus. In order to recognize the position changes of Hippocampus, improve the image contour quality, further improve the accuracy of convolution, and achieve the purpose of accurately extracting image information, convolutional neural network is introduced to recognize the Hippocampus region with brain magnetic resonance imaging, and a method combining multi-level 3D U-NET is proposed based on single-stage U-NET. The results showed that this model could enhance the segmentation performance, significantly improved the segmentation accuracy which had certain clinical significance for the brain to recognize the Hippocampus and the automatic discrimination of Alzheimer’s disease.

Deep learning of network traffic is a research method that mimics the structural functions of the human nervous system to identify, classify, and predict data. We propose a new model based on Conv-LSTM to improve the accuracy and efficiency of network encrypted traffic recognition. Based on the public CIC-ISD2017 dataset, the new model is tested and measured, and evaluated based on the constructed confusion matrix and ROC graph. Comparing it with traditional Conv-LSTM, decision tree method, and RF&LSTM methods, it was found that the new model performs better and can perform well in multi classification tasks, with an accuracy rate of up to 99.60%. This model provides a reference solution for relevant applications in the field of network security.