Int. j. inf. tecnol.

https://doi.org/10.1007/s41870-018-0116-1

ORIGINAL RESEARCH

A context-aware data fusion approach for health-IoT

Zartasha Baloch1 • Faisal Karim Shaikh1 • Mukhtiar Ali Unar1

Received: 11 December 2017 / Accepted: 16 February 2018

 Bharati Vidyapeeth’s Institute of Computer Applications and Management 2018

Abstract The technological advances in low-cost sensor 1 Introduction

devices and communication technologies bring rapid
increase in development of smart homes and smart envi- Internet of Things (IoT) is trending technology that attracts
ronments. The developments in wireless sensor networks many researchers around the world. It helps to make our
(WSN), body area networks (BAN), cloud computing and daily life more intelligent and smart by interconnecting
big data technologies trigger the use of Internet of Things physical objects. The main concept of IoT is the pervasive
(IoT) in healthcare industry. This poses many challenges presence of a variety of things (objects) around us, which
such as heterogeneous data fusion, context-awareness, can interact with each other and collaborate with their
complex query processing, reliability and accuracy etc. neighboring objects to achieve common target [1].
Data fusion techniques are used to extract meaningful IoT is a multidisciplinary field, which may include but
information from heterogeneous IoT data. It combines not limited to embedded systems, electronics, wireless
individual data from sensor sources to collectively obtain a sensor networks, communication networks, and computing
result, which is more reliable, accurate and complete. Apart paradigms. IoT serves as an umbrella above these tech-
from wearable sensors, additional context sensors need to nologies, which diversifies its applications area. Its appli-
be added to build a context. Health IoT applications has cations include smart cities, smart homes/smart buildings,
potential benefits of using context-aware data fusion. By environment monitoring, smart business and product
using context information, the behavior of the application management, emergency response systems, intelligent
can be customized according to the specific situation. This transportation, security and surveillance, energy and
paper provides a brief concept of context-aware data fusion industrial automation, and healthcare etc. [2].
and includes data management approach for context-aware The continuous growth of IoT projects towards its
systems for healthcare applications. Finally, a context- adaptation by diverse industries, even low cost organiza-
aware data fusion approach for health IoT is proposed. It tions are also feeling the effects. For the reason, IoT needs
includes context acquisition, situation building and rea- its own dedicated wide area network. Typically, wide area
soning and inference. networks require larger batteries or continuous power
supply, which is not feasible in all the situations. The
Keywords Data fusion
Context-awareness
Healthcare
development of low power wide area (LPWA) networks is
Internet of Things solution to this challenge, which provides benefits includ-
ing reduced total cost of ownership, increased battery life,
and long coverage range. Keith et al. [3] discussed basics
of low power wide area (LPWA) networks for IoT and
provides a survey of two LPWA technologies, namely
& Zartasha Baloch LoRa technology by Semtech and ultra-narrow band
[email protected]
solutions by SigFox, in terms of physical layer and medium
1
Mehran University of Engineering and Technology, access control layer. Konstantin et al. [4] analyzed the
Jamshoro, Pakistan capacity and scalability of LoRa wide area networks. On

