This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2017.Doi Number

FETCH: A deep Learning-based Fog Computing

and IoT Integrated Environment for Healthcare
Monitoring and Diagnosis
Parag Verma1, Rajeev Tiwari2, (SENIOR MEMBER, IEEE), WEI-CHIANG HONG3*,
(Senior Member, IEEE), and SHUCHI UPADHYAY4, and Yi-Hsuan Yeh3
1
Computer Science and Engineering, Chitkara University, Punjab 140401, INDIA
2Department of Systemics, University of Petroleum and Energy Studies, Bidholi via Prem Nagar, Dehradun (e-mail: [email protected])
3Department of Information Management, Asia Eastern University of Science and Technology, New Taipei 22046, Taiwan. (email: [email protected];
[email protected])
4Department of Allied Health Sciences, SoHS, University of Petroleum and Energy Studies, Dehradun (e-mail:[email protected])
Corresponding author: Wei-Chiang Hong (email: [email protected]).
This research is sponsored by the Ministry of Science and Technology, Taiwan (MOST 110-2410-H-161-001).

ABSTRACT These days cloud-based infrastructure is facing many challenges, out of which the major
issue is their syncing data before cutover and data migration. Due to the limited scalability in terms of
security concerns of cloud computing, the need for a centralized IoTs based environment has been
constrained to a limited extent. The sensitivity of device latency emerged during healthy systems such as
health monitoring, etc. is the main reason, because healthy systems require computing operations on high-
volume data. Fog computing provides an innovative solution to improve the performance of cloud
computing, providing the ability to take the necessary resources and those that are closer to the end-users.
Existing fog computing models retain several limitations, such as either considering result accuracy or
overestimating response time, but managing both together impairs system compatibility. FETCH is a
proposed framework that integrates with edge computing devices to work on deep learning technology and
automated monitoring and offers a highly useful framework for real-life health care systems such as heart
disease and more. The proposed Fog-enabled cloud computing framework uses FogBus, which
demonstrates utility in the form of consumption of power, network bandwidth, jitter, latency, process
execution time, and their accuracy as well.

INDEX TERMS Fog Computing, Edge Computing, Healthcare, Machine Learning (ML), Deep Learning
(DL), Internet of Things (IoT), Heart Disease Analysis

I. INTRODUCTION computing, and Edge computing have grown important due

In the digital technology era, the use of intelligent devices to their nature as warm-heartedness and ability to deliver
in daily routine is experiencing tremendous growth in all response properties that depend on the tracked target
sectors like medical, agriculture, etc. IoT-enabled devices applications.2] High volumetric storage capacity,
use their sensing technology to generate big data and then computation, and the reliable communication practices
transfer it via fog computing or cloud computing to provided by these emerging technology including edge
destinations on which decision-making capabilities can be gadgets, which encourage and improve mobility, safety
accurate by applying deep learning algorithms. However, concern, security, low scale latency, and bandwidth of the
fog computing, including cloud computing standards, has network so that fog computing seamlessly coordinates
emerged as the backbone of a cutting-edge economy that latency and continuous applications.[2]–[7] Currently,
uses the Internet to deliver services requested by users.[1] cloud computing structure similarly helps to generate the
These two fields seems significant. But due to the heavy applications standards, for example, through IoT devices,
delay in response time, cloud computing is certainly not a fog computing, edge computing, and big data support and
decent option for applications that need a continuous infrastructure.[8], [9] Routers, switches, computing nodes,
feedback scenario. Advancements in technology including and gateways are used by fog computing to offer the least
Big data handling with the Internet of Things (IoT), Fog

VOLUME XX, 2017 1

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

conceivable type of assistance with energy usage, network deactivation and the response time of the network.

VOLUME XX, 2017 1

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

Researcher K.C. Okafor[10] delivers a prediction algorithm expectation.[14] Likewise, some investment is required to
based on dynamic reliability that has the ability to accurately formulate this complex neural network and the data using for
estimate the average time to failure at the system level for the them must be isolated. Higher accuracy is required, the
CloudMesh architecture. The researcher focuses on the QoS organization is more refined, and higher expected time.[15],
Proactive Auto Scaling Algorithm (PASCQA), designed with a [16] This has been an important issue for medical services and
heuristic approach for CloudMesh Cyber-Physical Systems comparable IoT-based applications, where it is basic to get the
(CPS). This proposed algorithm is capable of reducing scaling results in a real-time scenario. As the computation at the edge
overheads, increasing the utilization of CPS resources, has an exceptionally favorable state of decreasing response
providing a highly flexible system for capturing user requests, time, it provides other directions to the test of incorporating
and reducing physical connections in terms of size. But still the edge computing into complex deep learning models with the
data delivery latency needs more attention in the context of ultimate goal that we have achieved with high accuracy results
high sensitive data like medical, etc. Also researcher Okafor et in real-time scenarios. One of the central points of this work is
al.[11] works on optimizing bandwidth with edge devices. The to overcome the issue and suggest a tangible platform that not
researcher extended the fog and cloud spine-leaf architecture only results in low latency using the edge computing-based
with the use of the Cisco Nexus platform to develop a scalalbe resources but in addition the use of the deep learning-based
IoT datacenter including a cost-effective processing, storage system to produce exceptionally precise results. These devices
and big data analysis environment. can give results of some work that has been done to perform a
Researcher Mutlag et al. [12] explores the difficulties of the calculation on edge computing-based devices near the patient
use of fog computing in the application domain of medical to reduce the resulting delivery time. A portion of this function
care that has been investigated and distinguished that lethargy actually depends on the simulations [4] and has not given a
and response time are the most important and difficult to single competent system. In this way, the medical care
streamline quality of services (QoS). Medical services are industry is expected to work towards this deficiency.
one of the surest application areas that require accurate and The traditional signs of heart issues are difficult to distinguish
ongoing results, and individuals have introduced fog and the patient needs an accomplished specialist to determine
computing in this field that promotes positive advancement in that a person has a heart issue. This is problematic to do for all
this technology. With the fog computing-based processing; intents and purposes due to the lack of experts as most nations
we move assets to adjacent customers and subsequently do not really consider the computer-based applications or
reduce inactivity and expand the measure of wellbeing framework as an option to identify heart issues with the
accordingly. Acquiring dizzying results suggests accelerated required accuracy and clarify capacity.[17], [18] Existing
movements for core heart patents. In any case, the quickness medical services frameworks, that are sent over the internet of
of the results is not sufficient with such sensitive data that we things technology and pre-configured devices that are suitable
cannot trade-off with the accuracy of the result. One approach for patient data processing to computing systems to the extent
to achieving high accuracy is by using the best in regular that results are communicated to customers within the cutoff
classroom exam programming that deep learning and their time. Several earlier works have attempted to use IoTs to
variance formulated on a vast dataset. Recently, deep estimate the medical conditions identified with the heart but
learning[13] has seen excellent development in areas cannot ascertain the accuracy required by the tougher
reaching from computer vision (CV) technology to the voice- guidelines of clinical generalization agencies. In the recent
based recognition system for the performance in general past few years, later advancement as deep learning has gained
language preparation, system forecasting, and mixed-work fame and also outperform experts in the accuracy of the
information settings, all from discourse acknowledgment. In location of coronary disease.[19], [20] The work plans to unify
addition, ensemble technology-based lelearning12] is used to deep learning in the medical services industry and IoT with
fetch the best multiple classifiers among the number of the hope that it will adopt a model that offers low inactivity
classifiers as per the target requirement. Among this one of and high precision to reduce the issue of absence of specialists
the ensemble, techniques are named bagging classifier, where in clinical normalization promotes organization. There does
the evaluator is on an arbitrary subset of the fit strains the not exist much such work, which means unifying these two
base classifier information and later expresses his or her parameters [21], although no one uses the idea of edge
personal expectations on the ballot or, on average, to obtain a computing to improve precision using in-depth study models.
final forecast. Such evaluators help minimize the change We present a more thorough correlation in section 2 and
when contrasted with a solitary evaluator by bringing subsection 7.9. Furthermore, the enhancement of the deep
randomness to the strategy of dataset dispersion. One other learning model is a specific extension to allow outcomes to be
advancement of deep learning technology is forecasting and ensured because it requires precise correction and a cautious
classifying the information from medical services with balance of lethargy enhancement to offer the most desired
surprisingly high accuracy.[13] As it may be, in the present support quality. Furthermore, expanding on previous works
scenario deep learning models for medical care applications such as [2], [21], [22], FETCH (Fog Enabled Technique for
are exceptionally sophisticated and required countless Clinical Healthcare) gives a design for medical care
computational properties for both preparation and computation that incorporates mixed backend structures such
VOLUME XX, 2017 1

