Adaptive entropy theory polymerization method for path optimization in edge computing educational systems

Abstract

In this paper, through the edge computing application path, the educational evaluation system was optimized using the adaptive entropy theory polymerization method which based on applying the path. By adding multiple constraints to filter out nodes and educational evaluation edges that do not meet the requirements, the improved algorithm is used to optimize the redundant paths to avoid loops and node detour problem. To improve the accuracy of education evaluation and evaluation, ensure the load balance in the domain, and solve the problems of single evaluation attribute and high overlap of education evaluation paths. This paper proposes a multi-attribute education evaluation model that refines the evaluation attributes of education evaluation and uses analytic hierarchy process perform weight distribution. The algorithm can improve the accuracy of the evaluation of the education evaluation system while ensuring the computational efficiency, and can ensure the load balance within the domain, and improve the network survival time.

Keywords

Edge computing education evaluation system application path optimization research

1 Introduction

With the development of the global informatization wave, education has entered a new historical period, and the development of personalized education has become the consensus of the international community [1]. The wide application of personalized education can make every student’s development conform to his characteristics, and it is the embodiment of humanization and bio-oriented. The object of educational evaluation is the educational and teaching ability of normal students. This ability is difficult to judge through traditional paper-and-pencil tests. Its evaluation is based on actual educational and teaching behaviors in real situations [2]. The performance evaluation requires students to complete the behavior we want to complete, not the behavior that looks like but did not happen. The performance evaluation is the most important evaluation method in internship evaluation [3]. A complete performance evaluation is composed of multiple links such as evaluation goals, performance tasks, and performance evaluation tools [4]. Among these links, the evaluation tool that determines the evaluation results is particularly important because of its guiding and evaluation functions [5]. The key point is that performance evaluation tools can tell evaluators what to evaluate and how to evaluate, which can help instructors and normal students make an accurate evaluation of internship performance [6]. The role of performance evaluation tools is not limited to the evaluation itself. Under the demand for teaching and evaluation integration, performance evaluation tools are also a learning tool that can support students’ learning, and the development of advanced thinking skills and evaluation tools can even be fuzzy the boundary between teaching and evaluation. The performance assessment tool will not only help normal life according to clear when the results of the evaluation with which the location of their former level, but also help normal students find direction for gaps and bridge the gap between the acceptable level [7, 8]. It is a very important way to improve educational evaluation by using performance evaluation tools as the starting point. Self-expression since the rise of the concept of sexual evaluation, the research on performance evaluation tools has been continuously developed and improved. Through the efforts of researchers, the theoretical trunk of performance evaluation tools has been formed, and some theoretical branches involving performance evaluation tools of specific disciplines have grown [9]. However, the performance evaluation tools related to the field of practice still lack special and systematic theoretical research. To make up for this deficiency, this research puts its foothold on the development of performance evaluation tools for educational practice. This will not only help to add branches in the practice field to the trunk of performance evaluation tools but also help make the performance evaluation tools more effective. The trunk of the evaluation tool becomes stronger [10].

