International Journal of Embedded Systems (16 papers in press)

Regular Issues

• A survey on latency and power consumption estimation for embedded systems  
  by Nejra Beganovic, Mattias O'Nils 
  Abstract: Performance evaluation of Internet of Things (IoT) platforms becomes inevitable as the number of IoT devices is constantly increasing. Discussing from the aspect of their interdependences, it is of utmost importance to provide an efficient framework for the analysis of causal relation between consumed power, processing latency, data size reduction, and algorithm computational complexity of embedded systems. As embedded devices, operating often on limited and unreliable energy sources such as batteries or other energy harvesters, are the devices with the highest need for optimal power use, the main focus of this contribution is to review energy consumption modelling approaches and their relation to a latency modelling framework. Such analysis is necessary to provide the basis for efficient system design from early design stage and to guarantee the fulfillment of all system requirements. Accordingly, the paper points out not only existing challenges but also the possibilities for improvements with respect to power/energy savings.
  Keywords: power estimation; energy consumption; internet of things; embedded systems;.
  
  
• Field-embedded database query system based on natural language processing  
  by Fei Long 
  Abstract: This research seeks to develop a paradigm that will improve user-database interaction. To convert the user's queries into structured query language (SQL), natural language processing (NLP) is needed, and then the SQL can be processed quickly by the query system in the embedded database. The primary goal of NLP is to facilitate human-computer interaction with little reliance on programming knowledge. To access the data efficiently, field embedded database query system (FEDQS) uses NLP to take in 2880 structured queries about train prices and seat availability from the train reservation database and turn them into a SQL query. Therefore, field embedded database query system (FEDQS) is suggested in this research to help the users access the data efficiently. The simulation findings show that the proposed method achieves a translation accuracy of 92%, precision of 91%, RMSE of 7%, and MAE of 9%.
  Keywords: field-embedded database; query system; natural language processing; NLP; structured query language; SQL.
  DOI: 10.1504/IJES.2023.10060443
   
  
• Application of machine learning algorithm in operator shop intelligent selection data Data  
  by Chao Liu 
  Abstract: In order to improve the accuracy of data analysis, this paper applies machine learning algorithms to the analysis of smart selection data in operator shops. This paper introduces common machine learning algorithms, analyses the data to be analysed for intelligent selection in operator shops, applies machine learning algorithms to intelligent selection data in operator shops, and finally analyses the effect analysis of the application of machine learning algorithms, finally concluding that the analysis of intelligent selection data in operator shops using machine learning algorithms can not only improve calculation speed and calculation accuracy, but also improve generalisation. It can also reduce the omission rate of data, in which the omission rate of smart selection data of shop 5 is reduced to 5.67%. Machine learning algorithms will need to be applied in many more ways in future life.
  Keywords: smart selection data; machine learning algorithms; operator stores; applied science.
  DOI: 10.1504/IJES.2023.10060689
   
  
• Mobile sensors-based detection of road conditions and quality  
  by Prabhat Singh, Abhay Bansal, Ahmed E. Kamal, Sunil Kumar 
  Abstract: As road infrastructure is a lifeline of transportation in modern society. Due to the frequent use of roads, maintenance, and monitoring at regular intervals become important. Indian roads have many anomalies factors such as poor construction quality, heavy traffic, poor drainage, weak sub grade, and large variations in temperature that can contribute to the creation of potholes, cracks, etc. Hence, authors are focusing on developing the most efficient and accessible application for road quality detection, that can focus on more problematic areas. In the first part the work is done on the collection of data sets with the help of Android in-built mobile sensors. The second part employs the machine learning algorithm on the dataset to depict the quality of the road. The third part focuses on the deployment of the machine learning model on the server-side and reverting the results to the application. The algorithm is based on machine learning algorithms and comparing the accuracies based on accelerometer data. Best accuracy was received by gradient boosting classifier technique. The accuracy obtained was 94.07% with 88% precisions core for detection of road quality so that accident can be reduced.
  Keywords: real-time road monitoring; smart phone; sensor; Android; machine learning; flutter.
  DOI: 10.1504/IJES.2023.10061009
   
