A preview of this full-text is provided by IGI Global.
Content available from International Journal of Grid and High Performance Computing
This content is subject to copyright. Terms and conditions apply.
DOI: 10.4018/IJGHPC.304908
Volume 14 • Issue 1
Copyright © 2022, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
*Corresponding Author
1
The main concern of fog computing is reducing data transmission on the cloud. Moreover, due to
the short distance between end-user and fog nodes, fog computing is considered more reliable to
handle time-sensitive situations like the critical data provided by the internet of things (IoT). This
may include sensory healthcare data which needs rapid processing to make decisions. However, in
healthcare monitoring systems, it is necessary to ensure the services’ availability when fog node failure
occurred. The issue of monitoring service interruption during fog node failure has not received much
attention. This paper proposes a multi-route plan that aims to identify an alternative route to ensure
the availability of time-critical medical services. Various scenarios have been designed to evaluate
the performance of the proposed strategy. The experimental results illustrate the superiority of the
approach in terms of latency, energy consumption, and network usage in comparison with the most
recent related work.
Fog Computing, Healthcare Monitoring Systems, IoT, Multi-Route, Node Failure, Time-Sensitive Application
The e-healthcare systems both web-based and mobile-based versions use wireless personal area
networks (WPAN) or/and wireless body sensor networks (WBSN) for delivering high-quality real-time
medical services and efficient medical treatments to patients (Sanna et al., 2014). In such systems,
sensors are fixed around the patient’s body to collect vital information on the patient such as oxygen
level, sugar level, heart rate, and pulse rate. This collected data is reported immediately to the remote
designated physician and/or to a healthcare service provider attempting to ensure taking suitable action
when detecting abnormalities, which is an application of Cloud-based e-healthcare systems (Xu et
al., 2018; Park et al., 2019). Healthcare data needs a cloud platform to manage the large volume of
generated data instead of relying on limited computing resources. However, this causes a high delay
Nour El Imane Zeghib, International Islamic University Malaysia, Malaysia
Ali A. Alwan, Ramapo College of New Jersey, USA*
Abedallah Zaid Abualkishik, American University in the Emirates, UAE
Yonis Gulzar, King Faisal University, Saudi Arabia
Volume 14 • Issue 1
2
that affects healthcare services negatively especially those requiring an immediate response which
is one of cloud computing drawbacks (Ahmad et al., 2016; Zhan et al., 2019).
In this regard, in 2014, Cisco announced a new computing concept which is fog computing, a
new infrastructure paradigm to go beyond the confines of cloud computing (Bonomi et al., 2014). Fog
computing infrastructure, on one hand, consists of many fog nodes, virtualized data centers, and IoT
devices, which have an established connection with the cloud for more implementation of permanent
storage and powerful computational capabilities (Dutta & Roy, 2017; Naranjo, Shojafar, Mostafaei,
Pooranian, & Baccarelli, 2017). Fog computing is held to be a suitable paradigm when it comes to
designing real-time or latency-sensitive healthcare applications. It is significantly contributing to
healthcare applications such as Ambient Assisting Living (AAL) applications, remote home nursing
that serves elderly people, and real-time tracking of chronic diseases (Al-Khafajiy, Webster, Baker,
& Waraich, 2018; Awad, Khanapi, Ghani, & Arunkumar, 2019). However, due to its infancy; several
other issues relevant to fog computing can be highlighted like sensor node failure or removal, external
attack on BAN/WBAN, environmental coincidences, loss or limited power, loss of connectivity, and
failure of the network, network congestion. Thus, system failure should be carefully addressed while
designing emergency or real-time e-healthcare systems (Kher, 2016).
Node failure has a direct impact on many aspects of the system including deteriorate of the
performance of the system, service disruption, increase the cost of operation, delay in service delivery
time and unpleasant user experience (Satria, Park, & Jo, 2017; Ullah, Sehr, Akbar, & Ning, 2018).
Most importantly, the issue of node failure has a significant negative effect on medical services in a
healthcare monitoring system. This is because any medical service disruption due to the node failure
might results in a potential loss of human life if the most appropriate action is not taken within the
required time. Hence, a significant design feature of these strategies and recovery techniques must
ensure service protection with low latency and less expensive execution costs (Khan, Parkinson, &
Qin, 2017). Most of the previous approaches designed for healthcare monitoring systems assumed
that the main fog node is always available and can entertain the user request, particularly for time-
critical cases. However, this assumption is not always true, and the designated fog node might fail to
entertain the user request and the latency to respond to the user becomes extremely long. Therefore,
preventing the patient from getting the necessary immediate aid. Hence, longer latency might result
in negative implications (death of the person) if the fog node fails to send the generated alert within
a reasonable time. Thus, an efficient approach is needed to consider such cases. The approach has
to identify and establish an alternative path through more than one fog node to be involved in such
critical situations to ensure system reliability requirements. This paper proposes an approach that
aims at establishing an alternative path to entertain the medical services for healthcare monitoring
systems. The approach tries to make sure that the system is reliable in which the critical alerts can
be reached to the fog node and the cloud instead of relying on a single path only that might leads to
signal loss in any failure case.
The following points summarize the contributions of this paper:
• We discuss the impact of the problem of node failure in fog-based healthcare monitoring systems.
• We conduct a comprehensive review of the remarkable studies and relevant works in the area
of fog-based healthcare monitoring systems. The review emphasizes examining in-depth the
strengths and the limitations of each work.
• A new approach that tackles the issue of fog node failure aiming at ensuring the sustainability of
healthcare monitoring systems’ services to handle the critical cases and avoid service disruption
is proposed.
• To evaluate the effectiveness and the performance of the proposed approach of identifying the
alternative route when node failure occurred in fog-based healthcare monitoring systems in terms
of latency, energy consumption, and network usage.
Volume 14 • Issue 1
3
The remainder of the paper is organized as follows: The necessary background and concepts
related to fog computing are explained in the next section. Then, the previous work relevant to the
healthcare monitoring systems in fog computing has been explained in the following section. Next,
the proposed framework for a fog-based healthcare monitoring system is described in the subsequent
section. Also, the detailed steps of the proposed approach for the multi-route plan for fog-based
healthcare monitoring systems have been presented in the next section. This is followed by reporting
the experiment results and discussing the results as well. Finally, the conclusion is outlined in the
last section.