Given the limited resources and openness of edge computing, many experts and scholars have adopted different methods to construct educational evaluation and evaluation models suitable for edge computing environments. However, the current research results are few and there is still a lot of room for discussion [11]. It proposed a party based on social evaluation of uncertainty reasoning education case, use of the flexibility and elasticity of uncertainty reasoning derived evaluation value education, combined with the Bayesian theory and DS evidence theory of direct [12]. Indirect education evaluation is used to evaluate and filter education evaluation opinions through reinforcement learning, which significantly reduces the adverse effects of biased opinions [13]. Evaluation Model of vehicle safety education in the fog node bearing load vehicle storage operating environment of local information, and education for added safety evaluation model of educational evaluation vehicle for local management to ensure the accuracy of the vehicle authorized to receive information. The three aspects of familiarity, similarity and timeliness weight different educational evaluation dimensions, and use vector machines and multi-weight subjective logic methods to maintain and update the educational evaluation information of local vehicles [14]. The interest similarity evaluation model proposed a social network education, use section of the community point of interest will be clustered, and evaluation by Kalman filtering technique to predict node education, to prevent possible malicious attacks [15]. However, the above model does not consider the unreliability of the education evaluation model itself and ignores the possibility of malicious recommendation. Based on the educational evaluation management framework of measurement theory, the measurement error of the educational evaluation is regarded as the confidence level, which measures the credibility of the equipment educational evaluation and ensures the accuracy of the educational evaluation value [16]. The model can also help service providers dynamically allocate resources based on real-time education evaluation information [17]. Blockchain consensus mechanism to prevent the number of educational evaluation data is forged or tampered with and does not rely on a trusted third party and inter-domain hypothesis evaluation of education while improving the accuracy of the assessment to ensure the security of the data [18, 19]. Degree feedback and feedback from malicious devices improve computing efficiency while resisting malicious attacks. Aiming at the single problem of the evaluation index, the evaluation index of the same quality service intensity of the node distinguishes the trustworthiness of the recommended node from the three perspectives of node similarity, evaluation difference, and self-education evaluation, filtering useless recommendation information, and reducing the impact of non-intrusive factors on education Evaluate the impact of the assessment [20]. The base station network edge is regarded as a trusted third party, the use of combined work to determine the extent of educational evaluation nodes and node density, energy levels of educational evaluation, and proposed a node-based single-hop clustering mechanism, the program fully into account. The energy consumption of the node reduces the resource consumption in the calculation of education evaluation [21].

A large number of excellent results have emerged in the education evaluation model in the edge computing environment, which effectively promotes the development of the education evaluation model in the edge computing, and improves the reliability and computational efficiency of the evaluation system. In the complex edge computing environment, the requirements for computing efficiency of education evaluation and evaluation models are getting higher and higher, and it is necessary to research computing efficiency from more angles [22]. In the process of acquiring an education evaluation, the multi-attribute problem of education evaluation is ignored, which leads to the fact that education evaluation information cannot reflect the subjectivity and complexity of the relationship between education evaluations between devices. This research takes teacher-training professional certification as the theoretical perspective and takes the design of the educational practice performance evaluation tool as the specific research object. The theoretical and practical meanings that can be expressed are as following [23]. Since the rise of the concept of performance evaluation, the research on performance evaluation tools has been continuously developed and improved. Through the efforts of researchers, the theoretical trunk of performance evaluation tools has been formed, and some theoretical branches involving performance evaluation tools of specific disciplines have grown. However, the performance evaluation tools related to the field of practice still lack special and systematic theoretical research. To make up for this deficiency, this research puts its foothold on the development of performance evaluation tools for educational practice. This will not only help to add branches in the practice field to the trunk of performance evaluation tools but also help make the performance evaluation tools more effective. The trunk of the evaluation tool becomes stronger.

2 Design of edge computing education evaluation system

2.1 Improved edge calculation

Aiming at the problems of computational load and redundant educational evaluation paths in existing education evaluation models, an edge computing education evaluation model based on the DFS algorithm is proposed. The educational evaluation relationship between devices is abstracted into a directed weighted graph, and it is defined and explained. At the same time, multi-source educational evaluation is aggregated based on the information entropy theory, and the degree of difference between multi-source educational evaluation is corrected. Secondly, by adding multiple constraints such as education evaluation threshold, path length limit, sliding window, etc., nodes, and education evaluation edges that do not meet the requirements are filtered in advance. Based on the modified DFS algorithm evaluation path redundancy educational path optimization, and the node ring detour to avoid the problem, reducing the computational cost of education evaluation path during formation. Finally, the proposed method is compared with the PSM model and the RFSN model. The results show that the model in this paper can effectively reduce the resource cost of edge devices and improve the effectiveness of the education evaluation model.