  
• Evaluation of CNN-based computer vision recommended treatments for recognised guava disease  
  by Vishal Kanaujia, Satya Prakash Yadav, Awadhesh Kumar, Victor Hugo C. De Albuquerque, Caio Dos Santos Nascimento 
  Abstract: Climate change poses a particular threat to the agricultural crop production sector. The entire food industry is affected by this issue, not just the farming sector. The diagnosis of plant diseases could be improved by using deep learning strategies, according to several studies. These samples are rarely analysed for their ability to predict quality. Extreme caution is required to organise agricultural output surgically. Detecting high incidence rates in commercial production is difficult because of the unfair model’s unpredictability, resulting in more difficulty in diagnosing reflex plant diseases. The proposed model is designed to identify the guava disease using convolutional neural networks (CNNs) and machine learning for classification. In which autoencoder is used to divide the neural network design in the encoder and decoder. The linear support vector machine is used as a classification to analyse the outcomes of our experiments. Preliminary results from the suggested model indicate a remarkable degree of accuracy (97.5%).
  Keywords: CNN feature extraction; guava disease; auto encoder preprocessing; data augmentation; plant disease detection.
  DOI: 10.1504/IJES.2023.10061388
   
  
• Simulation and application of computer network security monitoring based on multi-difference embedded model  
  by Yuping Li, Ke Li 
  Abstract: In order to strengthen the maintenance of computer network security, this article uses the multi-differential embedding model to monitor, simulate and apply research on computer network security. This article analyses the accuracy, stability and time period of network security through application experiments on two computers of different brands (Dell Precision 3551 and HP ZBook Fury 17 G7). The results showed that the neural network algorithm model had the highest average accuracy, with Dell Precision 3551 at 93.3% and HP ZBook Fury 17 G7 at 95.6%. The Math OS model had the highest average stability, with the Dell Precision 3551 at 77.5% and the HP ZBook Fury 17 G7 at 77.7%. The mathematical operating system model on the Dell Precision 3551 had the shortest average time period at 32.8 seconds, and the UML model on the HP ZBook Fury 17 G7 had the shortest time period at 30.6 seconds.
  Keywords: computer network security; neural network algorithm; embedded model; unified modelling language; UML; network security monitoring.
  DOI: 10.1504/IJES.2023.10061925
   
  
• Evaluation on application of intelligent traffic image recognition system in vehicle detection and tracking  
  by Cheng Liu 
  Abstract: This paper studied from three aspects: the structure of vehicle detection system and the use of intelligent traffic image recognition system video information collection and analysis, the use of intelligent traffic image recognition system to design vehicle detection algorithms, and the use of intelligent traffic image recognition system to track the application of moving vehicles. Through experiments and research, this paper built a new vehicle detection and tracking system, and the satisfaction rate was 19% higher than that of the traditional vehicle detection and tracking system. Compared with the traditional vehicle detection and tracking system, the accuracy of the new vehicle detection and tracking system was increased by 0.28, and the definition was increased by 0.4. This can be in order to better serve people and solve traffic problems such as urban congestion. Therefore, the construction of intelligent transportation system is very important.
  Keywords: intelligent traffic imagery; image recognition system; vehicle detection and tracking; video image processing; intelligent transportation system; ITS.
  DOI: 10.1504/IJES.2023.10062167
   
  
• Computer intelligent device adjustment and fuzzy controller design for embedded ARM  
  by Hansong Ge, Ke Li 
  Abstract: There are more and more researches on fuzzy control. Fuzzy controllers in all walks of life have very successful application cases, but they can be affected by quantification factors in the development process, so most of the control rules obtained are based on personal experience and have great uncertainty. To solve these problems, in this paper, the intelligent device fuzzy controller was designed and studied with the help of advanced reduced instruction set computer (RISC). The optimal control rules were searched by advanced RISC machines (ARM). These rules were used to generate the corresponding fuzzy controller. The experimental results suggested that the fuzzy controller based on embedded ARM was more accurate for the regulation of computer intelligence devices than the controllers based on ant algorithm and genetic algorithm. The accuracy of the controller studied in this paper was above 94%, while the other two adjustments were below 91% and 92%, respectively. The performance of the controller studied in this paper is also better, which is conducive to improve the performance of computer intelligent equipment, improve the use value of equipment, better improve the accuracy of equipment adjustment, improve the processing speed of fuzzy controller for subset rules, and the running speed is faster.
  Keywords: fuzzy controller design; intelligent device adjustment; embedded ARM; ant algorithm; genetic algorithm.
  DOI: 10.1504/IJES.2023.10062880
   