This section provides the necessary background and the fundamental concepts related to fog computing
and exposes its architecture. It also presents an overview of fog computing and its role in different
real-time healthcare systems Furthermore, the section explains and discusses the potential benefits
of redundancy for reliable healthcare services.
“Fog Computing” was initiated by Cisco Systems in 2012 as an exclusive platform that adds the
capabilities of the cloud to the edge of the network which allows remote data transmission and
processing among distributed devices in IoT network (Bonomi et al., 2014). Like the cloud, fog
computing provides storage on-demand, network resources, and computing services to the edge
devices which lessen the burden on the centralized cloud. However, fog computing provides additional
advantageous characteristics as its proximity to the user which reduces the latency by enabling the
data to be processed immediately on the fog instead of sending them to the cloud, which make it
important for delay-sensitive services, for instance, emergency cases and healthcare applications
(Aazam & Huh, 2015; Nath, Gupta, Chakraborty, & Ghosh, 2018). The geographical distribution
nature of fog commuting provides a multi routs options for data dissemination over the fog depends
on network conditions to ensure the availability of the services in any failure cases, which is an
important feature for critical services (Naranjo et al., 2017; Dziak, Jachimczyk, & Kulesza, 2017).
Fog computing also allows the mobility of the user and enables location awareness services by
exploiting its geographical distribution to keep track of network traffic information provided by
the connected fog nodes (Anderiopoulou, Tasos, & Orphanoudakis, 2017; Gao, Luan, Yu, Zhou,
& Liu, 2017). Fog computing adds a thin layer right between the IoT devices and a remote cloud
server through fog nodes and servers, which are the fundamental components in the fog computing
paradigm. Fog computing is used in a variety of applications such as healthcare monitoring that is
time-sensitive and requires a fast response by employing mobile nodes to transfer streaming data
from one point to another without considering time limitations (Kai, Cong, & Tao, 2016; Gao et al.,
2017). Figure 1 represents the typical IoT-fog computing architecture (Andriopoulou, Dagiuklas, &
Orphanoudakis, 2017).
The deployment of healthcare IoT largely uses fog computing for patients’ remote monitoring
services, as it potentially can accelerate the detection of early predictors and novel biomarkers that
help in making smart healthcare decisions in health associated situations, which lets the patients stay
in touch with their caregivers, and enable them to bring responsible, valuable, and timely care to
their patients(Cao, Hou, Brown, Wang, & Chen, 2015). Fog computing has been deployed in many
healthcare applications, in which they differ from each other concerning the users and stakeholders
involved as well as devices and connectivity. Firstly, mobile scenario, in which patient’s mobile
phones can act as a fog node between the sensor devices and cloud to have an expanded battery life
of the wearable sensor device, as the mobile device is placed near to those sensors (MasipMasip-
Volume 14 • Issue 1
4
Bruin, Marin-Tordera, Alonso, & Garcia, 2016). Secondly, in-home treatment patients are connected
through their internet plans. Similar to the mobile scenario, the fog node is utilized to gather, store,
and process raw data, before transmitting it to the cloud to be stored permanently. The impetus that
calls for fog computing here is to mitigate network traffic and latency by performing the regular
operations locally without a need to transfer them to a cloud provider (Paul, Pinjari, Hong, Seo, &
Rho, 2018). Thirdly, hospitals deployed fog nodes within their structure, fog nodes are possessed
and upheld inside the hospital. Fourthly, non-hospital sites which have less staff and infrastructure
such as clinics, or nursing homes also benefit from the geographical distribution of fog computing
with little communication overhead as well as reduces the load between the local area network and
the detection systems for seizures located inside a cloud center (Ullah et al., 2018). Lastly, vehicular
networks are also benefiting from using fog computing by connecting the patient’s vehicle to the
nearest ambulance and triggering the embedded system to park the car in the nearest emergency
parking area (Naranjo et al., 2017).
A dependable system is a system that can be trusted and able to ensure the availability of the services
whenever requested as well as ensure the continuity of operations to deliver those services and
avoid any frequent service failure that might occur (Laprie, Randell, Landwehr, & Member, 2004).
Implementing dependable real-time healthcare systems is a necessity to ensure their reliability
in performing the urgent operations needed when sudden human life threats occurred. However,
dependability requirements in fog computing are still not well defined as fog computing is still in
its infancy. Nevertheless, they can be achieved based on common dependability objectives such as
improving availability, reliability, and quality of service (QoS). The improvement of these features
is typically depending on different redundancy techniques that can be applied at different system
architecture levels. This includes redundancy in communication links that deals with network links,
redundancy in computing nodes to deal with node failure or redundancy of application software
that deal with process reallocation (Bakhshi & Rodriguez-navas, 2019). Given the interest in fog
computing, dependability, and reliability issues due to the failure of a potential node in the critical
systems are challenging. Most healthcare monitoring systems have strict real-time requirements and
any failure cause catastrophic consequence risking human life.
Figure 1. IoT-Fog-Cloud Architecture
Volume 14 • Issue 1
5
This section presents and examines the previous approaches proposed for healthcare monitoring
systems in fog computing. The discussion emphasizes the previous existing approaches that utilize fog
computing in the healthcare domain. The discussion demonstrates the significance of fog computing
in healthcare systems.
The work by (Fratu et al., 2015) discusses the issue of developing a fog-based health monitoring
system concentrating on Mild Dementia and COPD disease. Fratu et al. (2015) have adopted fog
computing for implementing the eWALL application to remotely monitor and support senior people
who are affected by Mild Dementia and COPD disease. eWALL application aims at improving their
quality of life in such a way it allows them to keep their daily habit without invasive support from
any relatives or caregivers. The fog node is responsible for all real-time operations, analysis, and
sending the notifications to eWALL health provider cloud to get the assistance of physicians. Fratu
et al. (2015) argued that using fog computing speeds up the real-time processing and allow receiving
emergency notifications in a short time compared to the conventional way of engaging the cloud.
