Virtual Sensors in Remote Healthcare Delivery: Some Case Studies

Nandini Mukherjee

, Suman Sankar Bhunia

and Sunanda Bose

Department of Computer Science and Engineering, Jadavpur University, Kolkata, India

School of Mobile Computing and Communication, Jadavpur University, Kolkata, India

Keywords:

E-Health, Sensor-cloud, Virtual Sensors, Remote Health.

Abstract:

Delivery of healthcare services to the people living in remote places is a challenging task. A remote healthcare

framework has been proposed in our earlier work based on sensor-cloud technologies. This paper explains

the scenarios for deployment of such a framework in a remote healthcare delivery system. In our sensor-

coud environment we propose to create virtual sensors and offer them on-demand to the health-care service

providers for use in their services. In this paper, we discuss the purpose of using virtual sensors in healthcare

domain and demonstrate how additional responsibilities that cannot be handled by physical sensor devices,

can be delegated to virtual sensors in order to improve the efﬁciency of the system. Results of preliminary

deployment of virtual sensors in two scenarios are discussed within limited scope and their advantages and

related issues are put forward for future implementation of the sensor-cloud infrastructure.

1 INTRODUCTION

Providing basic healthcare services in rural areas in

developing countries is a challenge. Primarily, this is

because doctors are not available in the same propor-

tion in the rural areas as they are available in urban

areas. Some studies in India indicate that there are

about four times as many trained doctors per ten thou-

sand population in urban areas as compared to the ru-

ral areas. This includes doctors in the public sector (in

primary healthcare centres set up by Government) as

well as in the private hospitals and nursing homes and

privately practicing doctors. Clinical testing facilities

are also unavailable in these areas.

Although, the situation can only be reversed

through government initiatives and some changes in

the societal structure, we propose to increase the

reachability of rural people to healthcare services with

the use of modern information and communication

technologies. In particular, our solution is based on

integration of cloud computing and sensor technolo-

gies and deployment of a sensor-cloud environment.

However, while building the sensor-cloud environ-

ment, we propose to create virtual sensors and offer

them on-demand to the health-care service providers

for use in their services. This paper presents few sce-

narios where instead of directly collecting data from

physical sensors and transmitting those to the cloud

environment, use of virtual sensors will be effective

and in some cases a necessary requirement. Such

scenarios for using virtual sensors in healthcare do-

main are discussed in Section 2. A scheme for re-

mote delivery of healthcare services on top of sensor-

cloud environment is discussed in Section 3. We also

demonstrate the use of some virtual sensors in the

kiosk-based system and discuss their advantages and

related issues. Section 4 discusses use of virtual sen-

sors in the proposed kiosk-based scheme and ﬁnally

Section 5 concludes.

2 DEVELOPING A

SENSOR-CLOUD

ENVIRONMENT

Cloud computing brings a change in the users’ vision

towards the computing world. Restrictions put by the

limited resources in a dedicated server are overcome

when computing and storage resources, software and

information are provided over a network as an utility

with an analogy with electricity or water supply. Ac-

cording to NIST (Mell and Grance, 2011), cloud com-

puting enables “ubiquitous, convenient, on-demand

network access to a shared pool of conﬁgurable com-

puting resources (e.g., networks, servers, storage, ap-

plications, and services) that can be rapidly provi-

sioned and released with minimal management effort

or service provider interaction”.

On the other hand, advancements in sensing tech-

484

Mukherjee, N., Bhunia, S. and Bose, S.

Virtual Sensors in Remote Healthcare Delivery: Some Case Studies.

DOI: 10.5220/0005823204840489

In Proceedings of the 9th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2016) - Volume 5: HEALTHINF, pages 484-489

ISBN: 978-989-758-170-0

Figure 1: Remote Healthcare: Use cases.

nologies, and possibility of connecting numerous spa-

tially distributed sensing devices wirelessly in a multi-

hop network create enormous scope for implement-

ing various applications which were never envisaged

earlier. However, instead of using sensor nodes only

for data transmission to cloud, we propose to create

a layer of abstraction of sensor nodes through virtu-

alization. Thus, with virtualization, it is possible to

delegate additional reponsisibilities to the virtualized

sensing devices and offer access to these devices on

demand basis and in shared and pervasive manner.

