See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/321826426

Cloud-Fog Interoperability in IoT-enabled Healthcare Solutions

Preprint · January 2018

DOI: 10.1145/3154273.3154347

CITATIONS READS
10 2,659

3 authors:

Md. Redowan Mahmud Fernando Koch

University of Melbourne University of Melbourne
17 PUBLICATIONS   260 CITATIONS    117 PUBLICATIONS   895 CITATIONS   

SEE PROFILE SEE PROFILE

Rajkumar Buyya
University of Melbourne
854 PUBLICATIONS   55,901 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Doutorado View project

Computational Intelligence in Perceivable Security View project

All content following this page was uploaded by Md. Redowan Mahmud on 15 February 2018.

The user has requested enhancement of the downloaded file.

 Cloud-Fog Interoperability in IoT-enabled Healthcare Solutions
Redowan Mahmud, Fernando Luiz Koch, Rajkumar Buyya
Cloud Computing and Distributed Systems (CLOUDS) Laboratory
School of Computing and Information Systems
The University of Melbourne, Australia.

ABSTRACT 1 INTRODUCTION
The issue of utilizing Internet of Things (IoT) in Healthcare solu- In recent years, the concept of IoT has been widely adopted on dif-
tions relates to the problems of latency sensitivity, uneven data ferent aspects of Healthcare specially in applications that involve
load, diverse user expectations and heterogeneity of the applica- ubiquitous sensors and actuators communicating through Wireless
tions. Current explorations consider Cloud Computing as the base Sensor Network (WSN) along with solutions for real-time data anal-
stone to create IoT-Enable solution. Nonetheless, this environment ysis and recommendation. When applied in critical scenarios, the
entails limitations in terms of multi-hop distance from the data services are very latency-sensitive and demand faster processing of
source, geographical centralized architecture, economical aspects, the generated data [18]. Moreover, the large utilisation of sensors,
etc. To address these limitations, there is a surge of solutions that mobility, and geographic distribution lead to issues of data volume,
apply Fog Computing as an approach to bring computing resources velocity, and variation, along with requirements for accuracy, secu-
closer to the data sources. This approach is being fomented by rity, Quality of Service (QoS), user expectations, and operational
the growing availability of powerful edge computing at lower cost costs [6].
and commercial developments in the area. Nonetheless, the imple- Cloud computing services have been widely accredited to sup-
mentation of Cloud-Fog interoperability and integration implies in port IoT-enabled Healthcare solutions, as presented in [4, 12, 19,
complex coordination of applications and services and the demand 23, 24], providing solutions for scalability, data analysis and relia-
for intelligent service orchestrations so that solutions can make bility [14]. However, the geographical centralisation of Cloud data
the best use of distributed resources without compromising stabil- centres requires that data collected from sensors to be transmitted
ity, quality of services, and security. In this paper, we introduce a through multi-hop distance for processing, which adversely affects
Fog-based IoT-Healthcare solution structure and explore the inte- the latency sensitivity of the solutions. Moreover, management of
gration of Cloud-Fog services in interoperable Healthcare solutions Cloud resources in heterogeneous Healthcare environments require
extended upon the traditional Cloud-based structure. The scenarios complex management tasks to avoid continuous revision of that
are evaluated through simulations using the iFogSim simulator and resource allocation in response to uneven and uncertain data loads
the results analyzed in relation to distributed computing, reduction coming from Healthcare solutions.
of latency, optimization of data communication, and power con- Fog computing is a promising solution in this scenario by ex-
sumption. The experimental results point towards improvement in ploring lightweight and customisable supplementary computing
instance cost, network delay and energy usage. resources closer to the IoT data source in Healthcare solutions
[15]. In this solution, traditional edge computing devices such as
CCS CONCEPTS switches, routers, low-profile computing devices, etc are equipped
• Computer systems organization → Embedded systems; Re- with computational infrastructure, services, and management mod-
dundancy; • Networks → Network reliability; els to implement local lean applications [3]. As result, some data
processing can executed closer to the data source, distributing re-
KEYWORDS source demands, reducing the need for multi-hope data communica-
Internet of Things, HealthCare, Fog computing, Cloud computing, tion, reducing latency, and promoting service flexibility. Although
Interoperable architecture Fog resources are constrained in terms of energy and computa-
tional power, they are flexible enough to customise according to
ACM Reference Format: the application context. The complexity arising in managing and
Redowan Mahmud, Fernando Luiz Koch, Rajkumar Buyya. 2018. Cloud-Fog
operation distributed computing scenarios, cope with the combi-
Interoperability in IoT-enabled Healthcare Solutions. In ICDCN ’18: 19th
International Conference on Distributed Computing and Networking, January
nation of variable demands and constrained computing resources,
4–7, 2018, Varanasi, India. ACM, New York, NY, USA, 10 pages. ensure performance, stability, and security.
https://doi.org/10.1145/3154273.3154347 Several applications of Fog computing in Healthcare solutions
exist in the literature, such as [1, 21, 22]. However, there is still a
Permission to make digital or hard copies of all or part of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
demand for methods to promote interoperability of services that
for profit or commercial advantage and that copies bear this notice and the full citation would allow for settle applications directly from Cloud elements
on the first page. Copyrights for components of this work owned by others than ACM into Fog elements coping with the inherent architectural differences.
must be honored. Abstracting with credit is permitted. To copy otherwise, or republish,
to post on servers or to redistribute to lists, requires prior specific permission and/or a In this context, the contribution of the paper is listed as:
fee. Request permissions from [email protected].
ICDCN ’18, January 4–7, 2018, Varanasi, India • A Fog-based IoT-Healthcare solution structure (system ar-
© 2018 Association for Computing Machinery. chitecture and application model) that is interoperable with
ACM ISBN 978-1-4503-6372-3/18/01. . . $15.00
https://doi.org/10.1145/3154273.3154347 general Cloud-based Healthcare solutions.
ICDCN ’18, January 4–7, 2018, Varanasi, India R. Mahmud et al.