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

as FogBus [23] and Aneka [24], which makes it an optimal architecture and front-end design of the proposed FETCH
model. model are described in detail in the next sections 4 and 5. The
Earlier works elaborate that there are two important types of various essential implementation modules of the model are
medical care information classification schemes for cardiac briefly covered under Section 6. The next section 7 deals with
patients using a variety of devices (for example, IoT-enabled the model performance evaluation part of the proposed model.
sensors and recorded input data). The first short length of The conclusion and future directions of the research work are
data is handled at the center of the fog node and the second is included in the final section 8.
the high volume data prepared in the cloud data center
(CDC).[1], [8] Medical services receive quiet data by the II. LITERATURE REVIEW
organization at high speeds (configured as 250 MB for each For the effective processing of healthcare data in the medical
moment or upgraded).[1] Existing structures are not flexible domain, fog computing provides an important paradigm, that
enough to capture and result in two types of information can be fetched from various IoT-enabled equipment. The fog
situations and thus need to use edge and cloud computing computing enabled devices or fog node for gauging gauze has
enabled devices to take an application with such high volume the ability to deal with the heart patients' data, greatly reducing
data. After classification and collation of information from the latency, delay, or response time as edge computing enabled
curious devices of IoT-enabled networks, information is devices are much closer to the IoT enabled devices to compare
stored away and handled by nervous nodes or cloud workers. to the cloud-based data centers.
To deliver productive data administration to heart patients An IoT-enabled medical services framework was proposed
and various clients in need of ongoing results, a corporate by the researcher He et al.[5]. The researcher introduced a board
edge to convey medical services and other haziness to model called FogCapCare that incorporated a cloud layer with a
delicate results with short response time, minimal energy sensor layer to detect patented heart health conditions and
usage, and high accuracy- Fog computing including cloud- shorten the execution season for runtime readiness. FogCepCare
based calculation model required. The absence of such uses a parceling and grouping approach and correspondence and
models or structures that incorporate the force of high similar handling strategy to advance exaction time.
precision of the deep learning model, simultaneously with the FogCepCare’s presentation is in contrast and the current model
low inactivity of the edge computing-based nodes prompted uses a reproducible cloud climate and streamlines exaction time,
this work. although it falls short on performance evaluation as far as
In the current work, we propose to FETCH: A Fog Enabled significant quality of services limitations, for example, power
Technique for Clinical Healthcare system which is capable usage, inactivity, accuracy, and so on.
for in-depth study and automated diagnosis of heart disease Another IoT-based e-health monitoring services framework is
including IoT and Deep Leaning ensemble technique. proposed by Researcher Ali & Ghazal[25]. This framework
FETCH gives medical care in the form of a mild mist relies on the use of the software-defined network (SDN), which
administration, additionally, it is effectively related to the is capable of collecting information via cell phone as a voice
data of cardiac patients collated from various IoT-enabled control system, enabling the health status of patients.
devices. FETCH provides this kind of support by celebrative An Autonomous-Monitoring-System (AMS) model proposed
using of FogBus system[23] and demonstrates this by the researcher Rajasekaran et al. [26] is particularly used for
application and designing to use the mist property to the Internet of Medical Things (IoMT). This model is capable to
accomplish the same. provide functionality for medical care. This research work is
Some major contributions of this research paper are listed as; designed based on a reward mechanism that uses the Analytics-
• Propose a deep learning ensemble technique for the Hierarchy-Process (AHP) for the proper dispersion of the flow
advancement of fog computing that supports a of energy between the considered nodes in the model. Cloud
generic framework design computing enabled simulated framework has been used to
• Built-up a lightweight computing technique for heart simulate and test the presence of the autonomous monitoring
patient diagnosis framework with the use of a deep system as far as energy use and the autonomous monitoring
learning ensemble technique named FETCH. This system are better in caparison to the FGCS strategy, but the
system is deployed with the utilization of the timing of the correspondence between nodes reflects the high
FogBus system to reconcile IoT-Edge-Cloud in a latency of patient solicitation preparation.
single unit for analysis the real-time data processing. A SMART-For-Gateway (SFG) model was proposed by
• The deployment of FETCH is investigating various researcher Constant et al.[27] to provides a potential filter, an
performance measurements such as shown and intelligent filtering system, deep examination, and specific
accurate execution, response time, rate of data information exchanges during the IoT-enabled devices
transmission of the network, and the usage of energy. (wearable) and a system of data modeling in SFG. The proposed
The overall research is summarized into 8 sections, where model streamlines the exposition about the execution of the
section 2 presents the previous related work in the same time and the usage of energy in the system as well, although it
direction. Research section 3 highlights the technical doesn’t think about the performance parameters of the system
background of the work and its importance. The system i.e. latency.
VOLUME XX, 2017 1

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

Azimi et al.[21] to explore the efficiency of Convolutional environment the use of FHIR delineation in the proposed
Neural Network (CNN) based models in conjunction with model provides the capability of accurately analyzing the
classification techniques for research for Hierarchical-Edge- multiple clinical opportunities from different focuses without
Based-Deep-Learning (HEDL) proposed for medical harmonizing the site-explicit data. In addition, the proposed
intensive as an illustration of a strategy for medical-care approach has been approved to use the clinical focus data in
despite the IoT enabled framework. Rajkumar et al.[28] the form of electronic health records (EHR) of two US
proposed a deep learning model with high scalability and scholars, with 216,221 adult patients in the hospital at any rate
accuracy for records the electronic health data of patents 24 hours admitted and the accuracy of the expectation
dependents on Fast-Healthcare-Interoperability-Resources improves. Table 1 summarizes the comparison of the
(FHIR) design. In the deep learning-enabled strategic proposed mode FETCH with the available models.
TABLE 1
A FEATURES COMPARISON TABLE OF FETCH (PROPOSED MODEL) WITH THE PREVIOUS EXISTING MODELS AND FRAMEWORKS
Features
Heart Performance Parameters
Disea
Ense Train Testi
Proposed Fog I Deep se Power
Authored by mble Late Jit ing ng
Models Comp o Lerar Predi Consu Executio Arbitrati Network
Lear ncy ter Accu Accu
uting T ning ction mption n Time on Time Bandwidth
ning racy racy
Syste
m
Gia et al.,
LCHM ✓ ✓
2017[29]
FogCepCa He et al.,
✓ ✓
re 2017[5]
IoT e-
Ali & Ghazal,
health ✓ ✓
2017 [25]
service
Akrivopoulos et
ECGH ✓ ✓ ✓
al., 2017[30]
Choi et al.,
GRAM ✓ ✓ ✓
2017[31]
Constant et al.,
SFG ✓ ✓ ✓ ✓
2017[27]
Azimi et al.,
HEDL ✓ ✓ ✓ ✓ ✓
2018[21]
Mahmud et al.,
FIH ✓ ✓ ✓ ✓
2018[23]
Barik et al.,
FogLearn ✓
2018[32]
Pham et al.,
CoSHE ✓
2018[33]
SLA- Sahoo et al.,
✓ ✓
HBDA 2018[34]
Rajasekaran et
AMS ✓ ✓
al., 2019[26]
Alam et al.,
EOTC ✓ ✓
2019[35]
Abdelmoneem et
CFBA ✓ ✓ ✓
al., 2019[36]
FETCH Proposed work ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓

An end-to-end scheme proposed by Moosavi et al.[22], to A Framework proposed by researcher Rahmani et al.,[2]
estimate the security for the IoT enabled devices under named Smart E-Health Gateway to use the required position
consideration of medical care, which uses the datagram of a gateway on the edge of a dedicated network to offer a
transport layer security (DTLS) handshake protocol, which is variety of support, e.g., over time, with the real data mining
separated without the need for reintegration at the physical process, data processing, data incorporation, and the local
layer of the devices- details established a secure data storage. In addition, the proposed framework fully
correspondence between different spectacular routes. The approves portable use to test IoT Enabled Warning Scores
proposed scheme results showed effective use in reducing (EWS). Expanding this work, our proposed model FETCH
communication overhead by 25% and latency below 16.00%. highlighted the option of using an in-depth learning model in
Expanding this work, our proposed work FETCH is expected an ensemble technique to increase the expected accuracy and
to express an all-around incentive real framework and to prove more accurate results for native heart patients.
medical care application at centers of fog computing-based The challenges that analyzed in the previous works[5],
nodes. [25]–[27], [29], [31], [37] needs to address the full potential
VOLUME XX, 2017 1