  
• Design and application of digital network teaching resource system for network environment  
  by Guobin Jun 
  Abstract: As the information technique developing, resource construction has become an unavoidable practical problem in college education. The systematic integration of teaching resources has become an important breakthrough to solve this problem. Therefore, this study first extracts hidden structural features of digital network teaching resources through data pre-processing, and adds split and merge operations to K-means algorithm to extract main features. Then use LSTM to optimise CNN to form LSCN. Finally, LSCN is combined with the improved K-means algorithm and applied to the digital network teaching resource system. The results show that the objective function value of the final solution of the improved K-means algorithm is 115. The accuracy of LSCN model in online teaching resource database can reach 94.6% at most, and the running time is 38.6s. After combining the enhanced K-means with the LSCN model, the accuracy of the integration of online courses, digital materials and other resources in the college network education system is more than 93%. It shows that the teaching resources integration method proposed by the research has good effect and efficiency, and can provide a reference method for the further informatisation of the education system.
  Keywords: network environment; teaching resources; K-means; convolutional neural network; CNN; LSTM; data mining; K-means.
  DOI: 10.1504/IJES.2024.10063172
   
  
• Psychophysiological state recognition of middle school students based on vibraimage technology and k-means cluster analysis algorithm  
  by Rui Huang, Xiaoquan Liu, Yunzhen Xue, Zhu Zhang 
  Abstract: Adolescence is a special period for middle school students to have rebellious psychology. How to effectively evaluate the mental health of middle school students and help middle school students successfully pass adolescence has always been the focus and difficulty of psychologists’ research. The typical emotion recognition of middle school students in adolescence is the basis for completing this work. In order to identify the psychological and physiological state of middle school students in adolescence, this paper proposes a method of adolescent psychological and physiological state recognition based on vibration imaging technology-K-means clustering analysis algorithm. In order to verify the feasibility of this method, 74,011 middle school students from 59 schools in Taiyuan City were selected as experimental subjects, and the experimental data were obtained by face-to-face interviews and capturing the facial expression video stream of the interviewees. The research results show that the vibration imaging technology-K-means clustering combination model is feasible for the identification of the psychological and physiological state of middle school students in adolescence, and has certain reference significance for the research work in this field.
  Keywords: K-means clustering; vibration imaging technology; descriptive statistical analysis; adolescence.
  DOI: 10.1504/IJES.2024.10063193
   
  
• Construction and application of online learning mental state diagnosis model based on student learning behaviour data  
  by Xiaohui Ma, Zhongwang Li 
  Abstract: This study addresses the issue of burnout psychology in online learning, which has become prevalent due to educational reforms and the push for educational informatisation, leading to a disinterest in learning among students. It defines the concept and dimensions of online learning burnout psychology using student data, and develops an early warning model using the XGBoost algorithm to predict student burnout effectively. Results indicate the XGBoost algorithm outperforms three other classification algorithms in iteration quality, with minimal difference between actual and training loss, and demonstrates an average absolute error between 1.5 and 2.0, and a mean square error around 1.0. In tests, the model’s accuracy, recall rate, and F1 score were 93.1%, 93.5%, and 0.93, respectively, surpassing comparative models. Thus, this early warning model is highly effective for diagnosing online learning burnout, offering significant improvements over existing methods.
  Keywords: learning data; online diagnosis; educational psychology; promotion of information technology; reform in education.
  DOI: 10.1504/IJES.2024.10063285
   
  
• Empowering intrusion detection in 5G embedded and cyber-physical networks  
  by Nitesh Singh Bhati, Manju Khari 
  Abstract: As intrusion detection systems (IDS) continue to evolve in response to emerging threats to edge devices and embedded devices, various approaches, such as anomaly-based and fuzzy logic-based techniques, have been employed to construct effective IDSs. More recently, with the introduction of 5G to the public usage, the data is dynamic and heterogeneous in nature due to which the integration of machine learning methodologies has gained prominence in IDS development. This research paper introduces a novel ensemble-based approach for enhancing intrusion detection within the context of modern 5G embedded and cyber-physical network security. The proposed technique leverages an optimised CatBoost classifier to fortify the defences of contemporary networks against potential breaches. To evaluate the efficacy of the proposed approach, experimentation was conducted using the KDDCup99 dataset. The results yielded by the proposed technique exhibit a remarkable 99.96% accuracy in detecting intrusions. This research contributes valuable insights to the realm of 5G embedded and cyber-physical by leveraging an ensemble-based approach with a focus on CatBoost optimisation, this study advances the field’s understanding of bolstering intrusion detection capabilities within the evolving landscape of modern distributed networks.
  Keywords: intrusion detection technique; 5G embedded; cyber-physical network; machine learning; CatBoost.
  DOI: 10.1504/IJES.2024.10063474
   