The work introduced by (Craciunescu et al. 2015) has proposed a fog computing system that
explored the probability of offloading storage and processing jobs of the healthcare cloud to the edge
of the network to lessen the latency related to performing these jobs inside the cloud. Craciunescu
et al. (2015) have implemented a laboratory e-Health scenario in the patient’s home to detect any
fall and gas leakage events that should be reported in real-time. The real-time processing of the
designed algorithms is performed by the fog node which is a personal computer at home, whereas the
extrapolated metadata is transmitted to the cloud to be processed thoroughly. Through the extracted
results, Craciunescu et al. (2015) have found that data fusion through fog computing enables real-
time delivery of patients’ information to their caregivers. Furthermore, Aazam & Huh (2015) have
focused on providing a simple and fast procedure for notifying different emergencies to keep the
victim from thinking too much. To achieve this aim the authors have implemented a smartphone-
based service, that offers rapid notifications to the applicable emergency department by making use
of different characteristics of fog computing for data offloading and pre-processing purposes. The
fog layer is responsible for the whole process of notifications. Information related to an emergency
is accordingly transferred from the fog to the cloud, permitting further analysis and better care for
the people and generating thorough portfolios of services for the relevant authorities and the users.
A fog-assisted cloud-based healthcare system for forecasting and averting the Chikungunya virus
has been proposed in the work introduced by (Sood & Mahajan, 2017). They have designed a system
that consists of wearable sensor technology, decision trees to categorize the users into dissimilar groups,
and the TNA graph that states the outbreak level of the virus. The system determines the likelihood
of a health index with respect to time, which allows sending alerts to the user’s mobile immediately
through the fog layer. The cloud layer is responsible for the complex processes and storage that can’t
be done by the fog. The experimental results demonstrate that the proposed system has achieved a
higher accuracy of the designed decision tree and lower response time in defining the state of an
event in comparison to other categorization algorithms that do not incorporate fog-cloud technology.
The work proposed by Gia et al. (2017) has introduced a fog-assisted health monitoring IoT system
that offers advanced services like data processing, classification, and push notification to enhance the
overall quality of healthcare services. The focus is given on providing continuous remote monitoring
of electrocardiography (ECG) and enhancing the operation time of data analysis and notification
generation. To achieve this aim, Gia et al. (2017) have designed sensor nodes considering software
and hardware elements to reduce power consumption as much as possible. Gia et al. (2017) have
concluded that by implementing this system at hospitals and homes, relevant doctors can be informed
in real-time to help prevent dire consequences by acting promptly. Moreover, with the help of fog
computing in smart gateways, the quality of healthcare service is drastically enhanced, and the energy
consumed by the sensor nodes is reduced because of their proximity to the gateway.
Volume 14 • Issue 1
6
The problem of designing a mobile-platform reliable healthcare monitoring system for heart attack
disease has been addressed by (Ali & Ghazal, 2017). They have proposed a Real-time Heart Attack
Mobile Detection Service named RHAMDS. An e-health IoT service leveraging Software Defined
Networks (SDN) powered VANET architecture that seeks to mitigate as well as avert vehicle collision
by identifying heart attacks that drivers may just encounter. In order to achieve the aforementioned
aim, the researchers have implemented two RHAMDS models, one controlled by voice and the other
by gesture through utilizing smartwatches. A fusion of these variations has been done through the
implementation of voice command followed by wrist gesture recognition for a higher consistency as
well as the correctness of the model.
The issue of fog node failure has been highlighted by the work in Souza et al. (2017). They have
addressed the failure recovery problem in fog-cloud architecture pointing at reducing the service
transmission delay as well as the protection cost. The work concentrates on assessing both proactive and
reactive strategies to recover the failure of network elements by modeling them as a Multidimensional
Knapsack Problem (MKP) to examine their impact of them on the service allocation time, recovery
delay, and computing resources load.
Satria et al. (2017) have also addressed the issue of failure recovery when mobile edge computing
fails. They have proposed two recovery systems for overloaded or broken Mobile Edge Computing
(MEC). The MEC provides real-time services in various field applications which makes MEC failure
a dire issue. The simulation results of this sharing of the workload with neighboring MECs via ad-hoc
relay nodes present that this MEC recovery strategy is likely to thrive in dense zones. Furthermore, it is
discovered that the capacity of ad-hoc relay nodes affects the MEC recovery system data meaningfully.
Singh et al. (2018) have proposed a Fog computing-based framework that categorizes patients
suffering from dengue based on the severity of the fever, and decides whether the patient is uninfected,
infected, or severely infected. The author aims to develop a latency-sensitive system that uses the
Internet of Things (IoT) sensors as well as generated audio and video files to categorize the patients
according to their respective symptoms. The fog devices located in the fog architecture reduce the
latency due to the proximity of the user. Data generated from IoT devices is at first processed by the
fog layer. Then, the generated data will be directed to a cloud to be processed deeply or to be stored
permanently, and then it is accessed by different entities such as users, hospitals, physicians, and
healthcare agencies.
Dar et al. (2019) have proposed fog-based detection of accidents and an emergency response
system that aims to keep up with emergencies in real-time to manage the huge chunks of data,
devices, and sensors without a need for human intervention. For this intention, fog-based accident
detection and an emergency response system named ERDMS that utilizes Android device sensors
are proposed. The proposed system attempts to exploit the fog computing characteristics to detect
an accident emergency, or accident location, and devise a way to manage the emergency effectively
as well as efficiently.
Last but not least, Huang & Guo (2019) have identified the challenge of providing a proactive
failure recovery mechanism for the Network Function Virtualization (NFV) enabled distributed edge
computing. To tackle this problem, they have proposed an innovative man agement architecture that
supports the proactive failover mechanism and provisions NFV ser vices in distributed edge computing
to allow failure prediction simultaneously. Huang & Guo (2019) have demonstrated in simulation the
significant advantages of the proactive mechanism through a comparison between failover latency
and the reactive mechanism. Table 1 summarizes the previous works presented and explained in this
section. The table shows the type of system, the technology used, the methodology, the services, the
considered parameters, and the limitation of each work.
Volume 14 • Issue 1
7
This paper focuses on realizing the reliability of healthcare services. In general, redundancy techniques
are considered an effective approach to ensure the dependability objectives of a system (Bakhshi
& Rodriguez-navas, 2019). Path and node redundancy, somehow, is seen as a potential solution to
ensure the continuity of operation during failure cases. Yet, the absence of an alternative plan in
some healthcare monitoring systems is considered a drawback, as these systems should be built
into a reliable manner due to its relation to human life that can’t be compromised. To the best of
our knowledge, there is a lack of comparative studies about engaging a fog-based redundant route
plan into the structure of this kind of time-sensitive healthcare services (Aazam & Huh, 2015; Ali &
Ghazal, 2017; Craciunescu et al., 2015; Dar, Shah, Islam, et al., 2019; Gia et al., 2017; Singh et al.,
2018; Sood & Mahajan, 2018). This paper, therefore, is conducted to fill such a gap. The comparative
analysis is done based on reviewing the strategy used by current healthcare monitoring systems.