• A reference architecture for Cloud-Fog service integration and inter-application interoperability are considered as the main
and orchestration from the perspective of interoperable IoT- features of the prototype.
Healthcare solutions. Renta et al. [23] focuses on storing Healthcare data received from
• Performance evaluation of Fog-based IoT-Healthcare solu- distributed IoT devices to remote Cloud. The data management
tions through simulation studies using iFogSim [8] in respect system lets the IoT devices to collect user critical data in real-time.
of deadline satisfied service delivery, cost, energy usage and Cloud-enabled techniques ensure faster processing of the stored
service distribution. data so that subscribed users can get quick notification during
The paper is organized as follows: Section 2 presents the mo- emergency. An alert service also runs in the system based on the
tivation through an analyze of the state-of-the-art and Section 3 predefined health rules and users reaction plan.
describes the general Cloud-based and proposed Fog-based Health- Mahmud et al. [19] presents a framework that enables data anal-
care solution structure. Section 4 introduces a integrated reference ysis and visualization for predicting health-shocks based on prede-
architecture of Cloud-Fog platform for IoT-Healthcare. Then, In fined health dataset. The framework is based on Cloud computing
Section 5, performance evaluation of the proposed solution in dif- platform. It incorporates Amazon web services (AWS), geographi-
ferent use configurations through simulation scenarios developed cal information systems (GIS) and Fuzzy rule based summarization
upon iFogSim. The paper concludes with Section 6. technique. The framework can classify health-shocks with interpret-
ability and accuracy using a data model. Besides it can explain the
2 MOTIVATION causal factors of health-shocks through linguistic rules.
Chen et al. [12] targets the security aspects of medical data
sharing through Cloudlet. Encryption is used in data collection. A
Table 1: Summary of the literature study
trust model is developed to identify reliable destinations (hospital,
doctor chambers, etc.) for sharing the data. The trust model helps
Work Cloud- Fog- Interoperable
based based (Cloud
Fog) to connect patients and medical professionals as well. In the data
Doukas et al. [4] X sharing destinations, data are segmented into three parts to store in
Renta et al. [23] X remote Cloud. In the whole process an Intrusion Detection System
Mahmud et al. [19] X
Chen et al. [12] X (IDS) works actively to prevent malicious attacks.
Zhang et al. [24] X Zhang et al. [24] proposes a patient-centric cyber-physical sys-
Fazio et al. [13] X tem named Health-CPS aiming to ensure convenient and efficient
Peddi et al. [20] X
Jindal et al. [11] X Healthcare service. The Health-CPS solely depends on Cloud com-
Muhammad et al. [7] X puting and data analytics to handle the Big data related issues of
Gupta et al. [9] X different Healthcare applications. The system is composed of sev-
Hossain et al. [10] X
Gia et al. [21] X
eral layers such as data collection layer, data management layer and
Gia et al. [22] X data oriented service layer. The system collects data in a unified
Ahmad et al. [1] X X standard. It supports distributed storage and parallel processing.
Chakraborty et al. [17] X X
Dubey et al. [5] X X
Fazio et al. [13] designs an e-health Remote Patient Monitoring
Negash et al. [16] X (RPM) system in a Cloud platform named FIWARE. The authors
Rahmani et al. [2] X emphasize to speed up the development of the RPM system with
Interoperable Cloud-Fog X X the facilities provided by FIWARE. The system targets to assist the
IoT-Healthcare (This work)
patients and optimize the responsibilities of the medical profession-
als. The association of FIWARE Cloud to the RPM system enhances
Table 1 provides a brief summary of our survey on the state- modularity, scalability and flexibility.
of-the-art in applying Cloud and Fog Computing paradigms in Peddi et al. [20] proposes a Cloud-based mobile e-health calorie
Healthcare solutions. As observed, there is a growing interest in the system. The system is able to classify different food objects from the
utilization of Fog computing in IoT-Healthcare solutions, however meal and with high accuracy can compute the overall calorie. The
a clear predominance around Cloud-based approaches. system incorporates computation offloading from Mobile e-health
Gia et al. [21] presents a Fog computing based health monitor- applications to the Cloud. In Cloud, it employs a broker entity to
ing system equipped with data mining, storing and notification manage the resources efficiently so that accurate outcome can be
facilities at the edge of the architecture. The authors explored the generated in tolerable latency. The broker imposes dynamic cloud
applicability of such arrangement in ECG feature extraction. A tem- allocation mechanism to engage and free computing instance in
plate based feature extraction technique is used in smart gateways real-time according to the demand.
to analyse the ECG signals. The experimental results indicate to Jindal et al. [11] propose a technique to calculate heart rate using
efficient bandwidth usage and service delivery. the smart phone embedded sensors (accelerometer) and Photo-
Doukas et al. [4] presents the online data management and pro- PlethysmoGraphy (PPG) signals. The technique is composed of
cessing of IoT-enabled pervasive healthcare applications on Cloud. three step data processing. Later, the technique requires Cloud asso-
The implemented prototype receives patient data from the IoT de- ciation to select perfect PPG signals through deep learning mecha-
vices and forwards to the Cloud for processing. Security issues are nisms and classify the signals to estimate heart rate. The technique
observed during every communications among the entities. Rep- is evaluated by processing the TROIKA dataset. According to the
resentational State Transfer (REST) API based access, scalability authors, the technique is able to predict heart rate accurately.
Cloud-Fog Interoperability in IoT-enabled Healthcare Solutions ICDCN ’18, January 4–7, 2018, Varanasi, India

