ADVERTISEMENT

Preschool education optimization based on mobile edge computing under COVID‐19 – Wei – – Expert Systems

548
SHARES
2.5k
VIEWS


1 INTRODUCTION

The COVID-19 pandemic has swept the world over. Preschoolers are susceptible to COVID-19 resulting from their low immunity. In order to make sure the well being and security of preschoolers under common pandemic, kindergartens perform numerous on-line instructing actions, in order that preschoolers can be taught information at house, and make enough preparations for the formal faculty time period. The instructor communicates with preschoolers by way of on-line instructing, and preschoolers talk with instructor at house utilizing mobile telephones or different transportable gadgets.

In latest years, the continual growth of rising applied sciences, comparable to massive information (Pita Costa et al., 2020), cloud computing (Kaplan et al., 2020) and fog computing (Whaiduzzaman et al., 2020), which has introduced new alternatives for clever applied sciences. In the age of massive information, the Internet is producing nice quantity of knowledge day-after-day. While the web of the whole lot (IoE) brings nice comfort to folks, however it additionally has issues such because the scattered geographical location of knowledge, which convey nice challenges to information processing and information safety. Cloud computing offers a centralized and environment friendly platform for giant information processing, however the charge of community bandwidth development is much lower than the speed of knowledge development, and the growing community bandwidth will enhance extra {hardware} prices. At the identical time, the difficult community surroundings will result in information processing latency.

Edge computing (EC) works on a time-pushed mechanism, whereas cloud computing is used to course of information that’s not time-pushed. By deploying gadgets on the edge of the community close to the information supply, EC is feasible to supply content material caching, information storage, and web of issues administration to acquire higher response occasions. Edge nodes collaborate with cloud centres to assist actual-time information processing, agile connection and optimization of software efficiency, which altering the way in which of centralized storage and processing of knowledge in information centres in conventional community structure (Wang et al., 2020; Zhao et al., 2021). Some EC gadgets use clustering technique, which additionally modifications the way in which of processing of knowledge (Ma et al., 2021). EC makes use of the gateway, base station, server, clever terminal and different edge node gadgets close to the information supply to finish a part of the information storage and processing which initially carried out by the information centre. EC as a complement to cloud computing quite than changing cloud computing. Its information processing is carried out on the edge of the community nearer to the information supply, and information doesn’t must be uploaded to the cloud over the community, which reduces the affect of community bandwidth on information transmission. EC can’t exist with out cloud computing, and cloud computing centre offers EC with highly effective computing capability and big storage assist, whereas EC offers cloud computing centre with storage providers for big quantities of personal information. Cloud computing offers providers to customers in a centralized computing means which causes latency. With the event of 5G mobile community, mobile edge computing (MEC; Keivani et al., 2021; Bi et al., 2021; Sun et al., 2021) has attracted large consideration within the business. Compared with cloud computing, MEC servers are sometimes deployed on the edge of the community. They are nearer to the customers within the distance, which may cut back latency, enhance the response velocity, and relieve the community congestion. As a key expertise in 5G mobile community, MEC can present low latency service (Shu, 2020; Slapak et al., 2021). As proven in Figure 1, the normal MEC structure reveals that customers offload computing duties to the close to-finish MEC server, in order to get providers with low latency and enhance person expertise.

Architecture of conventional mobile edge computing (MEC)

MEC has the next traits. (i) Nodes are deployed on the edge of the community. (ii) MEC server will get information from numerous consumer gadgets which may analyse and course of it to facilitate the event of latest enterprise. (iii) EC gadgets are near the tip system, to allow them to relieve the community congestion. Computation-intensive and latency-delicate purposes are growing with the event of the IoE. For instance, within the interactive on-line instructing between preschoolers and academics, preschoolers are very delicate to latency, and even the smallest one might disrupt their very own studying tempo, so it’s not advisable to run such purposes on mobile gadgets.

With little doubt, COVID-19 has introduced nice challenges to China’s financial system, tradition and folks’s life. During the pandemic, college students, particularly preschoolers, which can’t go to high school. Preschoolers are firstly of education, which performs an vital position of their future growth. However, on-line education has apparent benefits in contrast with conventional offline education, which isn’t simply restricted by area. Teachers do not need to return to the scene, which additionally saves many prices and solves the issue of uneven distribution of conventional excessive-high quality education sources. The interactivity of preschool education is essential. Online education has the traits of huge bandwidth, lengthy hyperlink transmission and large protection. Due to the true-time and interactivity of on-line education, it additionally has excessive necessities for the low latency and anti-jitter capability of the community. Assuming that after the jitter occurs in learning, there can be issues comparable to type of sucks, audio and picture out of synchronization, and latency of interplay, that are undoubtedly challenges to preschoolers. A really brief interval of latency will trigger preschoolers upset and lack of focus resulting from their poor adaptabilities, which ensuing within the observe-up programs can’t be carried out easily. Preschoolers of 1 or a number of kindergartens could also be scattered everywhere in the metropolis. We have to deploy servers in main areas based on the proximity precept.