Virtualized sensors can also be reserved for a time pe-

riod. In other words, virtual sensors enable us to offer

SENsing-As-A-Service with extra capabilities in com-

parison with what can be offered by physical devices

and also providing ways to overcome the limitations

of physical sensing devices. A sensor-cloud frame-

work with sensor virtualization and the issues related

to the development of the framework have been dis-

cussed in (Mukherjee et al., 2014). In this section our

endeavor is to present some rationale for use of virtual

sensors in healthcare domain.

Virtual Sensors: A virtualization layer of the

sensor nodes can be created for several purposes.

As mentioned earlier, the tiny sensing devices have

limited capabilities in terms of computing resources,

storage and energy. Thus, creating virtual abstractions

of physical sensors in cloud environment and thereby

enabling them to use additional resources, it is pos-

sible to enhance the capabilities of these sensing de-

vices. Furthermore, in case of healthcare applications,

we envisage other requirements which can be fulﬁlled

by virtualizing sensor resources. A virtual sensor ac-

cumulates the data from one or more physical sensor

nodes and analyzes and processes the data to make

a decision. Based on the complexity of the decision

making algorithm and whether it runs at the infras-

tructure level (IAAS) or at the platform level (PAAS),

different names are given to the respective virtual sen-

sors. In our earlier work (Bose et al., 2015), we have

already deﬁned eight types of virtual sensors. In this

paper, we describe applications of some of the virtual

sensors in healthcare domain.

1. Number of sensing units are deployed on a pa-

tient’s body to sense various clinical parameters,

such as blood pressure, body temperature and

oxygen level (or oxygen saturation) in the blood.

All these parameters are related to one single pa-

tient and an application should be able to access

these parameters as a single data unit. A single

virtual sensor is created to accumulate the sensed

data from different sensing units and represent

them as a single data unit. Such a virtual sensor is

named as an accumulator.

2. A virtual sensor may sense the blood pressure of

a patient and when the sensed data goes beyond

a threshold value, the virtual sensor generates an

alert. Such a virtual sensor gets data from a single

sensor node and can be implemented at the sensor

network level. This virtual sensor is named as a

qualiﬁer.

3. A context qualiﬁer virtual sensor is similar to

a qualiﬁer virtual sensor, but instead of consid-

ering data from a single sensor, it obtains data

from multiple sensors and applies a decision al-

gorithm. Such a virtual sensor may be imple-

mented at body sensor network level. An exam-

ple of a context qualiﬁer virtual sensor is as fol-

lows: “When body temperature sensor reads high

value and skin conductance sensor reads normal

(patient is not sweating), the values indicate that

the patient has fever”. A range of values is deﬁned

for describing the terms like high, normal and low.

Virtual Sensors in Remote Healthcare Delivery: Some Case Studies

485

Figure 2: Kiosk-based remote healthcare delivery.

We can also use fuzzy rules in such cases (Bhunia

et al., 2014).

4. A virtual sensor may be implemented with a pre-

dictor algorithm to predict possible sensed val-

ues based on time series data analysis on pre-

viously sensed data when actual physical sensor

goes down. For example, a heart rate monitoring

sensor may be attached with the patient’s body for

continuous monitoring and the patient may be on

move (traveling in a train). The monitoring data

can be regularly uploaded to the cloud environ-

ment and used by the care giver for taking care of

any unusual situation. Due to connectivity prob-

lem or any other reason data may not be avail-

able at every instant. The predictor algorithm is

implemented at the platform level (PaaS) and can

be used to predict the intermediate values. The

predicted values can later be compared / corrected

with the actual values which are locally stored and

uploaded later to the cloud environment. We name

such a sensor a predictor.

5. A virtual sensor may be equipped with some

contextual computation abilities that analyzes the

sensed trafﬁc from a set of sensors. The com-

putational procedure transforms sensed data to a

more understandable information. For example,

an image analysis algorithm may be executed on

the data obtained from a group of physical sen-

sors collecting images from a patient’s body. The

virtual sensor is then represented in an integrated

form which contains the data obtained from the

set of physical sensors, the algorithm and neces-

sary computation and storage resources. We name

this virtual sensor a compute virtual sensor. This

sensor is also implemented at the platform level

(PAAS).