Edge education computing architecture is shown in Fig. 1. The system comprises a trusted computing cloud edge, an edge server service layer, and an edge layer three. The data center of the cloud layer stores various data of users. The cloud layer can issue computing tasks to the edge server, and the task is returned to the cloud layer after the edge layer is executed. The edge server layer is mainly responsible for the communication and data transmission between the cloud layer and the edge layer. In this article, it assumes to be credible. Each edge server (ES) stores the interactive information of various devices in the area, and is responsible for the supervision, ensuring the operation of the education evaluation model and the accuracy of the education evaluation results, and aggregating the feedback from the equipment. We reduce the resource cost of equipment. The edge layer each species to perform tasks at the edge of the edge device (ED) configuration, comprising smartphones, wearable devices, vehicles, and so on. Edge devices are divided into different domains based on factors such as location and characteristics, and each domain is managed by an edge server. Devices can request services within or between domains according to task requirements to achieve interactive cooperation. Before the collaboration service, the device sends request information to the edge server to ensure the reliability of the partner. After the completion of collaboration services, equipment updates its list of educational evaluation and to report regularly to the upper edge servers pass education evaluation information.

Fig. 1

Edge computing model.

As shown in Fig. 1, the interaction between devices in the edge layer forms a directed network graph. The edge layer is evaluated between education may be seen as a large directed weighted graph, the participating device interaction pumping as to evaluate the relationship between education nodes in the graph. There is a device abstraction to the edges between nodes. Model Each node in the network can construct the education evaluation relationship between any nodes through path search and graph fusion, and further aggregate the education evaluation information between nodes to obtain the global education evaluation relationship.

Educational evaluation is a subjective judgment on whether a device can provide good network services. It provides guidance information when the device selects task objects, avoids establishing cooperative relationships with selfish nodes or malicious nodes, and ensures safe and reliable communication between devices. Generally, the education evaluation relationship between devices in the education evaluation model is divided into direct education evaluation, recommended an education evaluation, and comprehensive education evaluation. This article adds the concept of feedback education evaluation on this basis, which can better describe the education evaluation relationship between devices.

Direct education evaluation is an education evaluation score obtained by the device aggregation of multiple interaction results. The set of scores for the most recent service interaction provided by node i to node j is denoted as L_ij, where the success score of the interaction is 1, and the failure score is 0. The education evaluation sequence will be stored in the node. Direct education evaluation DT_ij uses the risk probability model [to calculate: ${DT}_{ij} = \frac{m^{s} \times (m^{f})^{(- 1 / α)}}{m^{s} + m^{f}}$ (1)

Among them, C _s is the number of successful interactions, C _f is the number of failed interactions, and the penalty factor f can effectively prevent malicious nodes from sudden attacks after the accumulation of higher education evaluations.

The third-party node regards its direct education evaluation of other nodes as recommended education evaluation and recommends it to the source code, which is stored in the edge server in the area in the form of a matrix and is managed and updated by the server. To prevent boasting, the server sets the value of DT₁₁ DT_nn to 0. $H = [\begin{matrix} {DT}_{11} & {DT}_{12} & . . . & {DT}_{1 m} \\ {DT}_{21} & {DT}_{2 m} & . . . & {DT}_{2 m} \\ . . . & . . . & . . . & . . . \\ {DT}_{m 1} & {DT}_{m 2} & . . . & {DT}_{mm} \end{matrix}]$ (2)

Feedback education evaluation is an education evaluation path from subject to an object formed by direct education evaluation and recommended education evaluation. Through comprehensive consideration of feedback from multiple education evaluation paths, objective credibility is described. The feedback education evaluation FT _ij calculation formula is as follows: $K_{ij}^{k} = {DT}_{ia} \times {DT}_{ab} \times {DT}_{bc} \times \cdot \cdot \cdot \times {DT}_{nj}$ (3) ${FT}_{ij} = \frac{1}{m} \sum_{k = 1}^{m} K_{ij}^{k}$ (4)

Comprehensive education evaluation is the overall education evaluation of one device to another. It is the final education evaluation value obtained by aggregating direct education evaluation and feedback education evaluation in some way. When there is no direct interaction record between nodes, the feedback education evaluation is regarded as a comprehensive education evaluation to construct the education evaluation between unfamiliar nodes. To improve the reliability of education evaluation, an adaptive aggregation method based on information entropy theory is used to aggregate comprehensive education evaluation values, which can effectively measure the degree of disorder between education evaluation values and correct the difference between education evaluation values. $T_{ij} = ϖ_{1} \times {DT}_{ij} + ϖ_{2} \times {FT}_{ij}$ (5) $ϖ_{1} = \frac{1 - \frac{M ({DT}_{ij})}{{log}_{2} {DT}_{ij}}}{[1 - \frac{M ({DT}_{ij})}{{log}_{2} {DT}_{ij}}] + 1 - \frac{M ({FT}_{ij})}{{log}_{2} {FT}_{ij}}}$ (6) $ϖ_{2} = \frac{1 - \frac{M ({FT}_{ij})}{{log}_{2} {FT}_{ij}}}{[1 - \frac{M ({DT}_{ij})}{{log}_{2} {DT}_{ij}}] + 1 - \frac{M ({FT}_{ij})}{{log}_{2} {FT}_{ij}}}$ (7)