The proposed framework for healthcare monitoring systems in the fog computing environment is
described in Figure 2.
Table 1. Summary of the Previous Approaches for Fog-based Healthcare Monitoring Systems
Author and
Year
Type of System Service Parameters
Considered
Limitations
Craciunescu
et al. (2015)
Fall detection and gas
leakage
Data fusion Accuracy + time Lack of a backup plan
Azam &
Huh (2015)
E-HAMC Alerting Time delay of
different file sizes
Lack of a backup plan
Sood &
Mahajan
(2017)
Predicting chikungunya
virus
Classification and
alerting
Accuracy + Time Lack of a backup plan
Gia et al.
(2017)
Remote monitoring of ECG Classification and
alerting
Energy consumption
+ time
Lack of a backup plan
Ali &
Ghazal
(2017)
RHAMDS heart attack
detection system in VANET
Alerting Notification receiving Lack of a backup plan
Souza et al.
(2017)
Fog-cloud architecture Failure of network
elements
Delay and cost Proactive allow
redundancy for real-
time fault tolerance
performance
Satria et al.
(2017)
Edge-fog Overloaded or broken
Mobile edge node fails
Time + throughout MEC provides real-
time services.
Singh et al.
(2018)
Dengue Detection Classification Respond time +
accuracy
Lack of a backup plan
Dar et al.
(2019)
Accident detection alerting Latency, network
usage, execution time.
Lack of a backup plan
Huang &
Guo (2019)
Edge computing Failure prediction The latency of failover
proactive vs reactive
Proactive Failover
prediction
Volume 14 • Issue 1
8
To test our proposed approach, we have designed and implemented four different testing scenarios.
The first two scenarios present the current system model being used in several industries and the
remaining two scenarios present the proposed work. A comparison between these approaches has
been carried out. The first scenario presents the normal state of the system in which the primary fog
node is assumed to be working properly with no failure to transmit the alert, this scenario represents
the ideal case. The second scenario presents the case in which the primary fog node fails to transmit
the alert and therefore, the cloud is used as an alternate to transmit the alert signal to notify the
responsible parties. However, this work assumes that more than one fog node can be placed at the
fog layer. Thus, this work proposed the third, fourth, and fifth scenarios based on the assumption
that the primary fog node might not be available, and referring to the cloud server to transmit the
alert signal should be avoided due to its prohibitive cost and the unacceptable latency. Therefore,
a secondary route using the next neighbor fog node should be established instead of relying on the
cloud that is placed further away (Fitzgerald, Pióro, & Tomaszewski, 2018; Satria et al., 2017). The
main focus was given to the third scenario by exploring the next neighbor fog node to take place if
the primary node fails, as recommended by many replication strategies in achieving better results
with maintaining the QoS requirements (Li, Yang, & Yuan, 2016). The fourth and fifth scenarios,
however, were designed to study the behavior of the system by utilizing more than one neighbor fog
node. We assumed that our proposed approach explores up to 4 neighbor fog nodes. The detailed
steps of the proposed approach are further explained in the following subsection.
Figure 3 clarifies the detailed steps of the proposed algorithm for a multi-route plan for a fog-based
healthcare monitoring system. The inputs of the algorithm include the heartrate value HRv received
from the edge device attached to the human body that triggers the system to start, the list of fog
nodes FN_list which can be up to 4 nodes (1 primary node and the remaining 3 secondary nodes),
and one Cloud datacenter Cdc. The output of the algorithm is the best route to transmit the HRv to the
designated destination, D (hospital). The algorithm starts by setting the first node, n0 in the fog node
list, FN_list to be the primary fog node to be checked first to transfer the heart rate value into the
designated destination (step 2). Then, the abnormal heartrate generated by the sensor device attached
to the human body is read and trigger an alert sent to the primary fog node n0 (steps 3-4). The edge
Figure 2. The Proposed Framework for Fog-Based Healthcare Monitoring Systems
Volume 14 • Issue 1
9
device keeps waiting for the acknowledgment message from n0 indicating that the heart rate value,
HRv has been received (step 5), then the sent HRv value is transmitted to the destination (hospital)
(step 6). Otherwise, if no acknowledgment message is received from n0 within the stipulated time;
the edge device attempts to explore the next neighbor fog nodes trying to employ them to send the
HRv to the designated distention (steps 7-16). The process of exploring the next neighbor fog node
continues until an acknowledgment is received from one of the fog nodes in FN_list, or all nodes in
FN_list have been visited. If an acknowledgment has been received from one of the secondary fog
nodes in FN_list, then the read HRv is transmitted to the hospital (steps 12-13). Otherwise, if none of
the fog nodes in the FN_list could accomplish the task; the edge device will send the HRv through the
cloud (steps 17-22). However, if no acknowledgment is received from the cloud, then the algorithm
will be terminated. Figure 4 demonstrates the detailed process flow of the proposed multi-route plan
for the fog-based healthcare monitoring system.
This section explains the proposed scenarios to evaluate the proposed approach for healthcare
monitoring systems based on the fog computing paradigm. Different scenarios have been designed.
These scenarios are further explained below.
Scenario (1) Primary Route With no Failure
This scenario has been implemented and simulated considering the normal condition of the network
in which everything is working properly. The edge device will send an alert to the Primary fog server
indicating a critical situation of the patient. The edge device receives an acknowledgment within 0.6
seconds which means the signal has been successfully reached the fog server. After that, the alert
is transmitted to the destination address which is the assigned hospital as depicted in Figure 5. This
scenario represents the normal states of the system when utilizing fog computing in such time-critical
applications.