In addition to the above sensors, some other virtual

sensors with various capabilities have been described

in (Bose et al., 2015). However, these virtual sensors

are more applicable in case of applications other than

healthcare applications (such as environment moni-

toring).

It is clear from the above description that when

a user application requests for a virtual machine in

our sensor cloud environment, it not only requests

for usual virtualized resources like CPU, memory and

storage, it can also request for virtual sensors, as nec-

essary on the basis of the patients’ requirements. APIs

are provided for requesting virtual sensors and they

are provided on demand basis.

3 A SCHEME FOR REMOTE

HEALTHCARE DELIVERY

The sensor-cloud environment described above is be-

ing implemented for a remote healthcare delivery sys-

tem. Three use cases are considered for implementa-

tion of such a system (Figure 1). These use cases are

described below:

• Kiosk-based Healthcare Delivery: In rural ar-

eas, where doctors are not available, a health kiosk

can be set up. The kiosk should be equipped

with e-health sensor kit and operated by a team

of health assistants. The health assistants must be

trained to use health sensors, gather data from pa-

tients’ bodies using the sensors and transmit data

to the cloud through an e-health application. The

e-health application will run on a virtual machine

with capabilities as requested by the application

itself including necessary virtual sensors. Doc-

tors may be located in urban areas. Our sensor

cloud-based application allows the doctors to vi-

sualize the patients’ data remotely. When a pa-

HEALTHINF 2016 - 9th International Conference on Health Informatics

486

tient comes to a kiosk, the health assistants col-

lect data, upload data to the cloud. Doctors, after

remote analysis of data, make diagnosis, suggest

medication and further investigation. All investi-

gations and tests may not be possible in the health

kiosks (such as X-ray) and in such cases the pa-

tients are advised to visit external laboratories (al-

though such situation arises only in a few number

of cases). A patient visits the kiosk more than

once until the treatment related to a complaint

is over or the patient is referred to a secondary

healthcare centre. The entire scheme is described

in Figure 2.

In the above scenario, virtual sensors can be de-

ployed in cloud environment and can be used to

represent patients’ clinical data in an integrated

form to the doctors.

• Continuous Monitoring of Patients on Move:

A single sensor (such as heartbeat monitor) or a

body sensor network can be put on a patient’s

body. The patient will carry out normal activities.

The patient can be in motion as and when nec-

essary. Therefore, it is required to have seamless

connectivity when the patient moves through dif-

ferent networks, including cellular network, Wiﬁ

etc. While transmitting data from various sensors

through a common channel, there may be interfer-

ences. Handling such interferences is another is-

sue. When the sensor data exceeds some threshold

values, it may also be required to generate alert

messages. Virtual sensors may be created to deal

with the above issues.

• Ofﬂine Data Collection: Health assistants may

move from dood-to-door and collect health-

related data from several households. At certain

instances, these data may be uploaded to cloud en-

vironment for analysis. The data, as well as anal-

ysis results may be used by caregivers (doctors

or health assistants), healthcare units (hospitals),

or by the government organisations (health de-

partments). While the caregivers and the health-

care units require the data for monitoring pur-

poses, health departments may use the analysis re-

sults for controlling any epidemic or introducing

new schemes for reducing the disease burden in

the state. Enhanced capabilities of virtual sensors

may run algortihms for data analysis and present

data in different formats to the users based on their

requirements.

Patients’ data contains ﬁve parts: (i) demographic

data, (ii) past medical history of the patient, (iii) the

complaint of the patient, (iv) data collected by virtual

sensors deployed at the kiosk or on the patient’s body

and (v) data collected by tests and investigations. The

last part, that is the data collected by tests and inves-

tigations are ﬁlled up on the basis of doctors’ advice

and are possibly ﬁlled up during the subsequent visits

of the patient. Data are stored using sensorML for-

mat. A description of the data model is given in (Sen

and Mukherjee, 2014).

