EUPaaS

Elastic Ubiquitous Platform as a Service for Large-scale Ubiquitous Applications

Fei Li

, Schahram Dustdar

, Jakob Bardram

, Martin Serrano

, Manfred Hauswirth

Vasilios Andrikopoulos

and Frank Leymann

Distributed Systems Group, Vienna University of Technology, Vienna, Austria

Cetrea A/S, Aarhus, Denmark

Digital Enterprise Research Institute, National University of Ireland Galway, Galway, Ireland

Institute of Architecture of Application Systems, University of Stuttgart, Stuttgart, Germany

Keywords:

Platform as a Service, Cloud Computing, Ubiquitous Computing, Healthcare, Hospital.

Abstract:

Large-scale ubiquitous computing applications are rapidly emerging in the ﬁelds of pervasive healthcare, smart

cities and so on. They present unprecedented challenges to state-of-the-art ubiquitous systems in the respects

of accommodating ﬂuctuating user demands, handling volatile data quality and adaptation to complex system

and user contexts. Driven by a motivating scenario in future mega-hospital environment, we propose to exploit

the potential of cloud computing in supporting large-scale ubiquitous computing applications. This position

paper will present the novel concept of EUPaaS (Elastic Ubiquitous Platform as a Service), outline the key

research topics, and propose a cloud-based ubiquitous application platform.

1 INTRODUCTION

In recent years, new business and research op-

portunities have been increasingly emerging from

the need for large-scale ubiquitous computing sys-

tems (Weiser, 1991), including for example pervasive

healthcare, smart cities, and so on. The development

and provisioning of these systems face a range of sig-

niﬁcant challenges. From a system’s point of view,

they are characterized by employing large numbers of

mobile devices and specialized sensors connected in

a volatile network setup. They generate and process

massive amounts of real-time data while coping with

ﬂuctuation of user demands. From the user’s point

of view, these systems need to ensure responsiveness

for ad-hoc usages, support user mobility, adapt to user

context, and provide highly personalized usage expe-

rience. Both viewpoints need to be coherently sup-

ported in large-scale ubiquitous computing systems.

Future mega-hospital

is a typical application en-

vironment of such systems. The scale of a future

mega-hospital will be in the range of 10,000 clin-

icians; 100,000 hospitalizations and 900,000 out-

Plan of the New University Hospital at Aarhus: http://

www.dnu.rm.dk/english

patient treatments per year; covering 380,000 m

;

35,000 ambulances and cars arriving every day; and

having 1,300 beds. Although ubiquitous comput-

ing technologies have been successfully applied in

hospital environments (Bardram, 2009), providing

context-awareness to such amount of users, with

many of them in time- and life-critical procedures,

poses unprecedented challenges, including elasticity,

large-scale context-awareness, reliable data process-

ing, application adaptation and so on.

This paper proposes to address the challenges

in mega-hospital and enable large-scale ubiquitous

computing systems by consolidating the core con-

cepts of two computing paradigms—Cloud comput-

ing and Ubiquitous computing. Conceptually, ubiq-

uitous computing implies using massive numbers of

small computational entities to reach all aspects of

human life. Intuitively, the virtually inﬁnite comput-

ing power of the cloud can help resource-constrained

ubiquitous systems to reach large-scale. However, the

state-of-the-art ubiquitous computing systems have

mostly been applied on a small, personal scale and

focused on personal mobility and context-awareness.

The ongoing work on combining ubiquitous comput-

ing and cloud computing still focuses on leverag-

ing the computing and storage resources of the cloud

309

Li F., Dustdar S., Bardram J., Serrano M., Hauswirth M., Andrikopoulos V. and Leymann F..

EUPaaS - Elastic Ubiquitous Platform as a Service for Large-scale Ubiquitous Applications.