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

of IoT based computing technique for medical care

framework; IV. FETCH PROPOSED SYSTEM ARCHITECTURE
• requires a productive IoT based healthcare The FETCH model is a fog computing-enabled technology
applications, which can handle the information of a based on IoT ensemble cloud computing specifically
large number of cardiac patients with minimal designed for medical care services, which can adequately
energy use and short response time, deal with the data of cardiac patients and analyze the
• An efficient estate planning method for processing situation well to distinguish the severity of the coronary
conditions require that the client with the greatest disease. FETCH coordinates various instrumentation tools
asset use to execute the remaining burden to meet through the programming segments and permits organize to
the cutoff time of the outstanding burden and end the coherence of edge-fog-cloud for quick and accurate
• The deep learning ensemble technique to description of the results. Figure 1 presents the designs of the
continuously analyze the severity of coronary FETCH which consists of various tools and programming
disease in patients. parts directly depicted.

III. TECHNOLOGICAL BACKGROUND

FogBus provides a framework that enables the
development and the deployment as well, of the coordinated
fog computing and cloud computing-based environment
including the phase-free exaction of organized
correspondences and uses.[23] FogBus can connect with
individual IoT-enabled devices that can be medical care
sensors with the gateway devices to transmit the data and
ventures to specialist nodes. The management of the
resources and the initiated tasks are done at the fog
representative nodes. To guarantee the data integrity, security,
blockchain-based FogBus, user authentication, and
encryption technology that enhance the unbreakable quality Figure 1. System Architecture of FETCH
and the robustness nature of the fog environment. FogBus
uses the HTTP RESTFul API for correspondence and directs
A. HARDWARE COMPONENTS OF FETCH
a continuous fog computing arrangement with the cloud
The FETCH model includes the following device parts:
computing by the use of the Aneka Software Platform.[24]
Aneka provides a software platform that encourages the
provisioning of applications that work on event and cloud 1) NETWORK ON BODY-AREA-SENSOR (BAS)
computing.[24] Aneka gives developers an API to use virtual Three types of sensors establish this segment: medical
resources on the cloud. The center sections of the Anneka sensors, actions sensors, and environmental sensors. This part
stage are planned and executed in a service-oriented fashion. provides the functionality to detect the data of the heart
A dynamic provision is the ability to powerfully accept assets patient and carries it to the respective passage gadgets.
and coordinate existing foundations and programming
frameworks. In the best-known case, accepted resources are 2) APPLICATION GATEWAY
virtual machines (VMs) that are sourced from an The proposed framework used the three different types of
infrastructure-as-a-service (IaaS) cloud supplier.[38] Aneka devices for application gateway specifically consist for
consists of dynamic provisioning as a component of fabric mobile phones, laptops, and tablets, which is going as fog
Services, which provides provisioning administration to enabled devices to collect the information from the different
deliver the virtual nodes from the public cloud suppliers to sensors and this information forwarded to the
complement neighborhood assets. This is basically Brokers/Workers node for additional information.
accomplished because of the relationship between the two
administrations: Scheduling services and resource B. FogBus ENABLING MODULE
provisioning services. Currently, Aneka supports four diverse A FogBus enabling module that includes
programming models[24], [38]: the bag of assignment
models, the distributed strings model, the MapReduce model,
1) NODE FOR BROKER
and the parameter sweeping model.
This part receives the work demand or potential input
At FETCH, we utilize the bag of assignment model for
information from a device of the gateway. The solicitation
task appropriation in cloud-based virtual machines. FETCH
input module receives the work demands from a gateway
utilizes the FogBus to bridge the cloud assets and Aneka to
device that does not last long before information is
cloud-based resources.
transferred. The assertion module (part of the resource

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

manager in the delegate node) takes all of the specialist give ideal execution. FETCH provides the control to the
nodes' stake management information and selects which node users to set the arbitration plans dependent on their own load
or sub-node to progressively send to the businesses. balancing and applications requirements. The proposed
scheme is depicted as the flow chart in figure 2 and the
2) NODE FOR WORKER pseudo code are as follows;
This segment especially works to complete the ventures
assigned by the resource manager of the node of the broker. Set Threshold T: sample data workload W over E
The node of the worker may have devices and single-board (past data experience)
computers (SBCs) installed such as the Raspberry Pi Input Tn (working task),
processor. In the FETCH, worker nodes can be incorporate wL1,wWL2, ..., wLn (Worker loads)
modern in-depth learning models to measure information and 1 compute tw = min(wL1, wL2, ..., wLn)
create and measure results. 2 if tw>T, handover Tn to cloud (for further decision
process)
3) CLOUD BASED DATA CENTER (CDC) 3 else, handover Tn to arg min tw (for further
When the framework of fog computing becomes sample data computation)
overloaded, due to that the arability of the services follows
the latency or the size of the information is much larger than 3) DEEP LEARNING MODULE
the normal size, at which point FETCH organizations harness Deep learning enabled model for the proposed system uses
the resources of the cloud-based data center. the data to design a neural network focusing or group
information that includes vectors before handling
information received from the body area sensor network. In
C. SOFTWARE COMPONENTS OF FETCH MODEL light of the undertaking specified by the resource manager, it
The programming leading parts of the FETCH model additionally predicts and creates results for the information
include; received from the devices on the gateway.
1) PREPROCESSING AND REFINING THE COLLECTED 4) ENSEMBLE MODULE
DATA This module incorporated with the deep learning model to
The initial step after gathering the data from the input is boosting the expected results from various models and uses a
the preprocessing to refine the data as per the expectation for majority voting method to select the yield class that would be
the model. It involves the separation of information that uses required if the patient had coronary disease. Similar to the
information test evaluations. The discrete information is weighted average, the majority vote base aggregates the
reduced to a more trivial measurement that is performed outputs of the learners. However, instead of taking the
using set partitioning in hierarchical trees (SPIHT) algorithm, average of the probability results, the majority voting base
principal component analysis (PCA), and the encryption counts the learners' votes and predicts the final label as the
technique using singular value decomposition (SVD) to label with the most votes. Compared to the weighted average,
remove key segments of information vectors that affect the the majority vote is less biased towards a particular base
patients’ health status. learner's results because the number of majority votes
outweighs the effect. However, event dominance in the
2) MANAGEMENT OF THE RESOURCES ensemble model results from the favoring of a particular
It consists of two modules, the remaining burden head, and event by most of the same base learners or dependent base
the arbitration module.[24] The outstanding burden continues learners. The base learners are the sum of weak learners (wl)
to solicit the tasks and undertaking to prepare instructional computed as of equation (1);
information. It additionally handles large chunks of
information to be prepared. The interferences module resides
in the node of the broker and selects whether the fog This module resides in the FogBus hub which is returned
computing enabled node should be set to achieve the results, to the assignment and is responsible for disseminating
the actual broker, the fog expert node, or the cloud-based information and collecting results from other nodes of
data centers.[23] The primary objective is to separate the workers.
work into different resources to accommodate the stack and