This paper investigates preschool education optimization based on MEC under COVID-19, the principle contributions of this paper are summarized as follows. (i) We examine the distributed process scheduling downside and mannequin. (ii) Stackelberg recreation strategy based on alternating path technique of multipliers (ADMM) is proposed to optimize preschool education.

The paper is structured as follows. Section 2 opinions the associated work. Stackelberg recreation strategy based on ADMM is proposed in Section 3. The experimental outcomes are proven in Sections 4 and 5 concludes this paper.

2 RELATED WORK

Some students have performed analysis on preschool education. In Ozdamli and Karagozlu (2018), authors decided the opinions of preschool academics on the usage of augmented actuality applied sciences in preschool science education. In Yue (2018), a balanced allocation technique of preschool education sources based on the coordinated growth of city and rural areas was proposed to supply a reference for the balanced allocation of preschool education sources and the sustainable growth of preschool education. In Li (2019), authors examined the present music curriculum reform of preschool education with an ecological perspective and proposed corresponding implementation methods, thus offering a brand new theoretical perspective and sensible strategy for the reform of preschool music instructing. In Maricic and Stamatovic (2017), given the truth that early arithmetic education was vital for additional studying of arithmetic, authors examined the results that preschool arithmetic education on growth of geometry ideas in youngsters. In Yildirim (2021), authors investigated the repercussions of the COVID-19 pandemic on preschool education and sought solutions to how preschool education was carried out, what sort of actions had been held, what sort of challenges wanted to be overcome, and what measures wanted to be taken to maintain preschool education.

MEC has attracted in depth consideration from students at house and overseas because it was proposed. The present analysis instructions primarily focus on process scheduling, content material caching, collaboration and so on. In phrases of process scheduling, in Zhou et al. (2021), authors introduced a workflow framework that captured the distinctive execution logic of sensing duties, then a staged scheme to decouple the unique scheduling downside was proposed. In Umber Saleem and Liu (2021), system-to-system cooperation based MEC was proposed to expedite the duty execution of mobile person by leveraging proximity-conscious process offloading. In Al-Habob et al. (2020), two algorithms had been proposed to unravel the scheduling downside which had been based on genetic algorithm and battle graph fashions respectively. In Chen, Zhang, and Yuan (2020), authors centered on collectively optimizing the choices to maximise the utility of the MEC system which accounted for each the computation throughput and the equity amongst completely different cells, by formulating a stochastic optimization downside topic to the constraints of queue stability and vitality finances. In Xie et al. (2020), authors proposed a peer-to-peer enhanced process scheduling framework to attenuate the typical process period in system-to-system enabled MEC community by collectively optimizing the duty scheduling determination and the computational useful resource allocation. In Jiang et al. (2020), authors proposed a hybrid deep studying based on-line offloading framework the place a big-scale path-loss fuzzy c-means algorithm was first proposed and used to foretell the optimum positions of floor automobiles and unmanned aerial automobiles. In Nath and Jingxian (2020), a brand new decentralized deep deterministic coverage gradient algorithm was developed to acquire the optimum designs for multi-cell MEC programs by leveraging on the cooperations amongst neighbouring MEC servers. In Huang et al. (2020), authors centered on the issues of dynamic process scheduling and useful resource administration in MEC surroundings with the precise goal of attaining the optimum income earned by edge service suppliers. In Xue et al. (2019), a joint optimization mannequin for process offloading and heterogeneous useful resource scheduling was proposed, contemplating the heterogeneous communication sources and computing sources, collectively minimizing the vitality consumption of person gear, process execution delay, and the cost.