DOI: 10.5220/0004427803090314

In Proceedings of the 3rd International Conference on Cloud Computing and Services Science (CLOSER-2013), pages 309-314

ISBN: 978-989-8565-52-5

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

(Satyanarayanan, 2010; Chun et al., 2011). To this

end, we present Elastic Ubiquitous Platform as a Ser-

vice (EUPaaS), which is focused on exploiting the po-

tential of clouds in supporting large-scale ubiquitous

systems. Speciﬁcally, this position paper will outline

a set of novel concepts in our ongoing research on

the convergence of the two complementary comput-

ing paradigms, and propose a cloud-based ubiquitous

application platform. Overall the research presented

in this position paper will achieve the following ob-

jectives:

• To genuinely transfer the key characteristics of

cloud computing, especially elasticity and service

orientation, into large-scale ubiquitous systems;

• To support the development and execution of

mobile, ubiquitous, context-aware applications in

PaaS paradigm;

• To realize cloud-native ubiquitous applications

that posses the characters of both cloud and ubiq-

uitous applications;

• To provide real-time data services to large

amounts of ubiquitous applications by quality-

aware, real-time data processing on clouds;

• To compose ubiquitous device services and real-

time data into adaptive applications through

service-oriented computing.

The paper is organized as follows: Section 2

presents a use case and corresponding challenges in

mega hospital environments; In Section 3 we present

the novel concepts of the work; then Section 4 pro-

poses the EUPaaS architecture under development;

the paper is concluded in Section 5.

2 UBIQUITOUS COMPUTING IN

FUTURE MEGA-HOSPITAL

Ubiquitous computing technologies have proven par-

ticularly applicable in a hospital setup (Bardram,

2009). Sensors are deployed hospital-wide for loca-

tion tracking of staff, patients, and equipment; video

surveillance; bed monitoring; and biomedical sens-

ing. All of these sensor inputs help build a real-time

overview of the clinical logistics at the hospital and

are used for context-aware information visualization

on large displays and on mobile devices. Moreover,

context-aware notiﬁcations are forwarded to relevant

clinical personnel on a timely fashion using the most

appropriate devices. However, as the size of a hospital

is increasing, the scalability of these types of ubiqui-

tous computing systems becomes signiﬁcantly chal-

lenging. This is especially true in the case of future

mega-hospitals—even the most simple context-aware

applications based on a location tracking infrastruc-

ture will need to handle location updates from 10,000-

20,000 entities in a large area with a sampling rate

of e.g. 1 minute. More importantly, a large amount

of time- and life-critical hospital procedures need to

be supported at the same time in the future mega-

hospital. The following use case illustrates one of

these procedures.

2.1 Clinical Logistics for an Emergency

Case

Monday morning, Mr. Hansen is not feeling well

and his left arm is hurting. His wife takes him to

the Emergency and Accident Department (E&AD) at

the local hospital for examination. At the E&AD, the

Emergency Clinical Logistic System (ECLS) helps

the head nurse to quickly assign a doctor, a nurse,

and an examination room. In the examination room,

his medical information and real-time data on ECG,

pulse, and O2 is displayed on the interactive ECLS

display. When examining Mr. Hansen, the doctor re-

alizes that this is a critical and acute case that needs

immediate transfer to the University Hospital (DNU)

for acute specialized surgery coronary angioplasty.

Together with the head nurse, the doctor uses ECLS to

initiate a transfer request from this local E&AD to the

Department of Cardiac and Thorax Surgery (Dept. T)

at DNU. The head nurse at Dept. T is notiﬁed about

the incoming patient via the ECLS system, and ac-

cepts the transfer.

Upon acceptance, the medical data of Mr. Hansen

is relayed through the regional healthcare backbone

infrastructure. At Dept. T, the ECLS interactive dis-

plays at the nursing station start to show real-time

information about Mr. Hansen. Porters and an am-

bulance are automatically scheduled and are notiﬁed

to pick up the patient in the local hospital and drive

him to DNU. The ECLS displays show the location

of the ambulance as it is approaching. Once in the

ambulance, medical information on Mr. Hansen and

his triage is entered in the ambulance medical record,

and real-time data on ECG, pulse, and O2 is obtained

through sensors in the ambulance. All of this data is

continuously displayed at the nursing station of DNU.