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

and a large portion is picked up by the specialist node to

which the information was sent through the bagging process
(Slave node). FETCH system includes all the edge
computing enabled devices that include the gateway devices,
nodes of a broker, and working on a common LAN. The
resource manager resides in the programming segment
broker node and as a result, the gateway of broker job
demands devices in a worker cloud manner that send work
demands to it. The discretion result obtained from the
resource manager is obtained by a gateway device that
provides information about where to send the data. Three
situations emerge here:
• Broker handling information to a node of worker
• Another node of worker for sending information
• Data processing based on the CDC (Cloud-based
Data Center)
Depending on the situation, the gateway device may send
information directly to the node of the worker or the broker
(with or without the use of a cloud-enabled environment).
Figure 2: FETCH based resource scheduling The agent can allow the enumerated administration to send
data only if it has enough devices and in addition, the worker
node is overloaded. If the information has to be sent to the
D. THE TOPOLOGY OF FETCH SYSTEM
All the major segments of the proposed FETCH system cloud, it passes through a broker node because the gateway
have been already described in the previous section including cannot access the VPN that has a cloud-based virtual
their required and processing amount of data, and control machine available. Additionally, workers sometimes send the
signals. In addition, correspondence should be considered broker a parcel of heartbeat to demonstrate that they are alive.
under fault tolerance level and in persistent mode. This entire The therapy data packet encapsulates a load of information in
situation considered, with a selection of segments and the the same way that it is used by resource managers for load
organization structure is presented in figure 1. Data balancing.
transmission across all the devices on the edge is encouraged
using the FogBus[23] and Aneka with the cloud-based virtual E. COMMUNICATION ARRANGEMENTS
machine (CVM).[24] In the FETCH system, all the hardware equipment
depends on the predefined convections as depicted in figure 3
Network topology covered under the FETCH system for three different conditions i.e. Brokers node, Worker node,
follows the hierarchy of master-slave style where the node of and Cloud Center that have been described in the previous
brokers acts as a worker to analyze and control the load. section. In each case, the gateway initially sends a job
Under consideration of the non-cloud case, the device under solicitation to the broker node. Depending on the situation,
gateway worker/broker sends work to the instance input the broker sends either the worker IP address (of a similar
information for the probe to run, at that time which runs the LAN) or the master IP address (with or without a cloud-
preprocessing, predication model and the computed results based environment) to the gateway nodes. In the case of a
send back to the gateway devices. In the case of cloud broker, the broker node can check the worker's load. On the
forwarding, since the gateway devices cannot be on the off chance that all workers have a substantial load or are all
virtual private network (VPN), it sends the information to the low and the cloud has been erased, the broker at that time
node of brokers and then brokers forward the collected data sends the gateway device without sending its IP to the cloud.
to the CDC. It likewise guarantees that the IoT sensors and If the workers there are not stacked as fast, the broker at the
the gateway devices are secure from the malicious elements point sends the IP address of the most un-stacked worker hub
and programmers as they cannot be connected to the web- to the gateway device. Expanding the number of workers will
based platform, but only local area networks (LANs) can be increase the time of discretion as the investigation of
connected with the other fog nodes. Because the cloud-based overloaded should end. Under consideration of the non-cloud
environment is capable to access a large set of resources, the case, the gateway device sends a task to the nodes of
timing of the execution is dependent on both the broker’s broker/worker for example input information for the exam,
node and the CDC having higher correspondence overhead which at that time runs the preprocessing, prediction model,
and less laziness due to the latency delay. At this point when and sends back the computed results to the gateway devices.
the organization is empowered, the information received by In the case of cloud sending, since the gateway device cannot
the nodes of broker/worker is routed to any remaining edge be on the VPN, it sends information to the broker node which

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

then sends it to the CDC. Likewise, it guarantees that the IoT The application-specific domain information is required to be
sensors and passage gadgets are separated from hazardous taken care of in the framework. Normalizing age information
materials and programmers as they may not yet be connected as it was demonstrated in figure 4. Essentially, the data of
with the internet and have LANs with other fog computing- rest blood pressure is similarly slow and patients with
based nodes. Because the cloud-based environment is coronary disease indicated more severe hypertension with
capable to access a large set of resources, the timing of the patients not having the coronary disease. In the same way
execution is dependent on both the broker’s node and the that understanding cholesterol levels reflects some objective
CDC having higher correspondence overhead and less clear behavior, good health patient’s communication is
laziness due to the latency delay. At this point when the leptokurtic. In fact, even with the most extreme pulse, all
ensemble is empowered, the information received by the patients who are considered under healthy status have
worker/ broker node is routed to any remaining edge and the significantly high pulse rate i.e. approximately 160 in
greater part is picked up by the worker node, using the data comparison with those patients having heart disease i.e.
sent through the bagging technique. approximately 150. Various highlights such as chest pain and
fasting glucose should be constantly changed from virtue to
absolute virtue. Similarly, the peak exercise ST section and
the cardiac position recovered from the thallium test.

Figure 3 FETCH based communication sequence arrangements

V. PROPOSED DESIGN OF FETCH SYSTEM

The model of fog computing that is already depicted in the
Figure 4: Age distribution of Heart patients in the Database including,
previous section considered the data of heart patients as a
(a) all considered patients in the dataset, (b) patients having no heart
contribution from the sensor and the computed results send
disease, and (c) patients having heart disease.
back to the data center in which the patient has the coronary
disease, with case certainty. It is executed with parts that
include information of preprocessing modules, ensemble B. APPLICATION OF DEEP LEARNING ENSEMBLE
deep learning modules, and the interface for gateway TECHNIQUE
depicted directly. A deep neural network (DNN) ensemble technique has
been used in this model for predictive analysis and our
A. DATA PREPROCESSING PHASE OF HEART PATIENT
thoughtful application. The model is used to solve the
Information from basic pule-oximeter or the devices of
purpose of binary classification problems. The first model
ECG is in a plain graphical organization and one must be
was formulated on the patient's heart continuous data from
prepared in advance to find the estimates for several
the Cleveland dataset with known output classes and then the
highlights of contributions to deep learning models.[39], [40]
corresponding training model is used to estimate the results

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

of the continuous information contribution as shown in C. COMMUNICATION MODE WITH ANDROID

Figure 5. INTERFACE
FastHeartTest and android executable interface have been
used to send the data to the broker or worker nodes in the
gateway devices. The appearance of the application interface
is shown in figure 6. This appearance allows the gateway to
act as an intermediary between the body sensor network and
the node that works as a worker. Correspondence is
accomplished using the HTTP RESTful APIs. We used the
HTTP Post method to try and download the result
information from the gateway devices. Every broker node,
worker node, and the CDC include a pre-training exercise
handily by the deep learning model and preprocessing
programming model.

Figure 5. FETCH model training and testing phase

The dataset is split into the ratio 7:2:1 with training, testing,
and validation respectively. The 70% ratio of the dataset is
used as the training dataset used to prepare the model, the
10% ratio of the dataset used as validation dataset that is used
to set the model, and the 20% ratio of the dataset used as the
testing dataset that is used to test how the model performs on
new information. The training model can be placed in all
nodes previously fitted for handling by placing them away in
a specific dataset. Another methodology may be used to train
the model independently, which focuses on datasets drawn in
a different model. In preparation for propagation, information
dissemination promotes processes, which test hastily
information from the dataset with replacement and send it to
different edge hubs to produce a singular model.[41] During
the time of diagnosis, at any point a hub has resumed an
undertone; it receives patient information that is a vector of
shape. This information is taken into account t as a
contribution to the model, making a forward pass on to the
DNN, for the binary output i.e. either 1 or 0 that means either Figure 6. FETCH model Gateway of a user interface for both negative

the patient has a coronary disease or not respectively. At the and positive patient heart-related issues

time of diagnosis, we are using the bagging process-an an

ensemble technique to join the result of different models to VI. MODEL IMPLEMENTATION MODULE
produce results with high accuracy. The worker gets the All the parts referenced in section 5 were executed in
information, which is handed over to other worker nodes. At different programming languages. The preprocessing and the
that point, every worker adds it to their line and the predicted deep learning with ensemble technique segments of the
results of every node of worker are sent back to the worker implementation were realigned to use a python programming
allowed for this assigned task. At that time the major part environment. The preprocessing module standardizes the
expectation class is sent from bagging to the devices of the information dependent on the min and max estimated value
gateway. FETCH model provides the functionality to the user of the area boundaries in datasets and their spread.
to cripple this element when the required results are inertia The deep learning with ensemble technique application
root. In the resulting section 7, we showed the performance enabled using the scikit learn – Machine learning Library in
of the deep learning model with ensemble technique that python environment.[42] Bagging classifier a method of
leads to better improvement and high response time with the scikit learn library has been used to implement the voting
network overheads. scheme in our proposed model. The model base takes the
basis of the classifier which is a deep learning network for
the considered situation and the number of classifiers used as