  
• Evaluation and design of changes in optical cable and fibre optic online monitoring system based on digital communications technology  
  by Qiong Cheng, Huaijun Li, Yurong Zhen, Qing Wang, Zhiyi Jia 
  Abstract: In order to solve the problem of low signal transmission efficiency and susceptibility to noise interference in online monitoring systems in fibre optic communication construction, this paper conducts effective research on the analysis and design of changes in fibre optic cable online monitoring systems using digital communications technology (CT). This paper conducted tests from three aspects: real-time analysis of changes, accuracy, and signal transmission efficiency. The test results show that at the accuracy level of change analysis, the average relative error level of digital CT used for fibre optic cable change analysis is about 6.381%, while the average relative error result of fibre optic change analysis in traditional online monitoring systems is about 7.595%. From the comparison of accuracy results, digital CT can enhance the stability and reliability of online monitoring systems, improve the accuracy of fibre optic change analysis, and promote the healthy development of fibre optic communication.
  Keywords: optical fibre; digital communications technology; online monitoring system; change analysis and design.
  DOI: 10.1504/IJES.2023.10063701
   
  
• Embedded detection system based on edge computing cloud platform image sensor  
  by Wei Xu, Yujin Zhai 
  Abstract: Aiming at the problems such as large volume, high price and insufficient flexibility of traditional imaging equipment in the current embedded inspection system, this paper aims to explore a method to optimise the embedded inspection system and make it better used in the industry. In this paper, a new embedded imaging device under cloud computing environment is studied first, and then image sensing technology is used to replace the traditional image data acquisition method, and image data is analysed and processed by cloud software. In order to verify the performance of the system, an empirical study is also carried out. The results show that in the embedded monitoring system of image sensor, the pixel detection rate of the smart operator in the night image detection is 7.4% higher than the original pixel detection rate. This shows that the system has good performance and can realise the night image detection well.
  Keywords: edge computing; cloud platform image sensor; embedded detection system; canny operator.
  DOI: 10.1504/IJES.2024.10064194
   
  
• Call-site tree and its application in function inlining  
  by Arthur Yang, Shih-Kun Huang, Wuu Yang 
  Abstract: Inlining avoids the overhead of function invocation and return and makes many other optimizations more effective because the inlined functions are larger. Intuitively, it is preferable to inline the call sites that are executed more often. We use profile information in deciding the call sites to be inlined and the order of inlining. We may keep an execution count at each call site in order to find the call site that is executed most often. A difficulty occurs because execution counts of all call sites may change after a call site is inlined. Consequently, profiling needs to be repeated after every inlining operation. We design a new data structure and new analysis algorithms. Then we can decide the new execution counts of call sites after one or more call sites are inlined without re-profiling. This also applies to call sites that are created due to previous inlining.
  Keywords: inlining; call-site trees; compiler optimisation; LLVM.
  DOI: 10.1504/IJES.2024.10064933
   
  
• Chatbot for mental health diagnosis using data augmentation techniques and deep learning  
  by Neel Ghoshal, Vaibhav Bhartia, Balakrushna Tripathy, Anurag Tripathy 
  Abstract: Statistical results obtained during recent surveys have indicated that about 8% of the total mass of the people in the world suffer from mental health problems. A scalable option that offers an interactive way to engage consumers in behavioral health interventions powered by artificial intelligence might be chatbots Although several chatbots have showed early efficacy results that are encouraging, their efficient utilization by people is not properly confirmed. In this paper a customized chatbot framework is proposed and developed using natural language understanding (NLU) mechanisms. The framework comprises of a unique two-tier modular functionality of an empathetic conversational model with a simultaneous implementation of a classification model. Along with this, the framework uniquely works on a data-driven knowledge based and predictive pattern, providing a holistic service to any user. The dataset used is completely scraped and prepared manually to inculcate the various mental health diseases and the appropriate responses provided by professionals
  Keywords: chatbot; mental health; natural language processing; NLP; deep learning; soft computing.