In the meanwhile, based on real-time knowledge

on the location, workload, medical proﬁle, and work

schedule of the doctors and nurses in Dept. T, the

ECLS system starts suggesting a surgeon and a team

of operating nurses to the incoming patient case. The

ﬁnal assignment is approved and done by the nurs-

ing station, and the involved staff is notiﬁed through

their ECLS mobile phones. Upon conﬁrmation of

CLOSER2013-3rdInternationalConferenceonCloudComputingandServicesScience

310

availability, the real-time information about the pa-

tient is available from the mobile phones of the as-

signed medical stuff. At the same time, an available

operation room and all the things needed for a surgery

are booked by the ECLS.

Mr. Hansen arrives and the porters can use their

mobile ECLS system to see where to take him. In the

operating room everybody is ready for surgery. The

surgery is successful and Mr. Hansen is moved to

the intensive care unit (ICU) for recovery. Once re-

covered, he is moved to the patient ward where his

wife is waiting for him. She has been able to follow

his treatment along the way using the patient/relative

ECLS mobile phone app.

2.2 Challenges

The use case alone can be realized with the technolo-

gies developed in the past research on context-aware

systems. However, supporting such procedure at a

very large scale presents unprecedented challenges to

both hospital IT platform and ubiquitous applications.

• Elasticity. The work load of a hospital is very

uneven across each day and week. Relatively

few activities are taking place on a Sunday af-

ternoon and night, mostly emergencies, whereas

there is a steep increase in activity (and hence in-

frastructure demand) starting Monday morning at

7 o’clock. Moreover, since a hospital to a large

degree is designed to handle critical situations

like emergencies, acute patients being transferred

from other smaller hospitals, and epidemic situ-

ations, its peak load cannot be predicted. There

is therefore a fundamental need for elasticity in a

hospital’s IT systems.

• Context-awareness in Large-scale. For a hospi-

tal at the scale of the DNU, a large number of

emergency processes, like in the use case, are per-

formed in parallel day and night. Besides emer-

gencies, many more routine and less-critical pro-

cesses of the hospital are also being carried out

all the time. Context-awareness is needed preva-

lently in these processes. There have been many

solutions to provide contextual information to ap-

plications through dedicated sensors. However,

it requires new approaches to acquiring and shar-

ing context for large amounts of applications and

users on a large-scale, heterogeneous sensory in-

frastructure.

• Data Quality and Data Processing. In a life-

and time-critical environment like a hospital, data

quality has signiﬁcant impact on users. For ex-

ample, in the use case, if the location information

of a doctor is not updated in 3 minutes by RFID-

based tracking (this could happen frequently be-

cause of signal interference), the ECLS needs to

alternatively use WiFi location tracking based on

the doctor’s wireless connection on mobile phone.

Furthermore, the system needs to accommodate

differentiated data quality requirements for dif-

ferent applications. In the use case, timeliness

and accuracy of biological signs are essential for

the treatment of Mr. Hansen. In future mega-

hospitals, these kinds of data will need to be de-

livered from large amount of devices to many doc-

tors reliably, timely, and accurately even when the

hospital is at peak load, e.g. at 7 o’clock Monday

morning.

• Application Environment. In mega-hospitals like

DNU, many applications like ECLS and other

clinical systems are used by a large number of

users at the same time. For different users, an

application may employ different sensors, require

different data processing schemes, and provide in-

formation on different user interfaces. To ensure

the required service quality for a user, an appli-

cation needs to adapt to many factors, including

availability of devices, ﬂuctuation of data qual-

ity, changes of user context, etc. Furthermore, a

mega-hospital is not only a centralized healthcare

facility, but also a center of regional healthcare

ecosystem. Many healthcare services need to per-

form beyond the boundary of a single hospital and

collaborate with other facilities. As suggested by

the use case, the system should allow applications

to seamlessly interact with authorized outside en-

tities like ambulances, devices and local hospitals.