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

input. Currently, the model haphazardly transmits the network and the energy overhead shows that the proposed
information among the classifiers to prepare the model. In the model is profitable and has low overheads.
time it accepts all the expected classes as input and the
predicted output of the major parts. The following parameters A. AN EXPERIMENTAL SETUP DETAILS
of the best base model on the considered dataset after tuning The experimental setups for evaluation of the proposed
to the next: system FETCH and configuration of the tools are depicted as
Input layer Size 13 # features of the considered
dataset follows;
Output layer Size 2 # 0 or 1 i.e. binary classifier • Gateway Devices: LETV S2 4G enabled with
#hidden layer 3 Android 10.
FC: Fully connected layer including • Nodes of Broker (Master node): Lenovo Yoga
20 nodes, 20 nodes and 10 nodes
ThinkPad X2 with Intel i7-250 GHz processing unit,
respectively
Model Optimizer’s "Adam" including 16GB RAM and 240 SSD ROM. The
Model Activation "ReLU" machine was configured with 64-bit professional
function windows 10. For the deployment toolkit, Apache
Learning rate of the 0.0001 HTTP server 2.4.34 has been used successfully.
model • Worker (slave node) nodes: Raspbian stretch
Operating System with Raspberry Pi 3B+ (μP),
The application of the Android interface was built using ARM Cortex SoC processing unit with 1.4GHz,
MIT's app inventor that establishes a communication link including 1GB LP-DDR2 SDRAM and a wifi unit
with the FogBus node of the broker. The Android interface (IEEE 802.11) with deployment toolkit apache
holds the information credit formatted documents in comma- HTTP server 2.4.34 have been used.
separated values (.csv) and transfers it to a broker node in the • Public Based Cloud: Microsoft Azure B1s Machine
Data Catalog module using HTTP POST. including 1v-CPU, 1GB-RAM, 2GB-SSD,
Arbitration modules are assigned to the node of brokers Windows Server 2016.
additionally that selects which worker node to perform the Figure 8 shows the actual performance of the proposed
task. This selection cycle of the worker node is performed system model. During the analysis process, specific data
according to the default FogBus strategy of selecting the parameters are considered and the data was recorded using
workers with the least CPU load. Any worker node which is the Microsoft Performance Monitor on the master and Azure
selected, a comma-separated value formatted record is sent enabled on virtual machines, although NMON performance
for examination. The execution interface module in each monitors are used in Raspberry Pi IC (integrated circuit)
expert receives information and initiates the deep learning (Source: Microsoft Windows performance toolkit.
ensemble technique code to check the information. The https://docs.microsoft.com/en- us/windows-
returned results are sent back to the worker/broker node that hardware/test/wpt/, SplunkBase performance monitor.
sent the information documents. The results are ensured to https://splunkbase.splunk.com/app/1753/). Microsoft
use the bagging technique and sent to the gateway decides. Network Monitor 3.4 was used in the broker's (Source:
A depiction of the various moduli and their collaboration Microsoft Network Monitor 3.4.
is shown in figure 7. https://www.microsoft.com/en-
au/download/details.aspx?id=4865) nodes and Raspberry Pi
in VnStat (Source: vnStat Network monitoring tool.
https://humdi.net/vnstat/) was used to use organization
transmission capability.

Figure 7. Various modules of the FETCH model

VII. MODEL PREFORMATION EVALUATION

The proposed FETCH model is demonstrated to check its
availability; feasibility and efficiency with executed,
communicated, and deployed the model on an actual Fog
enabled system of devices using the FogBus framework.[23]
The model has been used for a true use of identifying heart
issues for patients in the use of state-of-the-art deep learning Figure 8. The working environment of the FETCH model
strategies using an environment of fog-based learning. We
have dissected the precision and the response time with the

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

B. DATASET DESCRIPTION 5 53 1 0 140 203 1 0 155 1 3.1 0 0 3 0

To perform the experiments for computing the test results, 6 63 1 3 145 233 1 0 150 0 2.3 0 0 1 1
7 37 1 2 130 250 0 1 187 0 3.5 0 0 2 1
we considered the information from cardiac patients to
8 41 0 1 130 204 0 0 172 0 1.4 2 0 2 1
search for the presence of coronary disease in the patient[39], 9 56 1 1 120 236 0 1 178 0 0.8 2 0 2 1
[40], [43], [44], having a binary-valued 1 (presence of heart 10 57 0 0 120 354 0 1 163 1 0.6 2 0 2 1
disease) or 0 (no presence). The dataset Cleveland (Source:
https://data.world/uci/heart-disease) [43] is used to lead the
C. EXPERIMENT SYSTEM CHARACTERISTICS
analysis that was formed by Andras Janosi, Doctor of The dataset described in the previous section is used here,
Medicine at the Gottsegen Hungarian Institute of Cardiology, testing the model of whether a patient has a coronary disease
Hungary, and others. Patient’s names and their patient or not, and based on an estimate of the threshold set for each
numbers are classified. 14 relatively important details have patient. For complete information, the dataset was divided
been used in this research work for the classification of the into two-three different datasets, namely training, validation,
patients. These considered 14 relatively details about the and testing with ratios of 70%, 10%, and 20%, respectively.
status of the patient have been detailed in table 2 and The initial section was used to prepare the model, the second
exampled in table 3: for approval, and to tweak the model parameters. The
TABLE 2
DATASET DESCRIPTION OF 10 HEART PATIENTS previous part was used to test the model exposition. To
S. No. Parameter Description determine the amount of detail of the FETCH model, this
1 Age in years merit was noted.
Two values
2 Sex The accuracy of the model prediction: 1807 models
(1=male; 0=Female)
Chest pain type consist of the dataset, of which 1355 were used to train the
1. Typical Angina models and 452 were uses to testing the accuracy of the
3 Cp 2. Non Typical Angina model. The testing of the model was divided equally among
3. Non Angina Pain all the nodes of broker/worker to acquire their specially
4. Asymptomatic
Resting blood pressure (in mm Hg on tailored deep learning model. All hubs will need to be
4 trestbps transmitted to use all assets to produce a dataset model as a
admission to the hospital)
5 chol Serum cholesterol in mg/dl quantity of fog computing nodes. This reduces the
Fasting blood sugar > 120 mg/dl preparation time plus the accuracy of the test. To notice such
6 fbs
1 = True; 0 = False
Resting electro cardio graphic results effects, the accuracy of the preparation and testing was
0: normal investigated. We characterize absolute as the level of all
1: having ST-T wave abnormality (T official absolute patients, for which the model effectively
wave inversion and/or ST elevation or predicts if they have coronary disease. We think of
7 restecg
depression of > 0.05 mV)
2: showing probable or definite left improvement of various blur settings, by changing the
ventricular hypertrophy b Estes’ number of edge nodes and with or without ensuring the
criteria results.
8 thalach Maximum heart rate achieved
Exercise include angina ( 1= Yes; 0 =
Characteristics of time: Subsets that represent the specific
9 exang timing thresholds that appeared in figure 3 were additionally
No)
10 oldpeak
St depression induced by exercise observed and considered. These include interference time,
relative to rest latency, execution time, and jitter. We are comparing these
The slope of the peak exercise ST
11 slope segment timing parameters for various fog computing settings that
1. up sloping; 2. flat; 3. Down sloping have no node of edge computing or up to 2 nodes of edges
12 ca
Number of major vessels (0-3) colored (with or without ensemble) or a cloud having a calculation
by flourosopy framework.
3 = normal; 6 = fixed defect; 7 =
13 thal Range of bandwidth usage by the Network: For example
reversible defect
Diagnosis of heart disease as a node of broker, worker, or cloud and the number of
(angiographic disease status) worker nodes influence the network usage that was focused
14 Target (num) 0. <50% diameter narrowing
on exploring the use of the network in various cases. Like the
1. >50% diameter narrowing
(in any major vessel) analysis of deadlines, we analyze the network transfer speed
usage for diverse fog computing scenarios. It was never
TABLE 3 really out of the dependence of data transmission used with
SAMPLE OF CLEVELAND DATASET OF WITH BINARY TARGET VALUES OF the various fog computing configurations that FETCH
HEART DISEASE PATIENT
provides.
s.no age sex cp trestbps chol fbs restecg thalach exang oldpeak slope ca thal target
1 67 1 0 160 286 0 0 108 1 1.5 1 3 2 0 The system power consumption: Energy is an important
2 67 1 0 120 229 0 0 129 1 2.6 1 2 3 0 motive behind moving from cloudy to fog domain, in
3 62 0 0 140 268 0 0 160 0 3.6 0 2 2 0 addition to this we also considered the use of force in various
4 63 1 0 130 254 0 0 147 0 1.4 1 1 3 0
situations. In light of the force, use is considered and before