Figure 3. Proposed Multi-route Plan Algorithm
Volume 14 • Issue 1
10
Scenario (2) Primary Route with Primary Fog Server Failure
This scenario represents the single route plan adopted by (Aazam & Huh, 2015; Dar, Shah, Shahid, et
al., 2019; Gia et al., 2017; Huang & Guo, 2019; Satria et al., 2017; Singh et al., 2018; Sood & Mahajan,
2018; Souza et al., 2017) for healthcare monitoring applications. For the sake of the experiment, the
PS here is being failed by making its CPU load settings less than the processing load needed to send
the alert from the source to the destination (i.e. getCurrentCpuLoad < getCurrentModuleLoadMap).
So, when the edge device detects an abnormal heartbeat rate, it sends an alert to the PS and waits
to receive an acknowledgment within 0.6 seconds. However, due to the aforementioned reason, the
edge device will not receive any acknowledgment within the time constraint, and the only way to
carry on the process is to send the alert to the assigned hospital through the cloud which is assumed
Figure 4. Proposed Multi-route Plan for Healthcare Monitoring System Flowchart
Figure 5. Scenario 1- Single Route with no Failure
Volume 14 • Issue 1
11
to work properly. When the alert is received, the cloud transmits the alert to the destination which is
the assigned hospital address as shown in Figure 6.
Scenario (3) Secondary route with Primary fog Server (PS) failure
This scenario and the next scenarios have been implemented and simulated to evaluate our proposed
Multi route strategy. As with the previous scenario, the PS has given less CPU load capacity than
required to force the edge device to take the alternative route. When the edge device does not receive
any acknowledgment within the time constraint it scans the area for the nearest available fog server in
the area which we have formed as a predefined fog node array that contains three fog node elements
just for the sake of simplicity. The Edge device will send the alert to the first Next Nearest fog Server
(NNS1) and get back an acknowledgment within 0.6 seconds indicating the reception of the signal.
After receiving the alert, NNS1 will transmit the Alert to Destination which is the assigned hospital
address as illustrated in Figure 7.
Scenario (4) Secondary route with Secondary fog Server NNS1 failure
As with the previous scenario, scenario 4 has been implemented and simulated to evaluate our
proposed Multi route strategy that deploys more fog nodes in the fog layer to work as a spare route
for time-critical services. In this scenario, we assume the failure of both the PS and the NNS1 as well
by controlling the CPU load of both servers for the simulation. The same aforementioned process
will happen until the edge device doesn’t receive the acknowledgment from NNS1 within the time
constraint and it starts searching for the second Next Nearest fog Server (NNS2. After sending the
alert, an acknowledgment is received indicating the reception of the signal and there is no need to
Figure 6. Scenario 2- Single Route with Failure in PS
Figure 7. Scenario 3- Secondary Route with Failure in PS
Volume 14 • Issue 1
12
search for another fog node. After receiving the alert, NNS2 will transmit the alert to the destination
as described in Figure 8.
Scenario (5) Secondary Route with Secondary Fog Server NNS2 Failure
In this scenario, we assume the failure of the PS, NNS1, and NNS2 as well by shutting down those
servers for the simulation. The same process will happen until the edge device doesn’t receive the
acknowledgment from NNS2 within the time constraint and it starts searching for the third Next
Nearest fog Server (NNS3) which is the last sever in the list as we have assumed, an acknowledgment
received will indicate the reception of the signal and there is no need to search for another fog node.
After receiving the alert, NNS3 will transmit the alert to the destination as demonstrated in Figure
9. If a situation arises where all available fog nodes in the area (i.e. NNS3) are unable to transmit the
alert. Then our proposed algorithm uses the conventional method of sending the alert through the
cloud as the only way to accomplish the task.
This section presents and discusses the experimental results of the approaches proposed in this paper.
It elaborates on the experiment settings to evaluate the performance of the proposed approaches for
the multi-route plan for time-sensitive services in fog-based healthcare monitoring systems. Besides,
it explains and discusses the experiment results analysis of the proposed approach.
Figure 8. Scenario 4- Secondary Route with Failure in NNS1
Figure 9. Scenario 5- Secondary route with Failure in NNS2
Volume 14 • Issue 1
13
To evaluate the efficiency and the effectiveness of the proposed approach for the multi-route plan for
time-sensitive services in fog-based healthcare monitoring applications, the scenarios explained in the
previous section have been used. The proposed approach has been developed using an iFogSim-master
package based on the iFogSim simulator toolkit. iFogSim has been used by many research works in
the area of fog computing due to its convenience to simulate fog computing infrastructures. In this
paper, we have compared our approach with the current single route approach employed by several
researchers for fog-based healthcare monitoring applications. To carry out the experiment, we run the
simulator on a workstation with Lenovo YOGA 530 (Lenovo) and the experiment for each scenario
has been run five times and the average has been reported in the result. Table 2 describes the hardware
and software simulation requirements. Besides, the experiment has the same setting throughout the
entire simulation process considering the fog environment properties for all resources including edge
devices, fog servers, and the cloud. The specifications of the fog nodes used are illustrated in Table
3. Resources’ settings that have been employed in the simulation are presented in Table 4.
This section explains the experimental results of the proposed approach of the multi-route plan for
time-sensitive services in fog-based healthcare monitoring systems. We have generated a range of
random abnormal heart rate values to trigger the implemented system and send the alert from the edge
device that is connected to the patient’s sensor to the destination point. We have run each scenario 10
times to achieve a stable state of the performance and the average result of each scenario is computed.
The set of experiments aims at examining the effectiveness of deploying more than one fog node to
establish an alternative route ensuring the continuity of time-critical services in healthcare monitoring
systems. The proposed approach has been evaluated in terms of latency, energy consumption, network
use, and the execution cost as demonstrated in the following subsections.
Latency
The most important performance metric in healthcare monitoring systems is the latency of response
i.e. the time needed for transforming the critical alert from the patient’s side to the hospital. This
requires real-time communication between the edge device and the device hosting the hospital module.
However, the delay caused by the failure of these systems may cause a severe effect if not losing
the patient’s life. Thus, the latency indicates the time taken from the moment the value of the IoT
device attached to the patients is read until the value reaches the designated point (hospital). Figure
Table 2. Simulation Setup Requirements
System Requirements
Hardware
Processor Intel Core i5-8250U CPU
Clock speed 1.8 GHz
RAM capacity 8 GB
SSD 500 GB
Software
Operating system Windows 10 Home 64-bit
Language Core Java
Version JDK v1.8.0
IDE Eclipse Mars 2
Simulator package iFogSim-master
Volume 14 • Issue 1
14
10 demonstrates the experimental results of the latency for the implemented scenarios considered
in this research work. The first scenario was designed to demonstrate the ideal system. The second
scenario represents the single route strategy adopted by E-HAMC is considered in the comparison.