3 ELASTIC UBIQUITOUS

PLATFORM AS A SERVICE

To address the aforementioned challenges and enable

large-scale ubiquitous computing systems, this pa-

per proposes Elastic Ubiquitous Platform as a Service

(EUPaaS), which constitutes a set of novel concepts

to be investigated in the intersection of cloud comput-

ing and ubiquitous computing.

• Elastic Ubiquitous Computing

Elasticity is the infrastructural capability that

computing resources allocated to an application

can be scaled up and down according to the work-

load of the application while maintaining a certain

service level. Elasticity has not been considered

so far in ubiquitous applications because of their

limited scales and closed system environments.

EUPaaS-ElasticUbiquitousPlatformasaServiceforLarge-scaleUbiquitousApplications

311

However, with the emergence of large-scale ubiq-

uitous systems like future mega-hospital, it is im-

portant to recognize the need for elasticity in ubiq-

uitous computing environments and realize it on

the cloud. Elastic ubiquitous computing means

that computing resources allocated to an applica-

tion can be scaled up and down depending on the

number and types of devices the application em-

ploys, on the needs to acquire and process the data

provided by the devices, and on the quality level

needed by the application for a certain situation.

• Platform as a Service for Ubiquitous Applica-

tions

Platform as a Service (PaaS) (Liu et al., 2011)

is deﬁned as ”the capability provided to the con-

sumer is to deploy onto the cloud infrastructure

consumer-created or acquired applications cre-

ated using programming languages and tools sup-

ported by the provider. The consumer does not

manage or control the underlying cloud infras-

tructure including network, servers, operating sys-

tems, or storage, but has control over the deployed

applications and possibly application hosting en-

vironment conﬁgurations.” The PaaS model al-

lows application providers to focus on their ap-

plication logic by reusing software services on

cloud and transparently leveraging the elasticity

of cloud platform. For web applications, the tools

provided by PaaS usually include identity man-

agement, data as a service, messaging infrastruc-

ture and so on. For ubiquitous applications, it is

necessary to extend these basic platform services

with the key facilities of ubiquitous computing,

including virtualized device access, device life-

cycle management, real-time data services, and

adaptive application runtime environment. By of-

fering these facilities as services on the cloud,

the development of ubiquitous applications boils

down to a composition of these services.

• Cloud-native Ubiquitous Applications

The EUPaaS will enable cloud-native ubiquitous

applications that inherently possess the following

essential cloud and ubiquitous computing charac-

teristics:

– Elasticity—Applications will be consistently

responsive to users in need regardless of the

load on underlying resources.

– Multi-tenancy—Applications can provide each

user with a highly personalized and contextual-

ized usage experience by virtually partitioning

data and devices.

– Virtualization—Applications can dynamically

bind and use heterogeneous ubiquitous devices

that are virtualized as services on the cloud.

– Context-awareness—Applications will be re-

sponsiveto users surroundings, activities, phys-

ical conditions and any perceivable information

that may affect users needs.

• Real-time Quality-aware Data Service

Producing, processing, and consuming real-time

data is a fundamental aspect of ubiquitous ap-

plications. Thus, it is desirable to extend cloud

data services with the capability of processing

large-amount of real-time data. Furthermore, in

order to accommodate the volatile data quality

in ubiquitous environments while delivering re-

liable services, quality of real-time data should

be evaluated and assured for applications. The

ongoing research on data quality has been fo-

cused on database applications (Batini and Scan-

napieco, 2006), which generally deal with persis-

tent and relatively stable data. And in ubiquitous

computing ﬁeld, the research results on Quality

of Context (QoC) are not yet feasible enough to

be adopted in large-scale context-aware applica-

tions (Bellavista et al., 2013). It is necessary to

extend and adapt conventional data quality met-

rics to suit the needs of real-time data in ubiq-

uitous environments. Quality assurance mecha-

nisms should be developed based on these met-

rics and provided as quality-aware data services

for applications. Furthermore, the capacity of

cloud can be exploited to support real-time data

quality-awareness in large-scale ubiquitous com-

puting environments.