4 USE OF VIRTUAL SENSORS IN

REMOTE HEALTHCARE

DELIVERY

In this section, we discuss the use of virtual sensors in

our remote healthcare delivery application Healthsys.

In this paper, we will consider two particular scenar-

ios of deploying virtual sensors.

Case 1: Various physical sensors are put on the

patient’s body and a virtual sensor is deployed on

a gateway which is a laptop or an android-based

smart phone placed in the kiosk. The virtual

sensor accepts data from all the physical sensors,

interleaves them and forwards the combined data

for the use by an application or a service.

Beneﬁt: Parameters like electrocardiogram

(e.g. ECG) and other graph based parameters

require higher sampling rate, but have small data

size per sample. But parameters involving images

and audio (e.g. respiratory sounds) require low

sampling rate, but large data size per sample.

The required sampling rate must be maintained

for each sensor. Also, the data is sent through a

common channel. Therefore, a data interleaving

technique needs to be incorporated so that each

sensor gets a share of the available bandwidth

enabling each of them to maintain sampling rate,

data length and delay. If interleaving cannot be

done suitably (a sensor data may need to be split

between the samples of other sensors), then there

may be overlapping between the sensor data and

interpretation of the data may be deformed as

shown in Figure 3(a).

Implementation: A virtual sensor is de-

ployed at the infrastructure level that accepts

data from different physical sensors, interleaves

the data in such a way that none of the samples

from any physical sensor is missed and data are

delivered in time.

For example, ECG data requires high sample rate,

but data size is low. On the other hand, body tem-

perature data requires low sample rate and the data

Virtual Sensors in Remote Healthcare Delivery: Some Case Studies

487

(a) (b)

Figure 3: (a) ECG without using interleaving with other sensed data, (b) ECG with interleaving with other sensed data.

Figure 4: Framework for Fuzzy-based Data Collection.

size is large. Therefore, multiple ECG samples

can be interleaved with a sample of body temper-

ature data and can be transmitted through a com-

mon channel.

An example of the interleaving technique in a

speciﬁc scenario has been discussed in detail

in (Dhar et al., 2014) and some experimental

results are given. Figure 3 compares two ECG

results without and with the use of interleaving

techniques.

Case 2: A virtual sensor accepts data from one or

more physical sensors and applies fuzzy rules to

generate an alert or to activate another physical

sensor. Figure 4 shows a framework for fuzzy

assisted data collection in healthcare domain.

Example: An example of a fuzzy rule can

be as follows: “When body temperature is low,

and skin conductance sensor reads high (patient

is sweating), it is indicated that the patient is

in shock. Shock may arise from fear or heart

problems or psychological issues. As one of

the symptoms of heart attack is severe shock

of the patient, our system activates the heart

rate monitor when the above two conditions are

true.” (Bhunia et al., 2014).

Beneﬁt: This virtual sensor avoids unneces-

sary data collection from multiple physical

Figure 5: Reduced energy consumption in fuzzy-based

method.

sensors and saves energy consumption at the

physical level. An implementation of the fuzzy

rule-based system has been discussed in (Bhunia

et al., 2014). Figure 5 describes the reduction in

energy consumption with this implementation.

5 RELATED WORK

Delivering healthcare services remotely is being in

focus of the medical practitioners, as well as scien-

tists. Advantages of offering medical diagnosis and

monitoring services based on mobile health systems

and the challenges have been discussed in (Moham-

madzadeh and Safdari, 2014). In (Henderson et al.,

2014), it has been argued that nurse practitioners can

use technology through a telehealth service that can

improve quality and overcome geographic barriers to

health care access. They have also shown that tech-

nology can improve patients access to health care in

a cost-effective manner. Few critical questions have

been raised in (Puddu et al., 2014) in relation to re-

mote healthcare systems and the pertinent technolo-

gies implied in transferring surveillance and care to

the patients. The reviewed some use case scenarios

and attempted to integrate different points of views

of physicians, patients, academicians, health service

HEALTHINF 2016 - 9th International Conference on Health Informatics

488

organizations, industries, and the end users.