Muhammad et al [7] discusses an IoT-Cloud-based Healthcare set of IoT devices directly associated with the patients. The cluster
solution for voice pathology monitoring of the users. The proposed of smart e-health gateways are supported with data analytics and
solution incorporates a voice pathology detection system that ap- adaptive configurations. The proposed system is able to monitor
plies local binary pattern on voice signal represented through Mel- the patients independent to his/her movement.
spectrum technique. A machine learning classifier is also used to Another Fog-based smart e-health gateway is presented by Rah-
conduct the pathology. According to the authors, the association mani et al. in [2]. The authors exploit possible placement of smart
of Cloud computing improves the accuracy and the accessibility of e-health gateway so that it can offer real-time local storage, data
the Healthcare solution to a certain extent. processing and data analysis. The Fog-based system can effectively
A Cloud-based IoT-enabled predictive physical activity analysis cope up with mobility, energy and reliability related issues. Based
model for the users is discussed by Gupta et al. in [9]. The model on the concept a prototype of smart e-health gateway named UT-
encompasses embedded sensors, Cloud computing and XML Web GATE is also developed. An IoT-based Early Warning Score (EWS)
services for faster, secure and reliable data collection, processing is used to evaluate the performance of the system.
and communication. The model is evaluated from different per- We observed that there is an opportunity to contribute with stud-
spectives (service adaptation, prediction analysis, efficiency and ies around the interoperability between Cloud and Fog-based plat-
security). The model also enables alert to the ailing person notifying forms. We target a Fog-based IoT-Healthcare solution where both
abnormality or complications during physical activity. the system architecture and application model follows the Cloud-
The impact of Healthcare Industrial IoT (HealthIIoT) in observing like structure to promote the interoperability. Besides, we explore
real-time health issues of aged and disable people is discussed by how Cloud-Fog interoperable IoT-Healthcare solution structure can
Hossain et al. in [10]. HealthIIoT is able to monitor, track, and store lead towards enhanced service integration and orchestration.
users healthcare data for consistent treatment. The HealthIIoT-
enabled framework collects ECG data from smart phones and other 3 HEALTHCARE SOLUTION STRUCTURE
sensors. Later send the collected to the cloud so that Doctors can
In this section, at first based on the literature study, we discuss
access and assess the data seamlessly. Cloud-based data analytics
a general structure (system architecture and application model)
is used to detect the abnormality and error of the health data.
of the Cloud-based IoT-Healthcare solutions. Later, a Fog-based
Another work of Gia et al.[22] presents a Fog-based health mon-
IoT-Healthcare solution structure in proposed which can be inter-
itoring system that can provide consistent remote monitoring of
operable with the Cloud-based solutions.
cardiac patients at low cost. The system is comprised of energy
efficient IoT sensors and smart gateways. The sensors collect ECG,
respiration rate and body temperature data and wirelessly send to 3.1 General Cloud-Based Solutions
the gateways for automatic analysis and notification. Besides, it can Almost every Cloud-based IoT-Healthcare solution follows a com-
assist in visualizing the outcome in an effective and re-usable way. mon system architecture and application model. They only differs
Ahmad et al. [1] proposes a Fog-based Healthcare framework. from the functionality of the applications.
The framework acts as an intermediate layer between the Cloud The system architecture of a Cloud-based IoT-Healthcare so-
and end IoT devices. It enhances data privacy and security at the lution (Fig. 1) is usually comprised of several entities;
edge level along with Cloud Access Security Broker (CASB). In IoT sensors or wearable devices: In Healthcare solution, hand held or
Fog environment, the framework can be deployed in a modular body connected devices for example; pulse oximeter, ECG monitor,
approach. In addition the framework is capable of aggregating data smart watches, etc. perceive health context of the users. These de-
from multiple sources with adequate cryptographic assessment. vices can connect themselves with other user premises equipment
Chakraborty et al. [17] discusses a Fog-based computation plat- through Bluetooth, ZigBee and Infrared transmission. Generally
form that can deal with latency-sensitive Health data. A program- the data sensing frequency of these devices are fixed and seamlessly
ming model is proposed to handle geographically distributed large generate health data once turned on. However, most of such devices
scale Healthcare applications. The model is evaluated through pro- are subject to resource and energy constraint.
cessing of heart rate related Healthcare data. According to the Smart phones: Smart phones are used in many Healthcare solutions
authors, the proposed Fog-based Healthcare solution improves data significantly. Since IoT devices are lack of networking and process-
accuracy, service delivery time and retains data consistency. ing capabilities, Smart phones assist them in providing application
Dubey et al. [5] discusses about service oriented architecture interface and sending generated data towards Cloud datacentre.
of Fog computing to validate and evaluate raw health data sensed Smart phones maintain a persistent communication with the IoT
through IoT devices. In the proposed system, resource constrained devices to receive the sensed data. The data receiving frequency
embedded computing instances conduct the data mining and the of smart phones can be adjustable through the application. The
data analysis. The instances are also responsible to identify impor- embedded sensors on Smart phone like accelerometer, Global Posi-
tant patterns form the Health data and forward them towards Cloud tioning System (GPS), etc. can perceive the contextual data.
for further storage and usage. The basic intension of the work is to Cloud datacentre: The Cloud datacentre is the premier platform for
highlight the big data processing with low power Fog resources. IoT-enabled Healthcare solutions. In addition to large scale com-
Negash et al. [16] focuses on implementing a smart e-health putation, it facilitates storage, utility services with reliability and
gateway for Fog computing to assist IoT-enabled Healthcare ser- scalability. Cloud resources (computational infrastructure, services,
vices. Such smart e-health gateways are placed in geographically etc.) are orchestrated in a structural way and are virtualized. The
distributed network and each gateway is responsible to manage a components within a Cloud datacentre those assist in Healthcare
ICDCN ’18, January 4–7, 2018, Varanasi, India R. Mahmud et al.

Like the system architecture, the application model of different

Cloud-based Healthcare solutions are also similar. The generalized
model of Cloud-based Healthcare applications can be described
VM
through a flowchart as show in Fig.2.
Virtual
Machines
Server
er
The Cloud-based Healthcare application are usually divided in
Cloud to two parts. One part runs in the Smart phones of the users and
Datacentre
another part executes in the VMs of Cloud. At the initiation of the
Resource Manager

application in user’s Smart phone, it asks for the authentication

information. It can include user password, biometric recognition.
Not only in the Smart phones, this authentication information is
also applied to the Cloud for privacy protection.
Through the application, Smart phones are seamlessly connected
with the IoT sensors or wearable devices to receive Health data
Smart Phones of the users. Smart phones itself can perform some preprocessing
Pulse Oximeter
over the data. However, for extensive processing, the data is sent
Handheld ECG to the Cloud securely. The secured transfer of data can be ensured
through application and network centric cryptography.
In the VMs of Cloud, the other part of the application receives
Figure 1: Cloud-based Healthcare system architecture data from the authorized Smart phones and conduct data abstrac-
tion. Through data abstraction, essential information are extracted
from the raw sensed data and represented in a simplified form so
solutions can be listed as follows: that they become suitable for further analysis.
• Resource Manager: Resource manager is responsible for coordinat- At the later phase, data analysis is conduct. Data analysis can
ing the Cloud resources while dealing with IoT-enabled Healthcare include aggregation of users historical data, data mining, pattern
data. It actually deploys, manages and monitors both the infras- recognition, feature extraction, template matching techniques. In
tructures and services that made-up the Healthcare solution. It can this case, association of relevant information from data repository
schedule, terminate and scale the resources according to the de- of the Healthcare solution can be required. In some cases, external
mand, load and context. It also ensures higher level access control third-party software services are also used for analysing the data.
to the resources. In addition, Resource Manager defines the depen- After analyzing the data, users Health condition is evaluated
dencies between resources so that they can be run and executed in based on the outcome of the analysis. This evaluation can be done
the correct order. either by comparing the analysed data result with predefined sta-
• Servers: The Cloud datacentres encapsulate a set of Servers that tistics or by direct association of the medical professionals.
can either be homogeneous or heterogeneous in respect of hard- The final outcome of the Healthcare application can be a data
ware configurations (memory, cores, capacity, and storage). In tuple of the user’s Health context or a notification message to the
Cloud-based Healthcare system, two types of Servers are used pre- users. In general, the Health context is preserved for consistent
dominantly; Application Server and Database Server. In Application monitoring of the user and the message is send back to the user’s
Server the backend applications, web-services are hosted whereas Smart phone. This cycle continues until user terminate the applica-
Database Server solely handles the data repository and associate tion at his/her Smart phone.
operations. In a Server, a group of policies set by the Resource Based on the applicability of the Healthcare solution, associate
Manager is implemented for allocating bandwidth, memory and application can perform stream or batch data processing. Resource
storage to the residing instances. The Sever-applied policies include
important information regarding the type of processing cores, data-
base sharding and their mutual sharing or replication among the Start Healthcare application
instances. Service provisioning and load balancing policies are also Data abstraction