• Adaptive Context- and Data Quality-aware

Applications

The applications in ubiquitous environments need

to be adaptive to various situations while keep-

ing them functioning and responsive. As demon-

strated in the mega-hospital environments, it is

important to enable applications to adapt to com-

plex context changes (location, activity, users

concerned, and resources available), and to data

quality ﬂuctuations (accuracy, freshness, com-

pleteness). To cope with such multiple dis-

tinct volatilities of the environment, past research

on self-adaptive systems and context-aware com-

puting should be put into perspective of cloud-

based applications: in offering ubiquitous appli-

cations with a PaaS environment, such adaptive-

ness should be provided as part of the platform

capabilities and leveraged by applications through

their cloud-based run-time environment.

CLOSER2013-3rdInternationalConferenceonCloudComputingandServicesScience

312

Figure 1: The EUPaaS architecture.

4 THE EUPaaS ARCHITECTURE

This section presents the EUPaaS architecture that re-

alizes the aforementioned core concepts.

EUPaaS follows a federated service-oriented mid-

dleware architecture. The foundationfor this architec-

ture is the UCG (Ubiquitous Cloud Gateway), which

is dedicated to integrating large amounts of sensors

and devices within a cloud infrastructure. It pro-

vides a virtualization layer for heterogeneous entities

through existing service standards. Given the ad-hoc

and dynamic nature of ubiquitous environments, a

life-cycle model of virtualized services is being de-

veloped in order to monitor these entities. At the

core of UCG is a messaging infrastructure to manage

the reliability of communication and interoperability

of devices. The messaging infrastructure takes into

consideration both the diverse requirements of ubiq-

uitous applications, as well as the nature of mobile,

volatile, light-weight devices. The UCG is designed

as a cloud-native component with support for multi-

tenancy (Strauch et al., 2012) and elasticity. Overall,

the federation of UCGs is be the key to the success of

the EUPaaS architecture. The federation eventually

allows EUPaaS platform deployments to share load

between UCG instances, integrate with geographi-

cally distributed devices, and to connect different au-

tonomous domains in mega-hospital environments.

In the EUPaaS architecture, quality-ensured, real-

time data are provided to applications as a service.

In previous work, we have developed a set of feasi-

ble metrics and applied it to real-world datasets (Li

et al., 2012). Based on these metrics, formal se-

mantic annotations is being developed for annotat-

ing data streams with quality information. The eval-

uation framework particularly considers a) the data

quality change during data processing, e.g. aggrega-

tion of data streams with different accuracies, and b)

the data quality perceived by applications, e.g. the

completeness of multiple streams required in a cer-

tain situation. The data quality assurance approach

under development is based on statistical methods,

which can dynamically compensate for quality ﬂuctu-

ation according to application requirements, available

resources, and the characters of the stream (e.g. ag-

gregative, single, etc.). The data quality model, and

evaluation and assurance mechanisms will be imple-

mented into a large-scale real-time data processing

system on cloud

. Eventually, the quality-aware data

provisioning capability is open to ubiquitous appli-

http://www.gigaspaces.com/datagrid

EUPaaS-ElasticUbiquitousPlatformasaServiceforLarge-scaleUbiquitousApplications

313

cations through service interfaces. Applications can

then specify their data quality requirements while ac-

quiring real-time information of the application envi-

ronments.

EUPaaS applications carry out the adaptive exe-

cution of complex ubiquitous applications. An EU-

PaaS application model based on service composi-

tion is being developed to assess the impact of con-

text and data quality changes to the application ex-

ecution. Since the adaptation of applications can be

triggered by various factors, the architecture is not fo-

cused on supporting isolated adaptation mechanisms

for different situations. Instead, we aim at supporting

multiple adaptation mechanisms from a wide range of

options provided by the system, which can include re-

placement of services, changing of data sources, and

modiﬁcation of the application logic. Adaptation trig-

gers are deﬁned based on real-time quality informa-

tion from the data services, a system wide view of

resources, and the application context.