123
 Int. j. inf. tecnol.

the basis of their analysis, they supposed that LoRa can be altogether to be aware of some threat. Likewise, wireless
efficiently utilized in moderately dense network of devices sensor nodes are deployed to observe an environment,
with low traffic, and low latency and reliability require- combining these sensor readings is known as sensor data
ments such as environment monitoring and non-critical fusion.
infrastructure applications. It is a technique to integrate data from heterogeneous
As in IoT, every ‘thing’ is connected over internet and sources. It combines the common characteristics of datasets
information is shared among them, this imposes severe from multiple sources to produce better results. Sometimes
security challenges. Mostly, IoT is applied to critical data from one source may not be sufficient for taking
applications so their security requirements are quite high. decision in real-time environment. Data fusion supports
Hui et al. [5] reviewed security challenges in IoT and representation of data in a way that helps decision makers
discussed its security characteristics and requirements. in decision process [10]. It is a process that transforms data
They analyzed four layered security architecture for IoT to help taking correct decision on time. There are several
which includes perceptual layer, network layer, support definitions of data fusion in literature but most widely used
layer, and application layer. Quandeng et al. [6] also definition is from JDL. According to Joint Directors of
highlights security issues of IoT in perception, network and Laboratories (JDL) workshop [11] Data fusion is defined as
application layers. They also proposed secure construction
‘‘A multi-level process dealing with the association,
of IoT by describing security policies in each layer.
correlation, combination of data and information
Research in IoT security is still in its initial phase and more
from single and multiple sources to achieve refined
research is needed in this field.
position, identify estimates and complete and timely
With the advent of IoT in healthcare, many low-cost
assessments of situations, threats and their
devices are used to monitor patient’s health status remo-
significance.’’
tely. IoT has been widely adopted for both in-home and in-
hospital care. In medical systems, the use of wearable The aim of multi-sensor data fusion is to understand the
sensors and wireless sensor networks is increasing rapidly. environment and act accordingly. When more than one
Body sensor networks (BSN) or wireless body area net- sensor is used to gather more accurate and additional data,
works (WBAN) is a type of wireless sensor network that sensor fusion is used.
deploys to the patient’s body to collect physiological The use of a single sensor or multiple individually
measurements [7]. It facilitates developments in pervasive considered sensors can cause many limitations to any
monitoring systems by providing an integrated hardware physical measurement system. Likewise, a Health IoT
and software platform [8]. application may also suffer from several problems, such as
A generic BSN consists of a variety of sensor nodes [12]
which records patient’s physiological data. The different
• Sensor deprivation: in case of a node failure, that
wearable body sensors capture the patient’s data such as
specific observation will be lost.
blood pressure, electrocardiogram (ECG), blood sugar
• Restricted coverage: normally a single sensor covers a
level, galvanic skin response, and electromyography etc.
limited area or a restricted body part.
[9]. In case if any of these values crosses the threshold
• Imprecision: accuracy of individual sensors is limited
value, the system alerts the medical emergency response
to only those specific elements.
center. Such systems are helpful when a patient is not in a
• Uncertainty: an individual sensor may not determine all
condition to visit hospital for routine checkups or the
aspects of an incident, hence causes uncertainty.
patient needs in-home care e.g. elderly people. The sensed
data is then stored to the system for future processing. These challenges urge the need for context-aware data
As these devices consistently generate data, thus there fusion. In this paper, a novel approach has been proposed
are many issues and challenges to handle such an explosive to solve these issues. The rest of the paper is organized as
data. The data becomes more complex as BSN consists of follows. Section 2 discusses related work. Section 3 pre-
multiple heterogeneous sensor sources, so for the efficiency sents context-aware data fusion in healthcare and describes
of BSN applications the use of data fusion is non-trivial. data management steps for context-aware healthcare sys-
The concept of multi-sensor data fusion is not new. It can tems. In Sect. 4, a context-aware data fusion approach is
be compared with the use of senses of human body. In daily proposed. Finally, Sect. 5 concludes the paper.
routine, we use multiple senses together to understand the
environment more accurately. Sometimes a single sense
doesn’t provide enough information, e.g. we use smell,
sight and taste sense to assess food quality; even sometimes
we need to touch it. We use sight and hearing sense