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

drawing separate tests we examine how a unique FETCH creates two possibilities P0: no heart disease probability and
design can be used for different users and application P1: heart disease probability, an ultimate disease that P0 + P1
prerequisites. = 1 with. An expectation (P0, P1) is measured as 100 (2 max
P (P0, P1) 1) and thus has a range of [0,100]. In this way, on
D. ACCURACY OF THE PREDICTED MODEL
the prediction that the probability is (0.5, 0.5) the certainty at
Variation in the accuracy of the training module with the the confidence level is 0 and when they are (0.9, 0.1) the
different nodes of edge computing (brokers nodes other than predictor class is 0% and the level of certainty is 80%. The
worker nodes) is shown in figure 9. We can see that the confidence variance shown in figure 11 represents the binary
accuracy of the training steadily increases as the number of classifier for the total test dataset, the subset on which the
workers increases. It is on this basis that each node learns a model made an accurate estimate, and where the prediction
model for the information it receives, and as the number of was inaccurate.
nodes increases, the number of models obtained by each We see that the certainty is higher for information where
node decreases, and therefore the model for different epochs the focus was correct, where predictability was in contrast to
over fitted the samples and hence the accuracy of the training information where the expectation was false. The greatest
increases. In figure 10 we show the variation of test certainty with which the model predicts 49.7% by mistake is
information precision as the quantity of edge computing likely to be closed in such a way that the certainty is less than
nodes increments. As per the expectation, the accuracy of the 50%, at which time our model advises the patient’s consult
test decreased with a greater number of nodes because each because the prognosis can be problematic.
Figure 11. Confidence level of the model with a different set of
node gets a more modest subset of the training dataset and
samples of the dataset
therefore it cannot sum to the model. Another assumption is
that adapting ensemble learning gives consistently better
performance without the ensemble learning-best or average F. ANALYSIS OF THE TIMING CHARACTERISTIC OF
case. THE MODEL
Variation in the interference time of the model broker node

Figure 9. Variation in training actuary with number of edge nodes for different fog calculations conditions:
• Specifically for the broker,
• Individual worker nodes,
• Two worker nodes, and
• The cloud environment is shown in figure 12.
We have seen that when the assignment is to be sent
directly to the broker or master's node or to interfere with the
cloud the time is almost negligible which is approximately
being 115ms. As the amount of edge computing nodes
increments, the broker nodes need to check the load on each
worker node and find out the base load of the individual
working person, later interference time increases because of
edge computing. The amount of nodes increases. At this
Figure 10. Variation in test actuary with number of edge nodes point when the information is passed to labor nodes to
assemble learning, the broker does not need to examine any
stack at that point, as the decision for most classes is
E. CONFIDENCE LEVEL FOR PREDICTION MODEL
terminated by one of the worker nodes, Therefore, arbitration
At whatever point the prediction model with deep learning
is like time without ensuring the case.
techniques is enabled, if the patient has coronary disease, it

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

The variation of the latency shows in figure 13, which is

according to figure 3, is the extension of the corresponding
time and the latency delay. We can observe that if the task is
sent to the broker or any of the edge computing nodes, the
inactivity at that point is approximately equal to all
correspondence that occurs via single-node data transfers.
Among the ensemble cast, the value of latency is a bit higher.
Under considering the cloud-based system, latency is more
due to the multi-node transmission of the data outside the
LAN.

Figure 14. Results of Jitter (in milliseconds) with different fog

computing scenarios

Figure 12. Results of arbitration time (in milliseconds) with different

fog computing scenarios

Figure 15. Results of Execution time (in milliseconds) with different

fog computing scenarios
The variation in the execution time has been shown in
figure 15. As per the expectation, the execution time is
shorter in considered cloud systems due to higher asset
access. Representative execution time is shorter than expect
execution because workers in het FETCH system are
Raspberry Pi enabled with fewer clock repetition processors.
Similarly, when the prediction used in the ensemble
technique is stronger than the time of execution, the worker
Figure 13. Results of Latency (in milliseconds) with different fog node now needs to test which class is larger among all the
computing scenarios considered predicted classes.

The jitter value represents the variation of response time G. CHARACTERISTICS OF THE NETWORK
for consecutive task demands. This is a basic parameter for BANDWIDTH USAGE
most continuous applications, including test information. The variations of network bandwidth usage of all the
Figure 14 guided the variation of jitter with the configuration considered edge computing nodes under different conditions
of fog computing. After observing the results we identified are shown in figure 16. We see that increasing the number of
that the jitter is higher for the broker than the case in which worker nodes requires heartbeat data packets, similar to the
the undertaking sends data to labor nodes. It is on this basis network usage expert. Security checks and data transmission
that discretion in various functions, management of resources, tricks (including cloud) are highly essential. The most
and security checks are additionally carried out by the broker. notable case when ensemble technique is considered is the
As the number of worker nodes increases, there is some use of network transmission capability in gathering
increase for the two edge computing nodes as opposed to information, as it is sent to all workers' nodes.
single edge computing nodes due to differences in the heap
of workers. Jitter is additionally high in the condition of the
considered ensemble. When the CDR is turned off, the jitter
response is very high.

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

such models and deliver lower disease detection system with

accuracy. It is decisive in basic medical care applications,
particularly for people concerned about heart-related issues
such as heart attack, heart stokes, or arrhythmia disease to
deliver low sleepiness and deeply accurate results. In
addition, deep learning that uses in the previous works by[23],
[39], [42] utilizes the ensemble technique for similar
computation and fundamentally high accuracy, using the
strategies that ensure surprisingly superior results. As
mentioned in the demonstration section 7.4, with the
ensemble technique, the accuracy rate of the prediction is
Figure 16. Results of consumption of network bandwidth (in kbps) increased by 17% with the considered case including 5 edge
with different fog computing scenarios nodes, which cannot use exiting framework i.e. deep learning
with ensemble technique provided.
In addition, the dissatisfaction with the earlier work, the
H. THE SIGN OF POWER OF THE SYSTEM
FETCH utilizes the FogBus system[25] to present a hybrid
In addition, we tried that the FETCH system is using
design with various accuracy rates, response times, network
energy consumption properties in various conditions. The
characteristics, and the consumption of power characteristics
CDC’s force use is in high contrast with the Broker node (as
as well. In light of different applications and user
the laptop in our considered setup) or the worker node (as
requirements, the use of different setups can be illustrated in
Raspberry Pi microprocessor in our considered case). This
the accompanying sections. This allows the user to modify
indicates the use of exceptionally high forces in the cloud
the system structure according to their requirements. This
case as opposed to the edge computing case. As worker node
non-insignificant increase of cohesion and synchronization
volume increases, the consumption of the power used by the
between the nodes of fog computing allows the execution of
FETCH system increases (shown in figure 17).
deep learning using the ensemble technique that improves the
transition search precision as well as the energy according to
the various requirements. Thus, the FETCH system gives the
improved architecture for medical representatives of
healthcare computation that was not proposed by any
existing work.

J. RESULT DISCUSSION
In the prior work[25], examination with the strength of
FogBus and earlier such fog computing enabled systems
demonstrated how FogBus gives more efficient execution of
the use of dealing with edge and cloud assets. This work
Figure 17. Results of consumption of power in watts (W) with
utilized the FogBus system with simplistically designing the
different fog computing scenarios
latency and the precise sensitive use of the heart patients and
quiet examination and provided an ideal opportunity to use
I. WORK ANALYSIS WITH THE EXISTED WORK the resources of edge and cloud effectively. The application
The various works propose to compute the models for the development system gave various setups that provide better
application of medical care using the fog computing accuracy or inactivity depending on the user requirements.
technique but they don’t consider the different approaches Given exploratory results, we propose the FETCH system to
that the FETCH system does. Much earlier work [28], [31], be used as an on-target application configuration based;
[39] does not use the influence resources that are closer to the For latency-critical and the targeting tasks that are
edge of the considered network. According to figure 13, such considered lightweight or the energy bounded conditions, the
models give very high latency because all the calculations are considered worker nodes should be used to fulfill the
performed on the cloud and consequently transfer high data. requirement. This gives an exceptionally short result time
Including the advanced technology i.e. artificial intelligence- due to the proximity of the worker nodes. On the off chance
enabled deep learning algorithm-based prediction model, the that the energy and the network data transfer capacity or
proposed model FETCH can make the best use in the class network bandwidth constraints requirements are in place,
neural network models for exceptionally accurate prediction then the bagging - ensemble technique should be deactivated
of health characteristics of patients. Various other works such but in the case, if it is not deactivated it will enabling the
as [2], [9] or more related works done by other researchers bagging that will support the better accuracy.
like [5], [27], [28], [44] do not have the ability to include