There have been many methods on content material caching of MEC. In Liang et al. (2021), authors proposed a multi-head consideration based reputation prediction mannequin that thought of multi-dimensional options together with historical past and future reputation, social relationships, and geographic location to foretell content material reputation. In Zhang et al. (2020), authors proposed an built-in content material-centric mobile community framework for edge caching in 5G networks. In Zhang et al. (2018), authors proposed a brand new cooperative edge caching structure for 5G networks, the place MEC sources had been utilized for enhancing edge caching functionality. In Yu and Wu (2020), authors proposed a content material caching technique based on mobility prediction and joint person prefetching. In Liu et al. (2018), authors proposed a novel MEC enabled wi-fi blockchain framework the place the computation intensive mining duties might be offloaded to close by EC nodes and the cryptographic hashes of blocks might be cached within the MEC server. In Chen, Li, et al. (2020), Guo et al. (2019), Jianqiang et al. (2020), Li et al. (2019), Yang et al. (2021), recreation principle based in MEC have additionally been investigated.

3 TASK SCHEDULING CONSIDERING NODE SELFISHNESS IN MEC

Since the necessities of huge bandwidth, large protection, robust interplay, and low latency, on-line preschool education has the pure relation with EC. At first, EC can present providers on the edge of the community near the terminal. The node sources with full area protection appear to create an actual-time video forwarding community with top quality and low price masking a wider space. In the situation of on-line education, the shortening of area can cut back the latency and transmission time of varied forwarding and community gear processing within the complicated lengthy hyperlink transmission community, which may additionally keep away from the issue of disconnection and latency brought on by community jitter, in order that the low latency and powerful interactive expertise may be considerably improved. At the identical time, the processing of video providers with massive site visitors is accomplished on the edge, which is able to produce excessive concurrent entry within the on-line education. The distributed structure may successfully keep away from community congestion and cut back the bandwidth price of sending information again to the supply.

The MEC server is geographically nearer to the edge and preschoolers, which may present higher service for the preschoolers. For MEC server, its computing capability is way increased than that of EC gadgets. However, in contrast with cloud computing, the computing capability of EC gadgets are nonetheless restricted. Therefore, computing duties may be carried out in edge collaboration to enhance computing effectivity. In the method of MEC server and EC gadgets collaboration, the unique process scheduling strategy is not appropriate for on-line preschool education. Therefore, it is vitally vital to check the duty scheduling between EC gadgets and MEC server. At current, the analysis on distributed process scheduling principally considers the case of collaboration. EC gadgets can prolong the battery life as potential once they have selfishness, and they don’t seem to be keen to collaborate with MEC server. It is critical to think about not solely the variety of duties that the EC gadgets can collaborate with, but in addition whether or not the EC gadgets are keen to take part within the collaborative computing. The MEC server must schedule duties with a number of EC gadgets, which may be carried out on the MEC server facet or assisted by the EC gadgets.

3.1 Distributed process scheduling downside and mannequin

Task scheduling in MEC is a vital step earlier than computing. In order to realize extra environment friendly collaboration between MEC server and EC gadgets, it’s essential to design an efficient process scheduling technique. This paper research the duty scheduling downside in edge collaborative situation. The MEC server expects the EC gadgets to collaborate as many duties as potential, and it’s unknown whether or not the EC gadgets have the willingness to carry out duties for the MEC server. In on-line preschool education, mobile telephones and different gadgets in preschoolers are the EC gadgets. The price of collaborating with the MEC server will enhance when EC gadgets are egocentric, such because the vitality consumption of the gadgets. It is clearly that the EC gadgets are unwilling to collaborate with the MEC server, so the MEC server and the EC gadgets have inconsistent targets. When contemplating the duty scheduling of nodes selfishness, the right way to resolve the inconsistent targets between the MEC server and the EC gadgets is among the major analysis on this paper.

MEC server can solely carry out restricted computing duties, and there are a variety of accessible sources within the EC gadgets. MEC server collects accessible edge sources from mobile telephones and different gadgets in preschoolers for collaborative computing. The process scheduling diagram is proven in Figure 2, a MEC server with computing and storage capability is deployed close to the bottom station, and the EC gadgets have sure computing capability and information sources. In the method of process scheduling, the duty scheduling technique can decide the variety of duties that the EC gadgets can carry out for the MEC server. The MEC server and the EC gadgets collectively determine the ultimate scheduling coverage.

image

Task scheduling diagram

The major parameters used on this paper are proven in Table 1.

TABLE 1.
Symbols and implications
Symbol Implication
aok Data sizes supplied by EC system ok
ηok The weighting parameter of EC system ok to information measurement
ωok The weighting parameter of EC system ok to cost
Pok The cost consumed by EC system ok
T Interval
δok The technique of MEC server incentivizing EC system ok
C Data high quality rating
c Chip construction parameters
CRok Current frequency of EC system ok
ρ The computational density of the duty
i Iteration variety of the exterior loop
t Iteration variety of the interior loop
γ Penalty issue
λ Lagrange multiplier
  • Abbreviations: EC, edge computing; MEC, mobile edge computing.