123
Int. j. inf. tecnol.

2 Related work systems. Jain [18] presented implementation for telecom-

munication fraud detection using data stream analytics and
Sensor data fusion is well-known research field and there is neural network classification based data mining and com-
variety of techniques available in research literature but pared the accuracy of both techniques. The results of this
still there are many challenges that need to be addressed. research can provide a basis for big data analytics and
Two main approaches are possible for the process of data mining.
fusion, namely centralized and distributed/decentralized Based on above considerations, we propose a context-
[9, 13]. In the centralized data fusion approach, the fusion aware data fusion model for health-IoT. The novel concept
process is performed at a central node called fusion center. in this model is focus on dynamic management of context
All the sensor nodes pass their data to the fusion center and vital sign sensor data and it also includes user refine-
which is responsible for overall fusion process. The benefit ment and preferences.
of this approach is that it provides global vision of envi-
ronment; while the major drawbacks are; first the overall
fusion process depends on a single node, causing network 3 Context-aware data fusion in healthcare
congestion on that node and may be sometimes more
information is received than its capability. Second, in case For any health system, the accuracy and reliability of data
of fusion node failure it affects the overall system. In the is most critical issue. Sometimes wearable sensors do not
distributed or decentralized approach, the fusion process is provide sufficient information to build the situation. For the
distributed over some dedicated nodes. Those nodes reason, additional sensors are needed for context infor-
receives data from surrounding nodes and sends locally mation. By combining data from multiple heterogeneous
fused information to next module for decision level fusion sensors, sensor fusion creates complete picture of the sit-
and global analysis. In this case, the node failure doesn’t uation which enables context awareness. Health IoT
affect the overall system, as other fusion nodes can con- applications has huge potential benefits using context-
tinue their work. So, the fusion process can be performed aware sensor fusion, e.g. a pulse sensor measures an
on raw data (centralized approach) or on pre-processed increase in the pulse rate of a person; there may be some
data (distributed approach). situation where it is normal like if the person is running or
The use of context in sensor fusion is becoming essential doing some physical workout at gym, otherwise this may
part of the real-time interactive systems. Gravina et al. [9] be an alarming situation. By combining the location and
provides a comprehensive survey on multi-sensor fusion in motion (for activity) sensors, a context can be built to
body sensor networks and identifies the future research protect false alarms.
issues as autonomic, context-aware, collaborative, and Abowd et al. [14] defines context as it can be any sup-
cloud-assisted body area networks. Abowd et al. [14] cat- porting information to define the situation of objects that
egorized features of context-aware applications as (1) are related to the interaction between a user and an appli-
presentation of information and services to a user; (2) cation including them. So the context can be defined as
automatic execution of a service; and (3) tagging of context location, environment, time and identity of user. This
to information for later retrieval. identifies what is happening in the situation. Context plays
De Paola et al. [15] proposed and evaluated three-tier vital role in real-time systems. The sensor fusion method is
architecture of a context-aware self-optimizing adaptive supposed to deal with the changes in context, as they
system for multi-sensor fusion. They used Bayesian net- highly affect accuracy [9].
work for inferring the state of the world and using con- The healthcare systems generally consist of heteroge-
textual information to improve reasoning accuracy. Hong neous data sources that exhibit explosive data. Although,
et al. [16] proposed a framework for information man- due to massive data from sensors and other devices, data
agement in smart homes to support decision making pro- management is the main concern for many researchers; but
cess of activity recognition. They used equally weighted still there is no agreed upon definition [19]. The steps for
sum operator and Dempster–Shafer theory of evidence for data management highly depend on nature of data and the
uncertainty of sensor data. application used for.
An appropriate data fusion technique helps in building To manage heterogeneous IoT data a multilayered
intelligent systems. Chetty and Yamin [17] presented a approach is shown in Fig. 1, which includes data prepro-
smart fusion framework, based on hard and soft sensor cessing, context-aware data fusion, and data processing and
fusion, for combining heterogeneous, multimedia, multi- storage.
modal real-time big data streams to achieve actionable Data is collected at physical layer. Then, at data pre-
intelligence from the computer based decision support processing layer necessary cleaning and filtering is applied
to remove outliers (irregularities or unexpected values).

123
 Int. j. inf. tecnol.

Fig. 1 Data management for context aware healthcare systems

Sometimes collected data contain redundant data which The first step in context acquisition is data collection
must be compressed to reduce storage and communication from physical devices. Then, measurement preprocessing is
overhead. The next step is context-aware data fusion. This done by applying filtering and estimation techniques to
layer shows two types of data blocks, one is vital signals remove noise and other measurement outliers. After that,
such as pulse rate, blood pressure, electrocardiogram data from multiple sources are associated to extract context
(ECG), body temperature etc., and the other is context data. information. Finally, appropriate data fusion algorithm,
Context data contains additional information such as depending on the application, will combine data.
location of object, or environment temperature. The con- Situation building phase deals with the analysis of sit-
text source varies according to the situation. Intelligent uational and relational information. It also tracks historical
data fusion techniques combine heterogeneous data and data to validate current data and predicts future situation.
extracts context information which enables health appli- This level presents a virtual image of objects with their
cations to react accordingly. position and relationship.
Once fusion process is done and context is build, next Reasoning and inference completes overall target of the
step is to process that data. The data processing and storage system. It indicates how to interact with physical infras-
phase includes three sub-blocks, i.e. data analysis, query tructure and according to the requirement of applications; it
processing and storage. All of them are huge fields at their adjusts measuring parameters of sensors. It also includes
own, and are beyond scope of this paper. Healthcare user input to help increase system performance and deci-
applications can access and store that processed data. sion process; this is known as user refinement. Blasch and
Plano [20] revised JDL model by including level 5, i.e. user
refinement, which supports user interactions to improve
4 Proposed context-aware data fusion approach decision process.
for health IoT Different intelligent techniques may be used to boost the
decision process such as Dempster–Shafer theory, Baye-
This section discusses our proposed context-aware data sian network, artificial intelligence, and fuzzy logic etc.
fusion approach for healthcare applications. The approach
is based on JDL [11] data fusion process model. Our aim
is to design a model to fuse vital signal data with context 5 Conclusion
data to better understand situations. The model consists of
three phases, context acquisition and filtering, situation The advances in pervasive computing trigger the use of IoT
building, and reasoning and intelligent inference as shown in healthcare. Due to which developments in wearable
in Fig. 2. sensors and their use in health industry are also increasing