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

A CDC configuration with adequate and latency open expressed to improve accuracy. In addition, the proposed
views should be used in any case such that the task would not architecture can be made robust and nonexclusive to support
have the option of effectively completing on the resource other fog computing-based applications, for example, smart
constraint edge of worker nodes. production lines in the agricultural sector, weather
The proposed model is limited and dependent on the past forecasting domains, smart city forecasting domains, and
workload data values, which supports boosting the nature of traffic management for industry domains. FETCH can
decision-making over minimizing the error. Model reliability likewise be expanded to other important areas of healthcare,
is high over the low features noise and bias values. for example, cancer, diabetes, and hepatitis, which can
support efficient healthcare services to the concerned user or
patient.
VIII. CONCLUSION AND THE FUTURE DIRECTIONS
Medical care as an assistant is a vast and sensitive project. REFERENCES
In the current research work, our main objective is to focus
[1] S. M. R. Islam, D. Kwak, M. D. H. Kabir, M. Hossain, and K.-S.
on the healthcare system for heart patients, which propose a Kwak, “The internet of things for health care: a comprehensive
Fog computing enabled smart healthcare system that survey,” IEEE access, vol. 3, pp. 678–708, 2015.
incorporated with the latest technology i.e. deep learning [2] A. M. Rahmani et al., “Exploiting smart e-Health gateways at
the edge of healthcare Internet-of-Things: A fog computing
ensemble technique that provides an automated diagnosis of
approach,” Futur. Gener. Comput. Syst., vol. 78, pp. 641–658,
heart disease with the utilizing of FETCH system for IoT 2018.
enabled resources. FETCH delivers medical care in the form [3] S. Tuli, N. Basumatary, and R. Buyya, “Edgelens: Deep learning
of fog computing services and effectively manages the data based object detection in integrated iot, fog and cloud computing
environments,” in 2019 4th International Conference on
of a heart patient user that come from various IoT-enabled Information Systems and Computer Networks (ISCON), 2019, pp.
devices. FETCH incorporates the deep learning technique in 496–502.
edge computing enabled devices and deployed these devices [4] S. S. Gill, R. C. Arya, G. S. Wander, and R. Buyya, “Fog-based
smart healthcare as a big data and cloud service for heart patients
in the environment for real-life use of heart disease testing. In using IoT,” in International Conference on Intelligent Data
the prior art for such a heart patient examination did not Communication Technologies and Internet of Things, 2018, pp.
utilize the deep learning technique and resulted in an 1376–1383.
[5] S. He, B. Cheng, H. Wang, Y. Huang, and J. Chen, “Proactive
exceptionally low accuracy rate in the prediction power, personalized services through fog-cloud computing in large-
which rendered them redundant in practical settings. The AI- scale IoT-based healthcare application,” China Commun., vol.
based deep learning model with high accuracy requires 14, no. 11, pp. 1–16, 2017.
exceptionally high processing resources (that includes CPUs [6] I. Abdullahi, S. Arif, and S. Hassan, “Ubiquitous shift with
information centric network caching using fog computing,” in
or GPUs) for both purposes i.e. to train the model or Computational intelligence in information systems, Springer,
predicted the model. This work allowed the network with a 2015, pp. 327–335.
complex deep learning model to set in edge computing [7] M. Satyanarayanan, “The emergence of edge computing,”
Computer (Long. Beach. Calif)., vol. 50, no. 1, pp. 30–39, 2017.
standards that use the correspondence and model dispersion [8] A. Goyal et al., “Seasonal variation in 24 h blood pressure
methods, such as to ensure that high latency can be met with profile in healthy adults-A prospective observational study,” J.
low latency. It was approved for examining real-life heart Hum. Hypertens., vol. 33, no. 8, pp. 626–633, 2019.
[9] S. S. Gill, P. Garraghan, and R. Buyya, “ROUTER: Fog enabled
patient data through formulating neural networks on well- cloud based intelligent resource management approach for smart
known datasets and sending a working framework, which home IoT devices,” J. Syst. Softw., vol. 154, pp. 125–138, 2019.
provides real-time prediction capability. We have used the [10] K. C. Okafor, “Dynamic reliability modeling of cyber-physical
edge computing network,” Int. J. Comput. Appl., vol. 43, no. 7,
FogBus structure to approve the FETCH in an environment
pp. 612–622, 2021.
of fog computing and try the effectiveness of the proposed [11] K. C. Okafor, I. E. Achumba, G. A. Chukwudebe, and G. C.
framework regarding consumption of power, network Ononiwu, “Leveraging fog computing for scalable IoT
bandwidth, latency, jitter, the accuracy of the training model, datacenter using spine-leaf network topology,” J. Electr. Comput.
Eng., vol. 2017, 2017.
testing precision, and the time of execution. [12] A. A. Mutlag, M. K. Abd Ghani, N. al Arunkumar, M. A.
As per the future direction of this proposed work, the Mohammed, and O. Mohd, “Enabling technologies for fog
extension work of FETCH allows for cost-optimal exaction computing in healthcare IoT systems,” Futur. Gener. Comput.
Syst., vol. 90, pp. 62–78, 2019.
to provide a specific quality of service features and a hybrid [13] O. Faust, Y. Hagiwara, T. J. Hong, O. S. Lih, and U. R. Acharya,
fog-cloud cost model. In the present work, FETCH works “Deep learning for healthcare applications based on
with the record-based dataset that can be continuously physiological signals: A review,” Comput. Methods Programs
Biomed., vol. 161, pp. 1–13, 2018.
incorporated to capture the data directly from the sensor to [14] X. Zhao et al., “Deep learning based mobile data offloading in
make it easier to understand. In addition, the model mobile edge computing systems,” Futur. Gener. Comput. Syst.,
preparation system currently uses a different preparation on vol. 99, pp. 346–355, 2019.
each worker node in the system. The models formulated at [15] T.-S. Lim, W.-Y. Loh, and Y.-S. Shih, “A comparison of
prediction accuracy, complexity, and training time of thirty-three
each node are consolidated using various aggregated models old and new classification algorithms,” Mach. Learn., vol. 40, no.
of stowing. More curiosity models may be additionally 3, pp. 203–228, 2000.

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

[16] P. Datta and B. Sharma, “A survey on IoT architectures, 2018.