In spite of the above research works and studies

on remote healthcare delivery, application of sensor-

cloud and sensor virtualization have not received

much focus in healthcare domain. Sensor virtual-

ization for underwater event detection has been dis-

cussed in (Wang et al., 2014) in which the base station

collects measurements from multiple sensor nodes,

and makes a decision based on the sensors reports.

enables the collection of data streams from multiple

heterogeneous geographically dispersed data sources,

as well as their semantic uniﬁcation and streaming

with a cloud infrastructure. It has been proposed

in (Petrolo et al., 2014) to enable collection of data

streams from multiple heterogeneous geographically

dispersed data sources and their semantic uniﬁcation

and streaming with a cloud infrastructure for a smart

city solution.

Our research studies the purposes of using virtual

sensors for healthcare services and focuses on intro-

duction of a layer of abstraction to implement virtual

sensors. Use of virtual sensors have been demon-

strated in two scenarios. Currently, we are focusing

on other types of virtual sensors and their implemen-

tation.

6 CONCLUSION

This paper focuses on remote healthcare delivery on

top of a sensor-cloud framework. The paper particu-

larly discusses virtualization of sensors and their ap-

plications in healthcare domain. A remote primary

healthcare delivery application has been dicussed and

its implementation using virtual sensors has been con-

ceptualized. We are currently developing an architec-

ture for implementation of virtual sensors and APIs

for their uses in healthcare services.

REFERENCES

Bhunia, S. S., Dhar, S. K., and Mukherjee, N. (2014).

ihealth: A fuzzy approach for provisioning intelli-

gent health-care system in smart city. In Wireless and

Mobile Computing, Networking and Communications

(WiMob), 2014 IEEE 10th International Conference

on, pages 187–193. IEEE.

Bose, S., Gupta, A., Adhikary, S., and Mukherjee, N.

(2015). Towards a sensor-cloud infrastructure with

sensor virtualization. In Proceedings of the Second

Workshop on Mobile Sensing, Computing and Com-

munication, pages 25–30. ACM.

Dhar, S. K., Bhunia, S. S., and Mukherjee, N. (2014). In-

terference aware scheduling of sensors in iot enabled

health-care monitoring system. In Emerging Applica-

tions of Information Technology (EAIT), 2014 Fourth

International Conference of, pages 152–157. IEEE.

Henderson, K., Davis, T. C., Smith, M., and King, M.

(2014). Nurse practitioners in telehealth: bridging the

gaps in healthcare delivery. The Journal for Nurse

Practitioners, 10(10):845–850.

Mell, P. and Grance, T. (2011). The nist deﬁnition of cloud

computing.

Mohammadzadeh, N. and Safdari, R. (2014). Patient moni-

toring in mobile health: opportunities and challenges.

Medical Archives, 68(1):57.

Mukherjee, N., Bhunia, S. S., and Sen, P. S. (2014).

A sensor-cloud framework for provisioning remote

health-care services. In Computing & Networking

for Internet of Things (ComNet-IoT) workshop co-

located with 15th International Conference on Dis-

tributed Computing and Networking.

Petrolo, R., Mitton, N., Soldatos, J., Hauswirth, M., and

Schiele, G. (2014). Integrating wireless sensor net-

works within a city cloud. In Sensing, Communi-

cation, and Networking Workshops (SECON Work-

shops), 2014 Eleventh Annual IEEE International

Conference on, pages 24–27. IEEE.

Puddu, P. E., DAmbrosi, A., Scarparo, P., Centaro, E., Tor-

romeo, C., Schiariti, M., Fedele, F., and Gensini, G. F.

(2014). A clinicians view of next-generation remote

healthcare system. In Systems Design for Remote

Healthcare, pages 1–30. Springer.

Sen, P. S. and Mukherjee, N. (2014). Standards of ehr and

their scope of implementation in a sensor-cloud envi-

ronment: In indian context. In Medical Imaging, m-

Health and Emerging Communication Systems (Med-

Com), 2014 International Conference on, pages 241–

246. IEEE.

Wang, Z., Liu, M., Zhang, S., and Qiu, M. (2014). Sensor

virtualization for underwater event detection. Journal

of Systems Architecture, 60(8):619–629.

Virtual Sensors in Remote Healthcare Delivery: Some Case Studies

489