In general, the larger the amount of feedback information, the more reliable the calculation result of the node education evaluation value. The path search and aggregation require a lot of time and space resources, plus the dishonest feedback and collusion attacks of malicious nodes, etc., the calculation cost of the node is virtually increased, leading to deviations in the final education evaluation value. The optimization method of the education evaluation graph can be considered from the following two perspectives. One is to filter the nodes and education evaluation edges that do not meet the requirements of the education evaluation in advance, and greatly reduce the number of traversals of the education evaluation graph. The second is to delete redundant educational evaluation edges during the formation of educational evaluation paths to avoid node detours and loops and reduce the load of educational evaluation calculations.

2.2 Evaluation system education system designs

In personalized teaching based on big data, teachers, and students use interactive whiteboards and iPads to carry out various teaching and learning activities, realize the all-round digitalization of the teaching process, and comprehensively understand the process data of the main four links in the teaching process. Including the teacher’s pre-class preparation, classroom teaching, a subjective question review, review, and test data. The data of the students’ pre-class preparation, learning and listening during class, doing exercises, reviewing and summarizing after class, and checking for missing data. The refined teaching management process of education managers, teaching evaluation and reminders and helping front-line teachers adjust teaching data. The data on parents’ understanding of students’ situation and targeted assistance are comprehensively collected. Only by ensuring all-round collection of data can we ensure the good development of personalized teaching, make the teaching process use data to speak, and ensure the personalized development of students.

Based on constructing the knowledge system tree, different types of digital teaching media need to be used to present knowledge points, form digital learning resources, and determine a rich media space. These digital learning resources include texts, courseware, cases, videos, tutorials, and audios. Teachers can push different learning resources according to their teaching needs. Students can choose learning resources according to their learning style. Knowledge points can be presented in various forms of digital resources. Many knowledge points constitute a media space for personalized teaching based on big data. In the process of pushing students, it is necessary to choose the appropriate resource presentation form according to the different types of knowledge points: concept and theorem-based knowledge points can push text-type resources to students to help students grasp concepts and understand theorems. Complex skill-based knowledge points can push courseware PPT type resources to students. The application-oriented knowledge section of the key and difficult to make into an apparent frequency, push the animation resources students, help students establish a logical system of knowledge, as shown in Fig. 2.

Fig. 2

Education Evaluation System.

There are a series of tasks that need to be processed. These tasks upload basic information of the mobile device and the size, priority, and required CPU cycles of the tasks to be processed through a complex network environment. MEC server on the base station, MEC service after service is to obtain this information, the priority task of running the migration algorithm based on this chapter. The algorithm is divided into two steps. The first step is an initialization, and the second step is to execute the channel resource reverse auction algorithm to finally get the task migration decision, and then transmit the result of the decision to each mobile device, and the mobile device based on the decision information complete task scheduling and processing.

In task migration, the basis for deciding whether to process task k on the mobile device or the MEC server is that the task processing time overhead is minimal. Through the previous analysis, it can be seen that the processing delay of the control task of the MEC server is composed of two parts, the stay delay, and the transmission delay. When tasks increase, the transmission rate of the channel will decrease, and the required transmission delay overhead will increase. If task is in the exclusive channel, the processing delay of the MEC server is still greater than task. For the processing delay of the mobile device, the task is directly scheduled to the mobile device for processing.