This scenario assumes that when the edge device detects a failure of the primary fog server it sends
the alert directly through the nearest cloud which needs 288.5 milliseconds (» 0.3 seconds) to reach
the assigned hospital as shown in Figure 10. On the other hand, the results illustrate a noticeable less
time taken to re-transmit the critical alert through the secondary next nearest fog server in failure
cases which is estimated by 61.65 milliseconds (» 0.062 seconds) according to S3. This is due to the
shorter bandwidth links between the end-user and the fog server that is located closer to the user.
Consecutively, S4, and S5 achieve a slightly high latency than S3 because those fog servers are further
away. However, they maintain low latency as well when facing the failure of the previous fog server
as discussed in Section 4.2, and that is because all fog servers are placed in the users’ premises as in
the neighborhood that can be a few kilometers far, instead of relying on the cloud that may be placed
far away or might turn to a bottleneck issue while executing many tasks at one point, which caused
a notable increase in latency. According to the results, we conclude that our fog-based multi-route
approach outperforms the previous single-route approach in all the scenarios in terms of latency
which is a very crucial factor to provide reliable real-time services, where a second or millisecond
can make a difference in saving a life.
Table 4. Resources Settings
Parameter Value used
Node Name Edge Device Cloud Fog Servers
CPU load (mips) 1000 10000 2800
RAM (mb) 1000 40000 4000
UpBw (bit/s) 10000 100 10000
DownBw (bit/s) 270 10000 10000
Level 3 0 1
RatePerMips 0 0.01 0.0
BusyPower (W) 5 10*11.5 10
IdlePower (W) 5 8*10 10
Table 3. Destination node Specifications
Parameter Value
system architecture x86
operating system Linux
VMM “Xen”
Storage 1000000 mb
Time zone this resource located 10
Cost of using processing in this resource 3.0 mips
Cost of using the memory in this resource 0.05 mips
Cost of using storage in this resource 0.001 mips
Cost of using Bandwidth in this resource 0.0 mips
Volume 14 • Issue 1
15
Energy Consumption
Energy consumption denotes the sum of energy consumption of all recourses used in the simulation to
send an alert from the source to the destination for each scenario. From the results presented in Figure
11, we can conclude that around 2000 Joules of energy was consumed by the single-route strategy
adopted by the previous work E-HAMC presented by S2 in which the edge device has to pull the
data through long bandwidth links to the cloud if the primary fog server fails to send the critical alert.
In contrast, in our proposed multi-route plan we notice that utilizing an alternative closer fog server
resources consumes less energy estimated by half the energy consumed by the single route strategy
(1000 Joules) in S3. Consecutively, we can notice that the energy consumption slightly increases by
utilizing more fog devices in S4 and S5 as the Energy Calculator placed at the edge device calculates
more traffic caused by the re-transmission of the alert to the available fog servers each time a failure
occurs. The energy consumed by cloud deployment is greater because the output characteristics of
the cloud such that it emits tuples with greater CPU requirements and at a higher frequency than the
other scenarios of the proposed work to send the alert to the hospital. This leads to increased use
of the cloud hosting Energy Calculator module and causes greater energy consumption. Thus, we
conclude that our multi-route approach outperforms the previous approach in all scenarios. It provides
less energy consumption levels for sending the alert to the hospital because all fog servers deployed
in our scenarios were placed closer to the user compared to the cloud in the single-rout strategy. We
should also realize that by utilizing more resources in the fog layer the energy consumption in these
devices increases, too. However, it deserves notice that operating on the fog will not only consume
less energy, but also reduces the total energy consumed by the cloud as a part of the processing
burden is carried out by the fog.
Network Usage
Network usage represents the amount of data being disseminated across the network from the source
to the destination through different scenarios. Figure 12 demonstrates the experimental results of
network usage achieved by the scenarios considered in this research work. According to the results,
we notice less degree of network usage achieved by that the single-route plan adopted by E-HAMC
i.e. 5 Megabytes. That is due to the less amount of traffic needed to re-send the alert through the
predefined cloud (as one-hop). Therefore, few resources have been utilized to accomplish the task
compared to our proposed plan. Unlike latency and energy consumption, our approach has achieved
a reasonably higher degree of network usage in S3 i.e. 6.7 Megabytes. Furthermore, the value of
network use noticeably increases as utilizing more fog nodes in both S4 and S5 consecutively.
However, this result can be compromised as deploying an alternative route through the next nearest
fog server (S3) provides less latency which is the main concern of these real-time services that are
Figure 10. Latency
Volume 14 • Issue 1
16
related to human life. From the experimental results, we can conclude that increasing the number
of fog devices implemented in the system significantly increases the load on the network as more
fog devices are being utilized in each scenario. As shown in Figure 12 when more fog devices are
engaged, the network use considerably increases. This result can also be interpreted as a demonstration
of the scalability of fog-based applications. Uncontrolled growth of network use in case of fog-based
execution may result in wasting the resources which leads to further degradation of the application’s
performance. Such situations can be better avoided by adopting fewer fog devices to keep the process
of the information closer to its source as can as possible. Although these modules are separated by
long latency links in the E-HAMC scenario, the alert being transferred takes a single fog route to
the destination. However, the proposed work engages many fog servers that place the concentration
calculator on several fog servers, thereby leading to a high degree of network use.
This paper addressed the issue of service disruption for fog-based healthcare monitoring systems
when node failure occurred. The proposed solution applies the path redundancy concept which is a
way to ensure different dependability objectives, namely: availability, reliability, and quality of service
(QoS). Moreover, the proposed solution can also be applied to other domains in which time is the most
critical factor that needs to be considered in these applications such as disaster detection and accident
detection. Several real-life scenarios have been developed and discussed to evaluate the proposed
multi-route plan strategy. Finally, the simulation results indicate that deploying a multi-routing through
Figure 12. Network Usage
Figure 11. Energy Consumption
Volume 14 • Issue 1
17
another fog node to take place in failure cases would achieve a better latency to deliver an alert to
the suitable department when node failure occurred. Moreover, the results indicate that deploying a
fog-based multi-route plan as an alternative route would consume less energy and provide reasonable
network use which makes our proposed strategy meets the reliability requirements of the system.