observed within the server. User authentication

• Virtual Machines: Instances within a Server is termed as Virtual information

Data analysis
Machines (VM). Each VM has access to the hardware resources Receive data from IoT
devices Relevant information
provided by the host Server. A VM itself encapsulates some meta- from data repository

data regarding accessible memory, processor and storage size. In

Notification/Alert Result evaluation
Healthcare solutions, relevant applications and web services are
run in the VMs of Application server. The large volume of Health
Data store
data are managed distributively within the VMs of Database server. End Healthcare application

While operating a Healthcare solution, allocated instances of both

Cloud based
types communicate simultaneously. Images of the operating VMs Smart phone/devices
Virtual Machines
can be replicated so that the Healthcare solutions can be made fault
tolerant to a certain extent. Migration of the tasks are also possible
among the VMs. Figure 2: Cloud-based Healthcare application model
Cloud-Fog Interoperability in IoT-enabled Healthcare Solutions ICDCN ’18, January 4–7, 2018, Varanasi, India

requirements of the applications can also vary time to time with

Fog CN-5

Micro

the nature of incoming data load. Besides, a application can be run computing
instance

on a single VM or can be run in distributed manner on multiple Fog Cluster

VMs. However, most of the Cloud-based Healthcare applications Fog CN-1

Fog CN-2

perform serialized operation on the data. Parallel processing can

Micro Micro Micro
computing computing computing
Micro Micro
instance instance instance
computing computing

be initiated when the Healthcare solution is comprised of multiple instance instance

applications and deals with diverse Healthcare data. Cluster head node

3.2 Interoperable Fog-Based Solutions

Fog CN-3 Fog CN-4
Fog computing environment is comprised of specialized networking Micro Micro

devices named Fog nodes to perform diverse computational tasks at

computing computing
instance instance

the network edge (Fig. 3). Fog nodes are distributively arranged in
hierarchical Fog levels. A Fog node can be equipped with processing Figure 4: Fog-based Healthcare system architecture
cores, memory, storage and network bandwidth. The lower level
Fog nodes (Smart phones, cast devices, set top box, car media player,
etc.) reside very closer to the IoT devices and usually offer interfaces
of the associate applications. Therefore, for a particular Fog-based Multiple nodes from similar or different Fog levels can form
Healthcare solution, a lower level Fog node can be termed as the cluster among themselves with faster networking standards (Fig.4).
Application gateway node. Application gateway node can process While forming cluster inter-nodal communication latency are given
the sensed Health data or can forward to the upper level Fog nodes higher priority. In a cluster, some nodes execute the applications
named Computational nodes for processing. In a Fog node, resources while others either host database or maintains communication with
(cores, memory, storage, bandwidth, etc.) can be virtualized and the other clusters. Generally each Fog cluster is responsible for a
shared in the form of Micro Computing Instances (MCI). particular Healthcare solution. A single Healthcare solution can also
In Fog environment, all nodes are not kept computationally be run in multiple clusters. The isolated nodes those do not belong
active always. Computational unit of the Fog nodes can be turned to any cluster in such architecture only perform as a networking
off when the data load gets reduced and can be activated according device. The number of computationally active node within a cluster
to the demand. Hence, the Fog environment can be made scalable can be scaled up according to the load.
and energy efficient. Besides, on each communication link among In a cluster, all the secured inter and intra-cluster communica-
the nodes, security features can be applied for data privacy and tion are handled by a particular node named Cluster head node.
intrusion protection. Thus, reliable data transfer can be ensured. According to the general Fog architecture, each Fog node can re-
However, inherently Fog and Cloud differs from each other in ceive health data from other connected node. In a cluster, whenever
respect of resource capacity, capability and orchestration. Therefore, a node receives data, it checks the relevance of the data with the
Cloud-based IoT-Healthcare solutions loose interoperability when associate Healthcare solution and notifies the Cluster head node.
they are intended to place in Fog. To facilitate placement of Cloud- The Cluster head node based on the notification either forwards the
based IoT-Healthcare solutions in Fog environment, a cluster-based data to the corresponding cluster or schedules the data to MCIs of
Fog system architecture is discussed in the following paragraphs. the same cluster for processing according to the application model.
Besides, Cluster head node sets resource and service provision-
ing policies for the other nodes, balances the load among the nodes,
controls and secures access and communication, monitors activities
Cloud of the MCIs and preserve associate meta-data. In order to ensure
Datacentre consistency of the Healthcare solution during uncertain node fail-
ures, Cluster head node can replicate the image of MCIs from that
node to other node of the cluster. If a Cluster head node fails, an-
other node from the same cluster which is priorly defined can act
Fog
nodes as the Cluster head node. Cluster head node can also distribute its
responsibilities to the others ensuring no performance degradation.
Computational
nodes
Cluster
Moreover, MCIs from same node, cluster or Healthcare solutions
can share data and content with each other under the supervision
of corresponding Cluster head node. Since MCIs are not rich in
resource capacity, large scale health care applications and database
Application gateway are placed in them distributively. Allocated resources to an MCI can
nodes be provisioned dynamically based on the context of the Healthcare
IoT devices/ solution. Each MCI can be managed and configured independently
sensors
without violating QoS of the others.
The aforementioned cluster-based Fog architecture is resem-
blanced to the Cloud-based system architecture in some cases. The
Figure 3: Generalized Fog environment cluster itself symbolizes the Cloud datacentre where Fog nodes
ICDCN ’18, January 4–7, 2018, Varanasi, India R. Mahmud et al.