In this paper, we think about the edge community with a MEC server and urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0001 EC gadgets from preschoolers. The variety of EC gadgets may be very massive in future, and one of these on-line preschool education could also be prolonged to a metropolis or perhaps a province. Since a lot of the connections between MEC server and wi-fi community entry factors are through optical fibre that the transmission latency is negligible. Only one process may be carried out by EC gadgets when aiding process scheduling. The EC gadgets is not going to carry out the duty for the MEC server earlier than there isn’t any efficient process scheduling technique. Considering the selfishness of EC gadgets, the MEC server will encourage the EC gadgets to carry out the duty as a lot as potential. There is a process queue on the MEC server finish, which represents the record of duties that the MEC server must carry out. There are processing unit and information unit on the EC gadgets finish, and the information unit offers the information wanted by the processing unit for the computing duties. The general carry out course of is outlined as follows. (i) One process within the process queue is chosen by the MEC server. (ii) Sending the parameters of the chosen process to the EC system. (iii) The scheduling technique is decided collectively by the EC gadgets and the MEC server. (iv) The EC gadgets and the MEC server carry out their duties in response to the obtained scheduling technique respectively. (v) Returning the efficiency outcomes of the EC gadgets.

The goal operate of the EC gadgets finish is the weigh between the duty and system vitality consumption, which is outlined as follows.

urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0007(2)

the place urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0008 and urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0009 are the burden parameters of EC system urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0010. The cost consumed by the EC system urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0011 may be expressed as urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0012, and the worth of urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0013 relies upon on the chip construction of EC gadgets (Asad et al., 2020; Huang & Chen, 2020). In mobile telephones and different gadgets utilized by preschoolers, c is often set as 0.5. Let urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0014, urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0015 denote the computing density of duties carried out by EC gadgets, and there’s the connection between urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0016 and urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0017, which is urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0018. Each EC system wants to satisfy its personal most frequency constraints.

urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0019(3)

the place urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0020 is the utmost frequency allowed by the EC system urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0021. The MEC server expects the information supplied by EC gadgets with top quality, due to this fact, it should meet sure constraints.

urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0022(4)

the place urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0023 is the information high quality worth evaluated by MEC server for EC system urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0024, and urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0025 is a continuing which represents the worth achieved for information supplied by all EC gadgets. Given the above, we have now a particular optimization aim.

urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0026(5)

It is assumed that urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0027 is the optimum answer of the MEC server goal operate, and urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0028 is the optimum answer of the EC system finish. The aim of the EC system is totally inconsistent with that of MEC server. Therefore, the duty scheduling course of is constructed as a Stackelberg recreation downside by introducing recreation principle. The MEC server is the chief within the mannequin, and the EC system is the follower within the mannequin. The MEC server offers an incentive technique to facilitate the collaboration of the EC gadgets. The collaboration between EC gadgets and MEC server determines what number of duties can be carried out on the EC gadgets. EC gadgets are unwilling to collaborate with the MEC server with out incentive mechanism. We use distributed optimization ADMM expertise to unravel the issue, and the next is Stackelberg recreation strategy based on ADMM.

The chief must make the related technique urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0029 in regards to the incentive parameters, and the follower urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0030 must make the duty scheduling technique urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0031. The iteration is usually used to unravel the optimization downside in earlier research. However, there are a variety of interactions between the chief and followers when the scale of the community could be very massive, and a variety of iterations are wanted to achieve a consensus. Therefore, it is vitally essential to design an affordable incentive operate urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0032. The followers replace the duty scheduling parameters, and the chief updates the inducement parameters. After the chief designs an incentive mechanism for the follower finish, Equation 5 is remodeled as follows.

urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0033(6)

the place urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0034 is the inducement mechanism parameter variable managed by the chief, that’s, the chief’s technique. urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0035 is the variable managed by the follower, that’s, the follower’s technique. The chief will alter its technique appropriately, let urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0036, after which the follower will resolve the sport downside on the follower finish. When urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0037, the follower selects the present optimum technique urn:x-wiley:02664720:media:exsy12922:exsy12922-math-0038, thus minimizing the target operate on the chief finish.