The interaction between students and learning resources is also called individual interaction. Massive learning resources are the basis of individualized interaction. However, learning and interaction based on mass learning resources alone will dazzle students and get lost in the world of resources, based on big data. Through the collection and analysis of data in the whole process of student learning activities, personalized teaching can recommend personalized learning resources for students in line with their learning style and cognitive characteristics, which help students find their direction in the resource trek, and promote student personality Growth. This kind of two-way, diverse interaction between students and teachers, students and students, students and resources, based on real data, is an indispensable element in personalized teaching. Through this interaction, it not only can guide to increase the students to acquire knowledge and improve student achievement, but also can develop the innovative spirit and practical ability of students to help students grow in the competition. Progress in the cooperation helps teachers’ breeders to students of good sentiments, knowledge of meaning construction by changing the direction, and it improves the quality of education, and promote the development of students’ personality.

2.2 simulation experiments and analysis design

To verify the effectiveness of the education evaluation model and resource expenditure, this paper uses MATLAB to conduct simulation experiments and compares and analyzes the PSM model and the RFSN model. The simulation of the detection region is set to 200 m square-shaped, randomly placed 200 analog resources constrained nodes of the edge layer devices, particularly simulation parameters are shown in Table 1.

Table 1
Simulation parameters

Parameter 7 Description

N 10 Total number of nodes

ES 11 Number of edge servers

M 9 Percentage of nodes

W 8 Sliding window

a 8 Penalty factor

Parameter	7	Description
N	10	Total number of nodes
ES	11	Number of edge servers
M	9	Percentage of nodes
W	8	Sliding window
a	8	Penalty factor

To test the edge closer to the real computing environment, the success rate of interaction of the node-set to the normal 90%, die proposed node abnormality caused due to non-invasive factor. The malicious nodes in the experiment are randomly selected and divided into two categories. One category provides malicious services and provides false recommendation information to other nodes, and the other category provides honest services, but provides slander honest nodes and exaggerates similar nodes. The proportion of these two types of nodes in malicious nodes is 50%. The experimental part of the education domain does not distinguish between art. Between a root, edge server may be as necessary to exchange the respective evaluation information management education.

The process from the primary evaluation dimension to the second evaluation dimension, from the secondary evaluation dimension to the evaluation element, and from the evaluation element to the evaluation sub-element is a process of making the evaluation content from abstract to concrete. The hierarchical structure of the evaluation content allows us to know what performance of the teacher students in the internship should be observed, but what we need to know is how to make scientific and accurate judgments on the degree of the teacher students’ internship performance. The judgment of the performance level needs to rely on certain standards. In performance evaluation, this standard is composed of two types of things, one is the evaluation level and the other is the descriptor. They correspond to each other, that is, the evaluation level. Corresponding to the descriptor, the descriptor also corresponds to the evaluation level. Table 1 shows the relationship between observation points, evaluation levels, and descriptors.

The rating scale is representative of normal levels of student performance at a certain observation point identifier, which is used to distinguish normal student’s practice the table is composed of the superior to inferior to varying degrees. Regarding the number of evaluation grades and the setting of evaluation grade expression styles, based on consulting several masters and doctoral dissertations of the educational practice evaluation index system, set the number of evaluation grades to 4, indicating that the style was set as excellent. In terms of the degree of distinction, passing is the minimum requirement for internship performance, so there must be two grades of passing and failing. Those who perform exceptionally well above the pass are rated as excellent, and those who perform between excellent and passing are rated as good, as shown in Fig. 3.

Fig. 3

Evaluation steps.

The system provides teachers with a large number of lesson preparation resources. All the resources are labeled according to the knowledge system tree for easy retrieval. At the same time, it is connected with Baidu question bank and subject network and is provided to teachers of various subjects for free. Based on the syllabus and teaching materials, in-depth study of the subject, the establishment of the knowledge system of the subject, the teacher’s lesson preparation materials are all divided by each knowledge point, making lesson preparation easy and clear at a glance. Before preparing lessons, teachers can log in to the teacher terminal to show the comprehensive situation of the students’ knowledge, preview, and exams in the class, helping teachers to prepare lessons accurately. To have a deeper understanding of whether the personalized teaching system based on big data can help students’ learning effect, this study conducted classroom learning behavior records for the experimental group and the control group to observe and study the students’ classroom status. It is mainly recorded and observed from four aspects: students’ pre-class preview performance, classroom learning interaction, classroom learning order, and classroom learning status.