5 CONCLUSIONS

This position paper presented our vision and ongo-

ing work of EUPaaS (Elastic Ubiquitous Platform as a

Service). The work is motivatedby the need for large-

scale ubiquitous systems in future mega-hospitals.

The challenges we have faced mainly include elastic-

ity, large-scale context-awareness, reliable data pro-

cessing and application adaptation. To address these

challenges, we propose to enable the seamless con-

vergence of cloud computing and ubiquitous com-

puting by researching and realizing a series of novel

concepts, including elastic ubiquitous platform as a

service, cloud-native ubiquitous applications, real-

time quality-aware data service and context- and data

quality-aware applications. The research and devel-

opment work is in progress and will be integrated into

the EUPaaS software platform.

It is worth noting that the challenges addresses in

this work are not limited to hospital environments.

The need for large-scale ubiquitous systems, pro-

pelled by the emergence of Internet of Things, is

evident in applications in smart cities, smart health-

care, smart transportation and so on. The proposed

cloud-based approach will not only enable context-

awareness in large-scale, but also extend the applica-

tion domains of cloud computing to better support the

applications based on the growingly ubiquitous sen-

sors and smart devices.

ACKNOWLEDGEMENTS

This work is sponsored by Paciﬁc Controls Cloud

Computing Lab (PC

L), a joint lab between Paciﬁc

Controls L.L.C., Dubai, United Arab Emirates and the

Distributed Systems Group of the Vienna University

of Technology.

REFERENCES

Bardram, J. E. (2009). Activity-based computing for

medical work in hospitals. ACM Transactions on

Computer-Human Interaction, 16(2):1–36.

Batini, C. and Scannapieco, M. (2006). Data Quality-

Concepts, Methodologies and Techniques. Springer

Berlin Heidelberg New York.

Bellavista, P., Antonio, C., MARIO, F., and FOSCHINI, L.

(2013). A Survey of Context Data Distribution for

Mobile Ubiquitous Systems. ACM Computing Sur-

veys, 45(1).

Chun, B.-G., Ihm, S., Maniatis, P., Naik, M., and Patti, A.

(2011). CloneCloud-Elastic Execution between Mo-

bile Device and Cloud. In Proceedings of the sixth

conference on Computer systems - EuroSys ’11, page

301, New York, New York, USA. ACM Press.

Li, F., Nastic, S., and Dustdar, S. (2012). Data Qual-

ity Observation in Pervasive Environments. In The

10th IEEE/IFIP International Conference on Embed-

ded and Ubiquitous Computing (EUC 2012), Paphos,

Cyprus.

Liu, F., Tong, J., Mao, J., Bohn, R., Messina, J., Badger, L.,

and Leaf, D. (2011). Nist cloud computing reference

architecture. NIST Special Publication, 500:292.

Satyanarayanan, M. (2010). Mobile computing: The next

decade. In Proceedings of the 1st ACM Workshop on

Mobile Cloud Computing & Services Social Networks

and Beyond - MCS ’10, pages 1–6, New York, New

York, USA. ACM Press.

Strauch, S., Andrikopoulos, V., S´aez, S. G., Leymann, F.,

and Muhler, D. (2012). Enabling Tenant-Aware Ad-

ministration and Management for JBI Environments.

In (to appear in) Proceedings of the 5th International

Conference on Service-Oriented Computing and Ap-

plications (SOCA’12). IEEE Computer Society Con-

ference Publishing Services.

Weiser, M. (1991). The computer for the 21st century. Sci-

entiﬁc American, 265(3):94–104.

CLOSER2013-3rdInternationalConferenceonCloudComputingandServicesScience

314