are deployed like the Servers. The Cluster head node performs as • Client Module: Client Module provides the initial interface of the
the Resource Manager of the cluster. Inside the Fog nodes, MCIs, corresponding application. Connected IoT devices send the data
similar to VMs, execute the Healthcare applications and handle the signal to the application through this module. Authentication, data
database operations. However, some features of such Fog-based receiving frequency calibration, data aggregation from multiple
Healthcare solution, mentioned in Table 2, enhances its applicability IoT sources are handled by this module. Data pre-prossessing is
to the real world. also done by this module so that scattered data signals from IoT
devices transform to the formatted raw data. In addition, based on
Table 2: Enhanced features of the proposed Fog system the combined response from the subsequent modules, the Client
Module can manage the activities of the associate actuators.
Facts Fog Cloud • Data Filtering Module: Raw data forwarded from the Client Module
Cluster Datacentre
Confederation Easy Hard
contains some additional data (authentication, application meta-
Number of Nodes/Servers High Few data, etc.) with actual Healthcare data. Data Filtering Module extract
Inclusion/Exclusion of Nodes/Servers Flexible Difficult the Healthcare data and discard its irrelevant parts so that they can
Proximity to data source (hop) One/Two Multi
Geographical orchestration Distributed Centralized
be fed into the subsequent module for processing.
Communication latency from data source Low High • Data Processing Module: The filtered data from Data Filtering Mod-
Real-time interaction Possible Hard ule is processed in this module. This module actually incorporates
Nodes Servers multiple tasks such as data analysis, comparison and result evalua-
Tolerance to failure High Low
Communication path/link Variable Fixed tion. To assist this module external data, techniques and software
Energy usage Less High components can be applied. In this case, necessary communications
Cluster Resource should be handled by the corresponding MCI, Fog node and cluster.
head node Manager
Tolerance to failure High Low • Event Handler Module: After processing the data in Data Process-
Communication path/link Variable Fixed ing Module, the outcome can invoke any event of interest. The
Energy usage Less High Event Handler Module identifies the most appropriate response to
MCIs VMs
Management/Maintenance Simple Complex that event. The Event Handler Module can preserve the response for
Subject to Under/Over provision Less Higher further usage or can send back to the Client module to determine
Configuration Adjustable Rigid the action against the response.
Price Low High
Since working dimension of all the modules are different, their
resource requirement varies from one to another. Client Module
The Cloud-like arrangement of the components and the en- should be placed closer to the IoT devices for better performance of
hanced features from different perspectives make this cluster-based the application. In Fog, it can be placed on the Application Gateway
Fog architecture efficient to run Cloud-based IoT-Healthcare solu- Node. The subsequent modules can be placed to the particular
tion observing desired QoS and affordable service cost. cluster for that Healthcare solution without violating the data flow.
However, in this Fog-based system architecture, nodes are dis- All the modules can be placed on different MCIs of a single node or
tributed and their underlying MCIs are constrained in capacity. can be placed on different nodes. However, Data Processing Module
Placement of large-scale Healthcare application on single MCI is require additional resources compared to others module and failure
not feasible. Besides, on a single node there may not be additional to provide necessary resources for this module can crate a service
MCIs to accommodate resources for the whole application. Place- bottleneck for the solution. The Cluster head node should be made
ment of IoT-Healthcare application in such system will not be as aware of this fact prior to place that module.
straightforward as the Cloud. Therefore, it is required to transform In constrained Fog environment, distributed development and
the Cloud-based application model to a Fog compatible one without deployment is the best way to manage large-scale IoT-Healthcare
affecting the generality and the task consistency. The Fog-based applications. The proposed application model facilitates modular
application model for a Healthcare solution is described below. development of the applications and the associate inter-module
In Fog, a single application can be considered as a collection data dependency paves the way for its distributed deployment in
of Application Modules. As mentioned earlier, each Cloud-based constrained Fog environment. Thus Cloud-based IoT-Healthcare
Healthcare application performs some common and serialized op- applications can be customised to place in Fog environment.
erations on the received data. Each Application Modules can be
designed in such a way so that it can perform atleast one partic-
ular operation on the data. Besides, MCIs can be provisioned to
filtered data
Data Filtering Data Processing
Module Module
execute at most one module within it. Based on the observation,
each Cloud-based Healthcare application can be segmented into raw data processed data

four Application Modules as shown in Fig.5. IoT device

data signal

The data-dependency among the modules is drawn with a unidi- Event

rectional sequential data flow. The inter-module data-dependency

Client Module Handler
response Module

delay can affect the application service delivery adversely. There- Actuator
action

fore, in defining the service delivery deadline collective inter-module

data-dependency delay of the modules should be given higher pri-
ority. The detail of modules can be described as follows: Figure 5: Fog-based Healthcare application model
Cloud-Fog Interoperability in IoT-enabled Healthcare Solutions ICDCN ’18, January 4–7, 2018, Varanasi, India

IoT Sensors Fog Computing Cloud Computing

Data & Health Applications Data & Health Applications

Edge orchestration Service

Si(c) Reports App1 Appn Reports
... App1 Appn
Si(c) ……...
Fog/Edge Services Cloud Services
Si(c)
Serv1 Servn Serv1 Servn
... ……...
Si(c) Services & Services &
Micro Computing Virtual Machines
Si(c) Configuration Instances Configuration
VM1 VMn
MCI1 MCIn ……...
Sensor ... Edge App
Health Sensors
Demands Infrastructure Demands Infrastructure

Figure 6: A reference architecture of Cloud-Fog integration for interoperable IoT-Healthcare solutions