From the above observations, it can be seen that the experimental group performed better than the control group in four aspects: pre-class preview, classroom learning interaction, classroom learning order, and classroom learning status. Students in the experimental group can take advantage of the learning resources provided by the teacher and the system to actively complete the pre-class preview. In the classroom, the communication and interaction between students and teachers and students can be realized through collaborative inquiry learning methods. Through teachers’ active guidance, create a good classroom learning atmosphere, and maintain the best state of classroom learning.

3 Results and analysis

3.1 Comprehensive evaluation analyses

To analyze the ability of the model in this paper to distinguish between malicious nodes and normal nodes, the initial education evaluation value of the node is set to an intermediate value of 0.5, and the disguised node begins to provide malicious services after interaction cycles. The experimental results are shown in Fig. 4.

Fig. 4

Comparison of education evaluation values between normal nodes and malicious nodes.

As shown in Fig. 4, with the increase in the number of interactions between the nodes, the nodes of the comprehensive evaluation of the value of education in 10 weeks simulation showed a tendency period, the normal node evaluation value education and gradually rises close to one. The node’s educational evaluation value decreases and approaches zero. After a certain camouflage node education accumulated evaluation began offering malicious service, education evaluation value decrease rate is very fast, in the 7th cycle of an intermediate value less than 0.5, the evaluation value obtained by certain educational discrimination. Therefore, the mold -type nodes can effectively distinguish between normal and malicious nodes, comprehensive evaluation of the value of education was also well reflected in the behavior of nodes.

The successful interaction rate refers to the ratio of the number of successful interactions to the total number of interactions. It can reflect the ability of the education evaluation model to resist malicious code attacks. A larger successful interaction rate indicates that the education evaluation model has high reliability. To verify the effectiveness of the program, conducted simulation experiments interactive cycles, each node initiates a service in each cycle will service the request, the node at the same time as the requester and provider services. We set the proportions of malicious nodes to 10%, 20%, and 40%, respectively, representing different honesty network environments, to examine how the successful interaction rate of the education evaluation model varies with the number of interactions under different proportions of malicious nodes. The result is shown in Fig. 5.

Fig. 5

Successful interaction rate under different node ratios.

From the chart you can see the 6, with the increase in the number of interactions, the success rate will be interacting with the increase of the number of interactions and becomes high, the growth rate is gradually slowing down, eventually reach a steady state. You can see that with the increase in the proportion of malicious node, time to adapt educational evaluation model will become a long, successful interaction rate will decline. From the comparison of the models in the Fig. 6, the model in this paper can quickly reach a stable state as the number of interactions increases, indicating that the model is highly adaptable and can effectively identify malicious nodes. It randomly selects the source node and the target node from 200 nodes, we obtain the education evaluation subgraph between the two nodes, and observe the degree of change of the recommended education evaluation value between the nodes when the k value gradually increases.

Fig. 6

The influence of k value on the recommended education evaluation value.

When k is 20, the oscillation range of the education evaluation value is controlled within the range of [–0.02, 0.02]. With the increase of the k value, the degree of oscillation of the education evaluation value gradually decreases, slowly reaching a stable state. When the proportion of malicious nodes is 10%, the educational evaluation value tends to stabilize after the number of paths reaches 20. As the proportion of malicious nodes increases, the value of k that makes the education evaluation value reach a steady-state gradually increases. Taking into account the different proportions of malicious, herein the value of k is set to 40 then down experiments.

3.2 Evaluate the effectiveness analysis

For the repeatability limit factor, the smaller the initial value, the more severe the limit on the repeatability of the educational evaluation path, but it is also likely to cause the number of educational evaluation paths to be less thank, which causes the limit factor to increase continuously. Algorithm complexity. The repetition of the incremental factor is too small and cannot increase the value of the repeat evaluation path of education. The value is too large can lead to repeating too much of an increase. The relationship between repeatability and path length is shown in Fig. 7.

Fig. 7

The relationship between repeatability and path length.