This research received no specific grant from any funding body in the public, commercial, or not-
for-profit sectors.
Volume 14 • Issue 1
18
Aazam, M., & Huh, E. N. (2015). E-HAMC: Leveraging Fog computing for emergency alert service. 2015 IEEE
International Conference on Pervasive Computing and Communication Workshops, PerCom Workshops 2015,
518–523. doi:10.1109/PERCOMW.2015.7134091
Ahmad, M., Amin, M. B., Hussain, S., Kang, B. H., Cheong, T., & Lee, S. (2016). Health Fog: A novel framework
for health and wellness applications. The Journal of Supercomputing, 72(10), 3677–3695. doi:10.1007/s11227-
016-1634-x
Al-Khafajiy, M., Webster, L., Baker, T., & Waraich, A. (2018). Towards fog driven IoT healthcare:
Challenges and framework of fog computing in healthcare. ACM International Conference Proceeding Series.
doi:10.1145/3231053.3231062
Ali, S., & Ghazal, M. (2017). Real-time Heart Attack Mobile Detection Service (RHAMDS): An IoT Use
Case for Software Defined Networks. In 2017 IEEE 30th Canadian Conference on Electrical and Computer
Engineering (CCECE) (pp. 1-6). IEEE.
Andriopoulou, F., Dagiuklas, T., & Orphanoudakis, T. (2017). Integrating IoT and Fog Computing for Healthcare
Service Delivery. In G. Keramidas, N. Voros, & M. Hübner (Eds.), Components and Services for IoT Platforms:
Paving the Way for IoT Standards (pp. 213–232). doi:10.1007/978-3-319-42304-3_11
Awad, A., Khanapi, M., Ghani, A., & Arunkumar, N. (2019). Enabling technologies for fog computing in healthcare
IoT systems. Future Generation Computer Systems Journal, 90, 62–78. doi:10.1016/j.future.2018.07.049
Bakhshi, Z., & Rodriguez-navas, G. (2019). A Preliminary Roadmap for Dependability Research in Fog
Computing. 17th International Workshop on Real-Time Networks–RTN 2019, 16(4), 14–19.
Bonomi, F., Rodolfo, M., Natarajan, P., & Zhu, J. (2014). Fog Computing: A Platform for Internet of Things and
Analytics. Studies in Computational Intelligence, 546, 1–19. doi:10.1007/978-3-319-05029-4_7
Cao, Y., Hou, P., Brown, D., Wang, J., & Chen, S. (2015). Distributed analytics and edge intelligence: Pervasive
health monitoring at the era of fog computing. Proceedings of the International Symposium on Mobile Ad Hoc
Networking and Computing (MobiHoc), 43–48. doi:10.1145/2757384.2757398
Craciunescu, R., Mihovska, A., & Halunga, S. (2015). Implementation of Fog Computing for Reliable E- Health
Applications. 2015 49th Asilomar Conference on Signals, Systems and Computers, 459–463.
Dar, B. K., Shah, M. A., Shahid, H., & Naseem, A. (2019). Fog computing based automated accident detection
and emergency response system using Android smartphone. 2018 14th International Conference on Emerging
Technologies, ICET 2018, 1–6.
Dar, B. K., Shah, M. A. L. I., Islam, S. U. L., Mussadiq, S., & Khan, S. (2019). Delay-Aware Accident Detection
and Response System Using Fog Computing. IEEE Access : Practical Innovations, Open Solutions, 7, 70975–
70985. doi:10.1109/ACCESS.2019.2910862
Dutta, J., & Roy, S. (2017). IoT-fog-cloud based architecture for smart city: Prototype of a smart building.
Proceedings of the 7th International Conference Conf luence 2017 on Cloud Computing, Data Science and
Engineering, 237–242. doi:10.1109/CONFLUENCE.2017.7943156
Dziak, D., Jachimczyk, B., & Kulesza, W. J. (2017). IoT-based information system for healthcare application:
Design methodology approach. Journal of Applied Sciences (Switzerland), 7(6), 596. doi:10.3390/app7060596
Fitzgerald, E., Pióro, M., & Tomaszewski, A. (2018). Energy-Optimal Data Aggregation and Dissemination
for the Internet of Things. IEEE Internet of Things Journal, 5(2), 955–969. doi:10.1109/JIOT.2018.2803792
Fratu, O., Pena, C., Craciunescu, R., & Halunga, S. (2015). Fog computing system for monitoring Mild Dementia
and COPD patients - Romanian case study. 2015 12th International Conference on Telecommunications in
Modern Satellite, Cable and Broadcasting Services, TELSIKS 2015, 123–128.
Gao, L., Luan, T. H., Yu, S., Zhou, W., & Liu, B. (2017). FogRoute: DTN-Based Data Dissemination Model in
Fog Computing. IEEE Internet of Things Journal, 4(1), 225–235. doi:10.1109/JIOT.2016.2645559
Volume 14 • Issue 1
19
Gia, T. N., Jiang, M., Sarker, V. K., Rahmani, A. M., Westerlund, T., Liljeberg, P., & Tenhunen, H. (2017). Low-
cost fog-assisted health-care IoT system with energy-efficient sensor nodes. 2017 13th International Wireless
Communications and Mobile Computing Conference, IWCMC 2017, 1765–1770.
Gupta, H., Vahid Dastjerdi, A., Ghosh, S. K., & Buyya, R. (2017). iFogSim: A toolkit for modeling and simulation
of resource management techniques in the Internet of Things, Edge and Fog computing environments. Software,
Practice & Experience, 47(9), 1275–1296. doi:10.1002/spe.2509
Huang, H., & Guo, S. (2019). Proactive Failure Recovery for NFV in Distributed Edge Computing. IEEE
Communications Magazine, 57(5), 131–137. doi:10.1109/MCOM.2019.1701366
Khan, S., Parkinson, S., & Qin, Y. (2017). Fog computing security: A review of current applications and security
solutions. Journal of Cloud Computing, 6(1).