4 INTEGRATED ARCHITECTURE to distribute the processing between Fog Computing and

Fig. 6 depicts our proposal for a reference architecture that provides Cloud Computing demand This model works based on Edge
the elements to support integration between Fog and Cloud comput- App Demands , which provide public manifest describing
ing infrastructure while supporting interoperable IoT-Healthcare the resources demands, in terms of computational power,
solutions. The goal is provide the basic models to construct end- services, and infrastructure, that Cloud Services publish so
to-end solutions including sensors, and distribute application and that the Edge Orchestration Services can evaluate if it can
services, like data analysis, machine learning, context inference, accommodate part of the processing on the Fog Computing.
and recommendation systems. The key research questions are: • Service Orchestration in Sensor-Edge Service Management:
creating services to correlate resource demand between ap-
• How to promote the integration of IoT/Sensors and dis-
plications and Sensors and promote adjustment o sensor con-
tributed service environments?
figurations to cope with application demands. This model
• How to support service orchestration of distributed service
works based on Sensor Demands, which is a public manifest
environments and Fog Computing, considering the require-
describing configuration demands for the associated sen-
ments for local service support, limitations of computational
sors to accommodate the processing requirements on the
resources, networking and communication, and the local
Fog Computing and Cloud Computing services; e.g config-
environment?
uration about rate of data sampling, interval between data
• What are the issues of security and privacy in this environ-
transmissions, maximum data batch sizes, collect accuracy,
ment? How to develop strategies for distributed analysis and
and others.
security measures?
• Distributed Health Care applications: orchestration and lo-
We envision the following challenges and opportunities of Cloud- cal distribution of services between Cloud Computing and
Fog-based services while integrating them for interoperable Health- Fog computing elements to promote intelligent Health Care
care solutions: applications. The data which is critical to the operations of
• Intelligent Health Sensors: implementing basic, self-configurable the local infrastructure will be analysed and processed im-
Computational Intelligence services at the sensor devices mediately by the edge computing layer. This item will help
themselves; even considering the potential limitation of com- to address the challenge of turning big data into smart data
puting capability, it would be possible to implement micro- and stringent patient privacy and security rules.
services for contextualization, learning, and self-adjustment • Security and Privacy Solutions: security must become fluid
of sensor devices, providing solution on the lower-end data and adaptive an IoT and Edge Computing environments.
collecting and analysis process, i.e. between Sensors and This line of research encompasses distributed services to
Edge Services. correlate data from multiple layers to infer security issues,
• Service Orchestration Cloud-Edge Service Management: creat- intrusion detection, behaviour deviances, privacy threats,
ing services to correlate resource demand and performance and others.This item will help to address the challenge of (3)
information from Cloud Computing and Edge Service struc- stringent patient privacy and security rules.
tures to coordinate resource allocation, service management, The goal of the Cloud-Fog Orchestration Process is to balance
and adaptation to improve service performance and accu- the best distribution of applications and services considering: (i)
racy. The model of Edge Orchestrations Services must be able current requirements of executing applications in terms e.g. of QoS,
to relate information about Fog Computing infrastructure, frequency of data requirement, etc; (ii) computational capacity and
e.g. existing computation power, available services, and oth- service availability on Fog Computing devices, and; (iii) existing
ers and try to accommodate local applications and Services sensor and tier possible configurations.
ICDCN ’18, January 4–7, 2018, Varanasi, India R. Mahmud et al.

5 PERFORMANCE EVALUATION 160

Cloud based solution

In order to demonstrate the feasibility of our proposed Fog-based 140 Fog based solution

Average network delay (ms)

IoT-Healthare solution and iteroperation with Cloud-based solution,
we simulate both the environment and integrated architecutre using
120

iFogSim [8] simulation toolkit. In two folds we conduct the simula- 100

tion. At first the performance of Fog cluster-based IoT-Healthcare

solution is compared to the Cloud-based solution in terms of net- 80

work delay, energy usage and cost considering computational re- 60

sources are sufficient in Fog. Later the service distribution in Cloud-

Fog integration is demonstrated under different number of sensors 40

and CPU utilization of the services while executing applications 20

considering limited computational resources in Fog. In simulation, 5 10 15 20 25 30

synthetic workload is used as the real-world workload to simu- Number of placed applications

late such environment in large scale is not currently available. The

simulation metrics are summarized in Table 3. Figure 7: Network delay in Fog and Cloud-based solution

Table 3: Simulation parameters

low as there exist multiple communication link between data source
Parameter Value and proximate computing components. Moreover, the Cluster head
Simulation Duration 300 sec node can control the data flow to reduce the network delay.
Cloud Datacentre: In Cloud-based Healthcare solution, usually a single VM exe-
Network latency from source 100 ms cutes an application whereas in Fog-based solution, multiple MCIs
Cost of VMs 0.8-0.12 $/min collectively execute an application. Compared to a VM, an MCI
Energy consumption of VMs 10-15 Mega is lightweight and consume less amount of energy. Therefore, the
Joules overall energy usage of MCIs while executing increasing number
Average VMs per Server 10-15 of Healthcare application is less than the VMs (Fig.8).
Fog cluster:
Network latency from source 10 ms 400

Average Intra-cluster network latency 5-10 ms 350

Cloud based solution

Cost of MCIs 0.01-0.03 $/min

Fog based solution

Energy consumption of MCIs 2-3 Mega Joules

300
Energy usage (MJ)

Average MCIs per Server 3-10 250

Applications service delivery deadline 250 ms - 650 ms 200

Average data processing time of applications in 180-200 ms

Fog with a particular service
150

Average data processing time of applications in 100-120 ms 100

Fog with a particular service in Cloud 50

Data sensing interval of the IoT sensors 200-600 ms

0

5 10 15 20 25 30

In the simulation experiment, Fog resources can host different Number of placed applications

number of services of various CPU utilization rate to process data

coming from distributed sensors trough associate applications. Here Figure 8: Energy usage in Fog and Cloud-based solution
we assume, the applications are scheduled to the services based on
the frequency distribution (data sensing interval) of the correspond- In addition, single Application Module in Fog requires less re-
ing sensors and when the capacity of Fog exceeds, the applications sources than the whole application. If an MCI is provisioned accord-
are forwarded to Cloud-based services for execution. We conduct ing to the requirements of a module, resource over provisioning
the experiments by varying number of the applications, number of is less likely to be happen. In Cloud-based solution, it is difficult
the sensors and CPU utilization rate of the services. to provision VMs in such way as the configuration of the VMs are
predefined. Therefore, service charge for MCIs can be adjustable
5.1 Case Scenarios according to the context of the module whereas for VMs, whatever
In remote Cloud-based solution, sharing same communication link the usage is, service charge required to pay for the whole. As a
by multiple Healthcare applications reduces the bandwidth seg- result, in Fog-based solution, total instances cost is less compared
ment, creates network congestion and higher data round-trip time. to the Cloud-based solution (Fig.9).
As a result, average network delay perceived by the applications However, in Cloud-Fog integration for interoperable Healthcare
becomes high in Cloud (Fig.7). Conversely, in Fog-based solution, solutions, it is very important to determine the load distribution of
average network delay for data availability to the applications is the services between Fog and Cloud. In this phase of performance
Cloud-Fog Interoperability in IoT-enabled Healthcare Solutions ICDCN ’18, January 4–7, 2018, Varanasi, India