In the actual situation, since the average path length of the sparse matrix and the dense matrix is different, it is impossible to use a fixed value to measure the path repeatability. It can be considered to start with a stricter degree of repetition. Considering comprehensively, this article sets both the repeatability limit factor and the repeatability increment factor to 0.2.

With the increase in the number of interactions between nodes, the network energy of each model shows a different degree of decline. The network energy consumption curve of the model in this paper is relatively flat, and finally, the remaining energy of the overall network is also higher than the other two models. This is that the RFSN model needs to carry out a global education evaluation iteration, and the education evaluation value of this model is stored in the edge server, and the calculation and aggregation process of the education evaluation is completed by it, without the participation of resource-constrained nodes. Edge nodes only need to periodically send education evaluation information and query requests to the server, without the need for global education evaluation iteration. Therefore, this model can effectively reduce the energy consumption in the process of education evaluation transmission and improve the life cycle of the network, as shown in Fig. 8.

Fig. 8

Number of nodes.

As shown in Fig. 8, with the increase in the number of interactions, each with varying degrees of model nodes gradual decline case conditions. RLTS model and RFSN node model have died after 500, respectively, in 2000 and 2500 after the whole unit died. Because this will add multiple nodes energy attribute set, under similar circumstances other property values, prefer to interact with high energy nodes, the nodes to avoid high evaluation value of education early decline, in the beginning, emerge after section 1500 points deaths, Which shows that the multi-attribute model can ensure the load balance within the domain to a certain extent and extend the lifetime of the overall network.

As shown in Fig. 9, the migration ratios of tasks with different priorities are on a downward trend, which is caused by the large occupation of communication and computing storage resources in the mobile edge computing network. When the number of tasks is constant, the proportion of high-priority tasks selected for migration is greater than that of low-priority tasks, because high-priority tasks can prioritize communication, computing, and storage resources, ensuring the service quality of urgent tasks in telemedicine. As the number of tasks increases, the rate of decline in the proportion of task migration continues to accelerate. This is because the communication and queuing delays in the mobile edge computing network continue to increase, and more tasks are handled by mobile devices. For a priority task, in the best and worst cases, we choose to migrate to MEC service ratio of the task processing task is respectively 97% and 64%. The game theory can collaborative consumption optimization support for the high-priority task, while low-priority task can still choose to MEC service processing on the device, reducing the average processing delay task, which can be seen for remote monitoring field. Task priority is very important to ensure the service quality of high-priority tasks, as shown in Fig. 10.

Fig. 9

As the number increases, the migration trend of different priority tasks of this algorithm.

Fig. 10

Evaluation effect analysis.

In the survey on the implementation of teaching, 84% of students believe that a personalized teaching system based on big data can improve their classroom enthusiasm, and 80% of students believe that the use of this system can promote knowledge acquisition, internalization, and consolidation. Through real-time communication and exchanges with teachers and classmates and obtaining feedback, students can solve difficult knowledge points promptly, which enhances students’ interest in learning and fully mobilizes students’ enthusiasm and initiative in classroom learning. Through the study of the massive learning resources provided by the system, under the guidance of teachers’ personalized teaching, it will help students to internalize and consolidate the acquisition of knowledge.

4 Conclusion

Constructing an educational evaluation relationship between edge devices to ensure their interaction security has become an important research content in edge computing. Based on the DFS algorithm, an education evaluation model is proposed, which uses a directed weighted graph to construct the education evaluation relationship between nodes. In the path search process, a large number of irrelevant educational evaluation edges are eliminated through multiple constraints, and the accuracy of educational evaluation is improved. At the same time, the improved DFS algorithm is used to avoid loops and node detours, reducing the time for education evaluation aggregation, and effectively improving the computational efficiency of the education evaluation model. The KSP algorithm is introduced into the field of educational evaluation and evaluation, and constraints are added to the search process of education evaluation paths, so that the searched education evaluation paths have certain differences, and the problem of education evaluation dependence is solved. The use of algorithms makes the path search more directional and improves the search efficiency of the educational evaluation path. The experimental results show that the algorithm in this paper can improve the accuracy of education evaluation while ensuring the computational efficiency, and can ensure the load balance in the domain and improve the network survival time.