123
Int. j. inf. tecnol.

Fig. 2 A context-aware data

fusion approach for health IoT

very fast. Multi-sensor data fusion is emerging research 6. Gou Q, Yan L, Liu Y, Li Y (2013) Construction and strategies in
field with sound future scope to provide reliable data. This IoT security system. In: Green computing and communications
(GreenCom), 2013 IEEE and Internet of Things (iThings/
paper has aimed to provide basic concepts of IoT in CPSCom), IEEE international conference on and IEEE cyber,
healthcare and context-aware data fusion. Data manage- physical and social computing. IEEE, pp 1129–1132
ment steps for healthcare systems are also described, which 7. Amendola S, Lodato R, Manzari S, Occhiuzzi C, Marrocco G
includes preprocessing, context-aware data fusion and data (2014) RFID technology for IoT-based personal healthcare in
smart spaces. IEEE Internet Things J 1(2):144–152
processing and storage. A layered approach for context- 8. Lo B, Yang GZ (2005) Key technical challenges and current
aware data fusion for health IoT has been proposed, implementations of body sensor networks. In: Proceedings of the
including context acquisition and filtering, situation 2nd international workshop on body sensor networks (BSN 2005)
building, and reasoning and intelligent inference. This 9. Gravina R, Alinia P, Ghasemzadeh H, Fortino G (2017) Multi-
sensor fusion in body sensor networks: state-of-the-art and
explains how context-aware data fusion is performed by research challenges. Inf Fusion 35:68–80
following multiple subtasks. 10. El Faouzi NE, Klein LA (2016) Data fusion for ITS: techniques
As future consideration, we will work on specific tech- and research needs. Transp Res Proc 15:495–512
niques and algorithms for each phase. This framework will 11. White FE Jr (1987) Data fusion lexicon, joint directors of labo-
ratories, technical panel for C3, data fusion sub-panel. Naval
help us in creation and realization of intelligent algorithms. Ocean Systems Center, San Diego
12. Elmenreich W (2002) An introduction to sensor fusion. Vienna
Acknowledgements This work has been performed under IICT, University of Technology, Austria
Mehran University of Engineering and Technology Jamshoro funded 13. Veloso M, Bento C, Pereira FC (2009) Transportation systems:
by ICT Endowment Fund for sustainable development. multi-sensor data fusion on intelligent transport systems.
University of Coimbra, Coimbra
14. Abowd GD, Dey AK, Brown PJ, Davies N, Smith M, Steggles P
(1999) Towards a better understanding of context and context-
awareness. In: International symposium on handheld and ubiq-
References uitous computing. Springer, Berlin, Heidelberg, pp 304–307
15. De Paola A, Ferraro P, Gaglio S, Re GL, Das SK (2017) An
1. Giusto D, Iera A, Morabito G, Atzori L (eds) (2010) The Internet adaptive bayesian system for context-aware data fusion in smart
of Things: 20th Tyrrhenian workshop on digital communications. environments. IEEE Trans Mob Comput 16(6):1502–1515
Springer Science & Business Media 16. Hong X, Nugent C, Mulvenna M, McClean S, Scotney B, Devlin
2. Miorandi D, Sicari S, De Pellegrini F, Chlamtac I (2012) Internet S (2009) Evidential fusion of sensor data for activity recognition
of Things: vision, applications and research challenges. Ad Hoc in smart homes. Pervas Mobile Comput 5(3):236–252
Netw 10(7):1497–1516 17. Chetty G, Yamin M (2017) A distributed smart fusion framework
3. Nolan KE, Guibene W, Kelly MY (2016) An evaluation of low based on hard and soft sensors. Int J Inf Technol 9(1):19–31
power wide area network technologies for the Internet of Things. 18. Jain V (2017) Perspective analysis of telecommunication fraud
In: Wireless communications and mobile computing conference detection using data stream analytics and neural network classi-
(IWCMC), 2016 international. IEEE, pp 439–444 fication based data mining. Int J Inf Technol 9(3):303–310
4. Mikhaylov K, Petaejaejaervi J, Haenninen T (2016) Analysis of 19. Baloch Z, Shaikh FK, Unar MA (2016) Interfacing physical and
capacity and scalability of the LoRa low power wide area net- cyber worlds: a big data perspective. In: Mahmood ZH (ed) Data
work technology. In Proceedings of European wireless 2016; 22th science and big data computing. Springer, Switzerland, pp 117–138
European wireless conference. VDE, pp 1–6 20. Blasch EP, Plano S (2002) JDL level 5 fusion model: user
5. Suo H, Wan J, Zou C, Liu J (2012) Security in the Internet of refinement issues and applications in group tracking. In: Signal
Things: a review. In: 2012 international conference on computer processing, sensor fusion, and target recognition XI, vol 4729.
science and electronics engineering (ICCSEE), vol 3. IEEE, International Society for Optics and Photonics, pp 270–280
pp 648–651

123