protocols, security and smart city based applications,” in 2017 [34] P. K. Sahoo, S. K. Mohapatra, and S.-L. Wu, “SLA based
8th International Conference on Computing, Communication healthcare big data analysis and computing in cloud network,” J.
and Networking Technologies (ICCCNT), 2017, pp. 1–5. Parallel Distrib. Comput., vol. 119, pp. 121–135, 2018.
[17] V. H. Buch, I. Ahmed, and M. Maruthappu, “Artificial [35] M. G. R. Alam, M. S. Munir, M. Z. Uddin, M. S. Alam, T. N.
intelligence in medicine: current trends and future possibilities,” Dang, and C. S. Hong, “Edge-of-things computing framework
Br. J. Gen. Pract., vol. 68, no. 668, pp. 143–144, 2018. for cost-effective provisioning of healthcare data,” J. Parallel
[18] T. Panch, H. Mattie, and L. A. Celi, “The ‘inconvenient truth’ Distrib. Comput., vol. 123, pp. 54–60, 2019.
about AI in healthcare,” NPJ Digit. Med., vol. 2, no. 1, pp. 1–3, [36] R. M. Abdelmoneem, A. Benslimane, E. Shaaban, S.
2019. Abdelhamid, and S. Ghoneim, “A cloud-fog based architecture
[19] L. E. Juarez-Orozco, O. Martinez-Manzanera, F. M. Van Der for iot applications dedicated to healthcare,” in ICC 2019-2019
Zant, R. J. J. Knol, and J. Knuuti, “241 deep learning in IEEE International Conference on Communications (ICC), 2019,
quantitative PET myocardial perfusion imaging to predict pp. 1–6.
adverse cardiovascular events,” Eur. Hear. Journal- [37] O. Akrivopoulos, D. Amaxilatis, A. Antoniou, and I.
Cardiovascular Imaging, vol. 20, no. Supplement_3, pp. jez145- Chatzigiannakis, “Design and evaluation of a person-centric
005, 2019. heart monitoring system over fog computing infrastructure,” in
[20] U. R. Acharya, H. Fujita, S. L. Oh, Y. Hagiwara, J. H. Tan, and Proceedings of the First International Workshop on Human-
M. Adam, “Application of deep convolutional neural network centered Sensing, Networking, and Systems, 2017, pp. 25–30.
for automated detection of myocardial infarction using ECG [38] P. K. Donta, S. N. Srirama, T. Amgoth, and C. S. R. Annavarapu,
signals,” Inf. Sci. (Ny)., vol. 415, pp. 190–198, 2017. “Survey on recent advances in IoT application layer protocols
[21] I. Azimi, J. Takalo-Mattila, A. Anzanpour, A. M. Rahmani, J.-P. and machine learning scope for research directions,” Digit.
Soininen, and P. Liljeberg, “Empowering healthcare iot systems Commun. Networks, 2021.
with hierarchical edge-based deep learning,” in Proceedings of [39] D. K. Sanghera et al., “Testing the association of novel meta-
the 2018 IEEE/ACM International Conference on Connected analysis-derived diabetes risk genes with type II diabetes and
Health: Applications, Systems and Engineering Technologies, related metabolic traits in Asian Indian Sikhs,” J. Hum. Genet.,
2018, pp. 63–68. vol. 54, no. 3, pp. 162–168, 2009.
[22] S. R. Moosavi et al., “End-to-end security scheme for mobility [40] N. Kato et al., “Trans-ancestry genome-wide association study
enabled healthcare Internet of Things,” Futur. Gener. Comput. identifies 12 genetic loci influencing blood pressure and
Syst., vol. 64, pp. 108–124, 2016. implicates a role for DNA methylation,” Nat. Genet., vol. 47, no.
[23] S. Tuli, R. Mahmud, S. Tuli, and R. Buyya, “FogBus: A 11, pp. 1282–1293, 2015.
blockchain-based lightweight framework for edge and fog [41] T. G. Dietterich, “Ensemble methods in machine learning,” in
computing,” J. Syst. Softw., vol. 154, pp. 22–36, 2019. International workshop on multiple classifier systems, 2000, pp.
[24] C. Vecchiola, X. Chu, and R. Buyya, “Aneka: a software 1–15.
platform for .NET-based cloud computing,” High speed large [42] F. Pedregosa et al., “Scikit-learn: Machine learning in Python,” J.
scale Sci. Comput., vol. 18, pp. 267–295, 2009. Mach. Learn. Res., vol. 12, pp. 2825–2830, 2011.
[25] S. Ali and M. Ghazal, “Real-time heart attack mobile detection [43] D. Dua and C. Graff, “UCI Machine Learning Repository
service (RHAMDS): An IoT use case for software defined [http://archive. ics. uci. edu/ml]. Irvine, CA: University of
networks,” in 2017 IEEE 30th Canadian conference on California, School of Information and Computer Science, zuletzt
electrical and computer engineering (CCECE), 2017, pp. 1–6. abgerufen am: 14.09. 2019,” Google Sch., 2019.
[26] M. Rajasekaran, A. Yassine, M. S. Hossain, M. F. Alhamid, and [44] R. Malik et al., “Multiancestry genome-wide association study
M. Guizani, “Autonomous monitoring in healthcare environment: of 520,000 subjects identifies 32 loci associated with stroke and
Reward-based energy charging mechanism for IoMT wireless stroke subtypes,” Nat. Genet., vol. 50, no. 4, pp. 524–537, 2018.
sensing nodes,” Futur. Gener. Comput. Syst., vol. 98, pp. 565–
576, 2019.
Parag Verma, having experience of 9+ years in the
[27] N. Constant, D. Borthakur, M. Abtahi, H. Dubey, and K.
field of industry and academics. He is having 25+
Mankodiya, “Fog-assisted wiot: A smart fog gateway for end-to-
international papers in reputed conferences and
end analytics in wearable internet of things,” arXiv Prepr.
journals. He has contributed 3 books with publishers
arXiv1701.08680, 2017.
like Taylor and France (CRC Press), Nerosa and
[28] A. Rajkomar et al., “Scalable and accurate deep learning with
Alpha, and currently working on 2 more accepted
electronic health records,” NPJ Digit. Med., vol. 1, no. 1, pp. 1–
books. He had also contributed 5+ chapters for
10, 2018.
reputed publishers. He is also an editorial board
[29] T. N. Gia et al., “Low-cost fog-assisted health-care IoT system
member of many reputed conferences and journals including Scopus. He is
with energy-efficient sensor nodes,” in 2017 13th international
also a guest editor of IJNCDS (Scopus Indexed inderscience Journal) and
wireless communications and mobile computing conference
many more. He is associated with many societies and organizations for the
(IWCMC), 2017, pp. 1765–1770.
welfare of educationalist societies. His area of research is IoT, healthcare,
[30] O. Akrivopoulos, I. Chatzigiannakis, C. Tselios, and A.
image processing-based analysis using distributed computing platforms.
Antoniou, “On the deployment of healthcare applications over
fog computing infrastructure,” in 2017 IEEE 41st Annual
Computer Software and Applications Conference (COMPSAC),
Rajeev Tiwari, is a senior IEEE member and IEIM.
2017, vol. 2, pp. 288–293.
He has received Ph.D. from Thapar University,
[31] E. Choi, M. T. Bahadori, L. Song, W. F. Stewart, and J. Sun,
Patiala. He has 15+ years of research and teaching
“GRAM: graph-based attention model for healthcare
experience. He has published different articles in
representation learning,” in Proceedings of the 23rd ACM
international journals of repute with Elsevier,
SIGKDD international conference on knowledge discovery and
Springer, IEEE etc. He has chaired many
data mining, 2017, pp. 787–795.
International conferences globally and delivered
[32] R. K. Barik, R. Priyadarshini, H. Dubey, V. Kumar, and K.
keynotes in field of Sensor Network, IoT-Fog-
Mankodiya, “FogLearn: leveraging fog-based machine learning
Cloud Based Computing. He has edited 4 books
for smart system big data analytics,” Int. J. Fog Comput., vol. 1,
with Springer and Taylor and Francis. He has completed one project on
no. 1, pp. 15–34, 2018.
smart parking system using IoT devices. He has 3 patents by his name.
[33] M. Pham, Y. Mengistu, H. Do, and W. Sheng, “Delivering home
healthcare through a cloud-based smart home environment
(CoSHE),” Futur. Gener. Comput. Syst., vol. 81, pp. 129–140,

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2022.3143793, IEEE Access

WEI-CHIANG HONG, is a professor at the School of

Computer Science and Technology, Jiangsu Normal
University, China. His research interests mainly
include computational intelligence and application of
forecasting technology. Prof. Hong's articles have
been published in Applied Mathematics and
Computation, Applied Soft Computing, Decision
Support Systems, Energy, Energy Policy, European
Journal of Operational Research, IEEE Transactions on Fuzzy Systems,
and Neurocomputing among others. Prof. Hong is currently on the
editorial board of several journals, including International Journal of
Applied Evolutionary Computation, Neurocomputing, and Applied Soft
Computing. In May 2012, his paper had been evaluated as "Top Cited
Article 2007-2011" by Elsevier Publisher (Netherlands). In Sep. 2012,
once again, his paper had been indexed in ISI Essential Science Indicator
database as Highly Cited Papers. Prof. Hong is a senior member of IIE and
IEEE.

Shuchi Upadhyay, She is a Ph.D. in Food and

Nutrition, She has experience of clinical trials
analysis and experimentation. She has 14+ years of
research and teaching experience. She has many
papers in international renowned journals,
conferences and book chapters. She has completed
one Govt. of India Funded Project with Baal Vikas
department. Presently she is working in another
funded project as CO PI in healthcare and Pharma area.

Yi-Hsuan Yeh is an assistant professor at the

Department of Information Management, Asia Eastern
University of Science and Technology, Taiwan. Her
research interests mainly include computational
intelligence and application of information technology.
Dr. Yeh’s articles have been published in Utilities
Policy, Applied Soft Computing, among others. She
has 20+ years of research and teaching experience.
She has completed 10+ enterprises’ projects in
Taiwan.

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/