Kher, R. K. (2016). Mobile and E-Healthcare: Recent Trends and Future Directions. Journal of Health & Medical
Economics, 2(3), 1–8. doi:10.21767/2471-9927.100010
Laprie, J., Randell, B., Landwehr, C., & Member, S. (2004). Basic Concepts and Taxonomy of Dependable
and Secure Computing. IEEE Transactions on Dependable and Secure Computing, 1(1), 11–33. doi:10.1109/
TDSC.2004.2
Li, W., Yang, Y., & Yuan, D. (2016). Ensuring Cloud Data Reliability with Minimum Replication by Proactive
Replica Checking. IEEE Transactions on Computers, 65(5), 1494–1506. doi:10.1109/TC.2015.2451644
Masip-Bruin, M., Marin-Tordera, E., Alonso, A., & Garcia, J. (2016). Fog-to-cloud Computing (F2C): The
key technology enabler for dependable e-health services deployment. 2016 Mediterranean Ad Hoc Networking
Workshop, Med-Hoc-Net 2016 - 15th IFIP MEDHOCNET 2016.
Naranjo, P. G. V., Shojafar, M., Mostafaei, H., Pooranian, Z., & Baccarelli, E. (2017). P-SEP: A prolong stable
election routing algorithm for energy-limited heterogeneous fog-supported wireless sensor networks. The Journal
of Supercomputing, 73(2), 733–755. doi:10.1007/s11227-016-1785-9
Nath, S. B., Gupta, H., Chakraborty, S., & Ghosh, S. K. (2018). A Survey of Fog Computing and Communication:
Current Researches and Future Directions. IEEE Communication Surveys and Tutorials, 1–47.
Park, M., Choi, E. J., Jeong, M., Lee, N., Kwak, M., Lee, M., Lim, E.-C., Nam, H., Kim, D., Ku, H., Yang, B. S.,
Na, J., Jang, J. S., Kim, J. Y., & Lee, W. (2019). ICT-based comprehensive health and social-needs assessment
system for supporting person-centered community care. Healthcare Informatics Research, 25(4), 338–343.
doi:10.4258/hir.2019.25.4.338 PMID:31777678
Paul, A., Pinjari, H., Hong, W. H., Seo, H. C., & Rho, S. (2018). Fog computing-based IoT for health monitoring
system. Journal of Sensors, 2018, 2018. doi:10.1155/2018/1386470
Sanna, A., Serafin, R., & Maganetti, N. (2014). E-Health: Current Status and Future Trends. Lecture Notes
in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in
Bioinformatics), 6545, 697–720.
Satria, D., Park, D., & Jo, M. (2017). Recovery for overloaded mobile edge computing. Future Generation
Computer Systems, 70, 138–147. doi:10.1016/j.future.2016.06.024
Singh, S., Bansal, A., Sandhu, R., & Sidhu, J. (2018). Fog computing and IoT based healthcare support service for
dengue fever. International Journal of Pervasive Computing and Communications, 14(2), 197–207. doi:10.1108/
IJPCC-D-18-00012
Sood, S. K., & Mahajan, I. (2018). A Fog-Based Healthcare Framework for Chikungunya. IEEE Internet of
Things Journal, 5(2), 794–801. doi:10.1109/JIOT.2017.2768407
Souza, V. B., Masip-Bruin, X., Marin-Tordera, E., Ramirez, W., & Sanchez-Lopez, S. (2017). Proactive vs
reactive failure recovery assessment in combined Fog-to-Cloud (F2C) systems. IEEE International Workshop
on Computer Aided Modeling and Design of Communication Links and Networks, CAMAD. doi:10.1109/
CAMAD.2017.8031528
Ullah, A., Sehr, I., Akbar, M., & Ning, H. (2018). FoG assisted secure De-duplicated data dissemination in smart
healthcare IoT. Proceedings - 2018 IEEE International Conference on Smart Internet of Things, SmartIoT 2018,
166–171. doi:10.1109/SmartIoT.2018.00038
Volume 14 • Issue 1
20
Nour El Imane Zeghib received a master’s degree in Information Technology from International Islamic University
Malaysia, Malaysia in 2020. Her research interests include Fog Computing, IoT, and data management in cloud
computing.
Ali A. Alwan is currently an assistant professor at School of Theoretical and Applied Science, Ramapo College of
New Jersey, United States. He received his Master of Computer Science in 2009 and Ph.D. in Computer Science
in 2013 from Universiti Putra Malaysia (UPM), Malaysia. His research interests include databases (mobile,
distributed, and parallel), preference queries, web databases, probabilistic, incomplete and uncertain databases,
query processing and optimization, data management, data integration, location-based social networks (LBSN),
recommendation system, data mining, database in Cloud, Big data management, and crowd-sourced database.
Abedallah Zaid Abualkishik received the Master and Ph.D. degree in Software Engineering. Dr Abedallah is
passionate about the managerial process of software development, coding themes, Big Data, Blockchain, and
Data Science. His research interests include software functional size measurement, conversion, cost estimation,
empirical software engineering, database, Big Data, and Data Science. Currently, he is working as an associate
professor at the college of computer information technology, American University in the Emirates, Dubai, UAE. He
serves the scientific community as a regular reviewer for several high-impact journals.
Yonis Gulzar received a master’s degree in computer science from Bangalore University, India, in 2013, and the
Ph.D. degree in Computer Science from International Islamic University Malaysia, in 2018. He is currently an
Assistant Professor with King Faisal University (KFU), Saudi Arabia. Before joining KFU, he was a part-time Lecturer,
a Teaching Assistant, and a Research Assistant with the Department of Computer Science, International Islamic
University Malaysia, Malaysia. His research interests include preference queries, skyline queries, probabilistic and
uncertain databases, query processing and optimization and management of incomplete data, data integration,
location-based social networks (LBSN), recommendation systems, and data management in cloud computing.
Xu, Q., Li, S., Zhou, Y., & Yang, X. (2018). Delay Analysis of Wireless Body Sensor Network Based on
Congestion Control. Proceedings of 2018 IEEE 3rd Advanced Information Technology, Electronic and Automation
Control Conference, IAEAC 2018, 2076–2080. doi:10.1109/IAEAC.2018.8577913
Zhan, Y., Xia, Y., & Vasilakos, A. V. (2019). Future directions of networked control systems: A combination of
cloud control and fog control approach. Computer Networks, 161, 235–248. doi:10.1016/j.comnet.2019.07.004
Content uploaded by Ali A. Alwan
Author content
All content in this area was uploaded by Ali A. Alwan on Jul 27, 2022
Content may be subject to copyright.