20

Cloud based solution

Fog based solution

15
Total cost ($)

10

5 10 15 20 25 30

Number of placed applications

Figure 9: Instances cost in Fog and Cloud-based solution

evaluation, the integration of Cloud-Fog in service distribution is

discussed.
Fig.10 depicts the Cloud-Fog service distribution on increasing Figure 11: Integration of Cloud-Fog varying number of ser-
number of sensors. For lower number of sensors, associate applica- vice CPU utilization
tions can be handled by the Fog-based services. With the increasing
of sensors, the number of services in Fog also increases and until a
particular point, it is not required to send the applications to Cloud. For Cloud, the number of running services also increase with the
However Fog resources are not abundant. After reaching a maxi- CPU utilization rate of the services and number of sensors.
mum number of running services with specific CPU utilization rate,
for Fog resources it is not possible to accommodate more services 6 CONCLUSIONS AND FUTURE DIRECTIONS
and applications are required to send to Cloud for services. In this In past several years the Cloud computing and its association to
case, number of Fog running services become constant whereas the many aspects of education, industry and medical services has been
number of Cloud services begin to increase. studied extensively. Many potential and widely accepted Cloud-
based solutions are currently available. With the inclusion of mod-
20
In Cloud
ern techniques like the Internet of Things (IoT), these Cloud-based
In Fog solutions become more dynamic and user oriented. However, Cloud
itself has some limitation due to its geographically centralized ar-
Number of services

15
chitecture and multi-hop distance from the IoT data source. Real-
time interaction between user and computing platform often get
10
disrupted by these inconvenience of the Cloud. For IoT-enabled
Healthcare solutions sometimes failure of real-time interaction can
provoke life threatening incident. Therefore, a new computing par-
5
adigm named Fog is coined at the edge network. It helps to meet
the limitations of Cloud computing. However, due to differences
0
between these two computing platforms, available Cloud-based
10 20 30 40 50 60 70 80 90 100 solution can not be directly placed to Fog environment.
Number of sensors In this work, we analyzed the literature review and gets the moti-
vation to generalize the Cloud-based IoT-Healthcare solution struc-
Figure 10: Integration of Cloud-Fog varying number of sen- ture both in terms of system architecture and application model.
sors Later, we propose an interoperable Fog-based IoT-Healthcare solu-
tion that extends the general Cloud-based IoT-Healthcare solution
Fig. 11, shows number of available services in Fog computing structure with some enhanced features. The integration of both
platform and the Cloud-Fog integrated service distribution by vary- the interoperable solution structure is discussed through a refer-
ing CPU utilization rate of the services on a particular number of ence architecture. However, we analysis the performance of both
sensors (three scenarios of sensor number in this case; 50, 100, 200). solution structure and their integration through simulation studies
With low CPU utilization rate of services, Fog itself can handle in iFogsim. The performance of Fog-based solution is improved
majority of the applications for lower number of sensors. However, in terms of service distribution, instances cost, energy usage and
as the service CPU utilization rate and number of sensors increase, network delay.
with all the services available in Fog, the application demand can The proposed Fog-based IoT-Healthcare solution (system archi-
not be satisfied. In this case association of Cloud becomes necessary. tecture, application model) can be extended for further research.
 ICDCN ’18, January 4–7, 2018, Varanasi, India R. Mahmud et al.