References

Zhang

, Zhang

and Yen

N.Y.

, Recommendation System of Micro-Blog Topic Based on User Clustering[J], Mobile networks & applications 22(2) (2017), 228–239.

Leuenberger

, Kocak

, Jordan

D.W.

, et al., Medical Physics: Quality and Safety in the Cloud[J], Health Physics 115(4) (2018), 512–522.

Noor

A.K.

, Intelligent Adaptive Cyber-Physical Ecosystem for Aerospace Engineering Education, Training, and Accelerated Workforce Development[J], Journal of Aerospace Engineering 24(4) (2011), 403–408.

Pecori

, A Virtual Learning Architecture Enhanced by Fog Computing and Big Data Streams[J], Future Internet 10(1) (2018), 4.

Nivala

, Simple answers for complex problems: education and ICT in Finnish information society strategies[J], Media, Culture & Society 31(3) (2009), 433–448.

Pecori

, A Virtual Learning Architecture Enhanced by Fog Computing and Big Data Streams[J], 10(1) (2018).

Lin

, Yang

, Pan

, et al., An Optimal Energy-Saving Strategy for Home Energy Management Systems with Bounded Customer Rationality[J], Future Internet 11(4) (2019), 88.

Zhou

, Ke

and Luo

, Multi-camera transfer GAN for person re-identification, J Vis Commun Image Represent 59 (2019), 393–400.

Fang

, Wang

and Ying

, A Novel Model for Mining Frequent Patterns Based on Embedded Granular Computing[J], International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems 26(05) (2018), 741–769.

10.

Moskowitz

E.J.

and Nash

D.B.

, Accreditation Council for Graduate Medical Education Competencies: Practice-Based Learning and Systems-Based Practice[J], American Journal of Medical Quality 22(5) (2007), 351–382.

11.

Panda

, Agrawal

, Samantaray

, et al., An evolutionary gray gradient algorithm for multilevel thresholding of brain MR images using soft computing techniques[J], Applied Soft Computing 50 (2017), 94–108.

12.

Blaise

, Knowledge management, organizational culture and Anglo-American higher education[J], Journal of Information Science 27(3) (2001), 129–137.

13.

Pedrotty

R.D.

, Mathematics Education and Students with Learning Disabilities[J], Journal of Learning Disabilities 30(1) (1997), 2–19.

14.

Blakely

E.J.

, A New Role for Education in Economic Development[J], Education and Urban Society 29(4) (1997), 509–523.

15.

Allegrante

J.P.

, SOPHE: At the Intersection of Education, Policy, and Science and Technology[J], Health Education & Behavior 26(4) (1999), 457–464.

16.

Englert

R.M.

, The University’s Role in the Improvement of Inner-City Education[J], Education and Urban Society 29(3) (1997), 373–402.

17.

Weintraub

and Ziejewski

, North American Brain Injury Society[J], Journal of Head Trauma Rehabilitation 35(2) (2020), 156–252.

18.

Canales

, Lee

and Cannesson

, Science Without Conscience Is but the Ruin of the Soul: The Ethics of Big Data and Artificial Intelligence in Perioperative Medicine[J], Anesthesia & Analgesia 130(5) (2020), 1234–1243.

19.

Hill

R.B.

and Somers

J.A.

, A Process for Initiating Change: Developing Technology Goals for a College of Education[J], Journal of Teacher Education 47(4) (1996), 300–306.

20.

Nolte

W.H.

, Guaranteed Student Success: General Education and Occupational Programs[J], Community College Review 19(1) (1991), 14–23.

21.

Bhandari

and Reddiboina

, Augmented intelligence: A synergy between man and the machine[J], Indian Journal of Urology 35(2) (2019), 89–91.

22.

Yannuzzi

, Siddiqui

M.S.

, Saellstroem

, et al., TEFIS: A single access point for conducting multifaceted experiments on heterogeneous test facilities[J], Computer Networks 63(Apr.22) (2014), 147–172.

23.

Lambert

C.U.

and Andrew

W.P.

, Computer Literacy in Hospitality Education[J], Cornell Hotel and Restaurant Administration Quarterly 25(2) (1984), 48–51.