Some potential future research scopes in this direction are listed 4th International Conference on Wireless Mobile Communication and Healthcare -
below: Transforming Healthcare Through Innovations in Mobile and Wireless Technologies
(MOBIHEALTH). 263–266.
• Intelligence at the Cluster: Artificial intelligence can be ap- [7] G. Muhammad and S. M. M. Rahman and A. Alelaiwi and A. Alamri. 2017. Smart
Health Solution Integrating IoT and Cloud: A Case Study of Voice Pathology
plied to the Fog cluster to predict the future data load, chances Monitoring. IEEE Communications Magazine 55, 1 (January 2017), 69–73.
of nodal failure, network topology and associate data flow. [8] Gupta, Harshit and Vahid Dastjerdi, Amir and Ghosh, Soumya K. and Buyya,
• Placement of the Application Modules: Different policies to Rajkumar. [n. d.]. iFogSim: A toolkit for modeling and simulation of resource
management techniques in the Internet of Things, Edge and Fog computing
place the Application Modules within the cluster can be environments. Software: Practice and Experience, 2017 ([n. d.]).
explored. Latency, QoE and QoS-aware module placement [9] Gupta, P. K. and Maharaj, B. T. and Malekian, Reza. 2017. A novel and secure IoT
can enhance the acceptability of Fog in practical world. based cloud centric architecture to perform predictive analysis of users activities
in sustainable health centres. Multimedia Tools and Applications 76, 18 (01 Sep
• Module sharing and re-usability: Applications belonging to 2017), 18489–18512.
the same Healthcare solution can share executing modules [10] M. Shamim Hossain and Ghulam Muhammad. 2016. Cloud-assisted Industrial
Internet of Things (IIoT) âĂŞ Enabled framework for health monitoring. Computer
among themselves. As a result computational load can be Networks 101, Supplement C (2016), 192 – 202. Industrial Technologies and
optimized. Besides, deployed modules of recently terminated Applications for the Internet of Things.
application can be re-used for other application. To perform [11] Jindal, Vasu. 2016. Integrating Mobile and Cloud for PPG Signal Selection to
Monitor Heart Rate During Intensive Physical Exercise. In In Proceedings of
such operations, necessary techniques and policies are re- International Conference on Mobile Software Engineering and Systems (MOBILESoft
quired to be developed. ’16). ACM, New York, NY, USA, 36–37.
• Cluster resource management: In order to meet increasing [12] M. Chen and Y. Qian and J. Chen and K. Hwang and S. Mao and L. Hu. 2017.
Privacy Protection and Intrusion Avoidance for Cloudlet-based Medical Data
service demand, Fog cluster is required to be scaled up. Effi- Sharing. IEEE Transactions on Cloud Computing PP, 99 (2017), 1–1.
cient approaches for managing resource can be investigated [13] M. Fazio and A. Celesti and F. G. MÃąrquez and A. Glikson and M. Villari. 2015.
Exploiting the FIWARE cloud platform to develop a remote patient monitoring
those can meet the service demand without scaling up the system. In In Proceedings of the IEEE Symposium on Computers and Communication
cluster. Such resource management approaches can reduce (ISCC). 264–270. https://doi.org/10.1109/ISCC.2015.7405526
the energy usage within the cluster and cost of deploying [14] M. Hassanalieragh and A. Page and T. Soyata and G. Sharma and M. Aktas
and G. Mateos and B. Kantarci and S. Andreescu. 2015. Health Monitoring and
additional resources. Management Using Internet-of-Things (IoT) Sensing with Cloud-Based Process-
• Pricing and billing: A framework for pricing and billing of ing: Opportunities and Challenges. In In Proceedings of the IEEE International
Fog cluster resources can be pursued. Incentive and reward Conference on Services Computing. 285–292. https://doi.org/10.1109/SCC.2015.47
[15] Mahmud, Redowan and Ramamohanarao, Kotagiri and Buyya, Rajkumar. 2017.
based resource sharing among multiple Fog cluster can also Fog Computing: A Taxonomy, Survey and Future Directions. In Internet of
be explored. Everything: Algorithms, Methodologies, Technologies and Perspectives, Di Martino
Beniamino, Yang Laurence, Kuan-Ching Li, and Esposito Antonio (Eds.). Springer.
• Mobility and Edge-centric affinity: The mobility of the user [16] Negash, Behailu and Gia, Tuan Nguyen and Anzanpour, Arman and Azimi, Iman
and edge-centric affinity of the applications should be han- and Jiang, Mingzhe and Westerlund, Tomi and Rahmani, Amir M. and Liljeberg,
dled together by the Fog cluster for better performance. Pasi and Tenhunen, Hannu. 2018. Leveraging Fog Computing for Healthcare IoT.
Springer International Publishing, Cham, 145–169.
• System prototype: We plan to develop a prototype of IoT- [17] S. Chakraborty and S. Bhowmick and P. Talaga and D. P. Agrawal. 2016. Fog
Healthcare solution for Sleep Apnea analysis comprising the Networks in Healthcare Application. In In Proceedings of 13th IEEE International
interoperable and integrated Cloud-Fog architecture. Conference on Mobile Ad Hoc and Sensor Systems (MASS). 386–387.
[18] S. M. R. Islam and D. Kwak and M. H. Kabir and M. Hossain and K. S. Kwak. 2015.
The Internet of Things for Health Care: A Comprehensive Survey. IEEE Access 3
ACKNOWLEDGEMENT (2015), 678–708.
[19] Shahid Mahmud and Rahat Iqbal and Faiyaz Doctor. 2016. Cloud enabled data
The authors would like to thank Verra Mukty for her initial work analytics and visualization framework for health-shocks prediction. Future
on integrating pulse oximeter with the Cloud-based Healthcare Generation Computer Systems 65, Supplement C (2016), 169 – 181. Special Issue
on Big Data in the Cloud.
solution for Sleep Apnea analysis hosted in Microsoft Azure. [20] Sri Vijay Bharat Peddi and Pallavi Kuhad and Abdulsalam Yassine and Parisa
Pouladzadeh and Shervin Shirmohammadi and Ali Asghar Nazari Shirehjini.
2017. An intelligent cloud-based data processing broker for mobile e-health
REFERENCES multimedia applications. Future Generation Computer Systems 66, Supplement C
[1] Ahmad, Mahmood and Amin, Muhammad Bilal and Hussain, Shujaat and Kang, (2017), 71 – 86.
Byeong Ho and Cheong, Taechoong and Lee, Sungyoung. 2016. Health Fog: a [21] T. N. Gia and M. Jiang and A. M. Rahmani and T. Westerlund and P. Liljeberg
novel framework for health and wellness applications. The Journal of Supercom- and H. Tenhunen. [n. d.]. In In Proceedings of the IEEE International Conference
puting 72, 10 (01 Oct 2016), 3677–3695. on Computer and Information Technology; Ubiquitous Computing and Communi-
[2] Amir M. Rahmani and Tuan Nguyen Gia and Behailu Negash and Arman An- cations; Dependable, Autonomic and Secure Computing; Pervasive Intelligence and
zanpour and Iman Azimi and Mingzhe Jiang and Pasi Liljeberg. 2017. Exploiting Computing.
smart e-Health gateways at the edge of healthcare Internet-of-Things: A fog [22] T. Nguyen Gia and M. Jiang and V. K. Sarker and A. M. Rahmani and T. Westerlund
computing approach. Future Generation Computer Systems (2017). and P. Liljeberg and H. Tenhunen. 2017. Low-cost fog-assisted health-care IoT
[3] Bonomi, Flavio and Milito, Rodolfo and Zhu, Jiang and Addepalli, Sateesh. 2012. system with energy-efficient sensor nodes. In In Proceedings of 13th International
Fog Computing and Its Role in the Internet of Things. In In Proceedings of the Wireless Communications and Mobile Computing Conference (IWCMC). 1765–
First Edition of the MCC Workshop on Mobile Cloud Computing (MCC ’12). ACM, 1770.
13–16. [23] Tsiachri Renta, Pelagia and Sotiriadis, Stelios and Petrakis, Euripides G.M. 2017.
[4] C. Doukas and I. Maglogiannis. 2012. Bringing IoT and Cloud Computing towards Healthcare Sensor Data Management on the Cloud. In In Proceedings of the 2017
Pervasive Healthcare. In In Proceedings of the Sixth International Conference on Workshop on Adaptive Resource Management and Scheduling for Cloud Computing
Innovative Mobile and Internet Services in Ubiquitous Computing. 922–926. (ARMS-CC ’17). ACM, New York, NY, USA, 25–30.
[5] Dubey, Harishchandra and Yang, Jing and Constant, Nick and Amiri, Amir Mo- [24] Y. Zhang and M. Qiu and C. W. Tsai and M. M. Hassan and A. Alamri. 2017.
hammad and Yang, Qing and Makodiya, Kunal. 2015. Fog Data: Enhancing Health-CPS: Healthcare Cyber-Physical System Assisted by Cloud and Big Data.
Telehealth Big Data Through Fog Computing. In In Proceedings of the ASE Big- IEEE Systems Journal 11, 1 (March 2017), 88–95.
Data & SocialInformatics 2015 (ASE BD&SI ’15). ACM, New York, NY, USA,
Article 14, 6 pages.
[6] F. Fernandez and G. C. Pallis. 2014. Opportunities and challenges of the Internet
of Things for healthcare: Systems engineering perspective. In In Proceedings of the

View publication stats