DEVELOPMENT TOOLS FOR PERVASIVE COMPUTING IN

EMBEDDED SYSTEMS (PECES) MIDDLEWARE

Ran Zhao, Kirusnapillai Selvarajah and Neil Speirs

School of Computing Science, Newcastle University, Newcastle, NE1 7RU, U.K.

Keywords: Smart space, Middleware, Pervasive computing, Wireless networking, Context ontologies, Modelling,

testing, Dynamic addressing, Communication gateway, Eclipse.

Abstract: The main objective of the PECES project is the development of system software to enable the

communication among heterogeneous devices across multiple smart spaces, breaking the traditional barrier

of “smart islands” where only the services offered in a nearby spatial area can be used easily. PECES

development tools help the application developer to build and test the PECES middleware based

applications. This paper presents a set of tools, namely Peces Project, Peces Device Definition, Peces

Ontology Instantiation, Peces Service Definition and Peces Role Specification Definition which enable

application developers to build the PECES middleware based application using the novel concepts such as

role assignment context ontologies and heterogeneous communication technologies. Furthermore, this paper

proposes new tools which enable to model and simulate the smart space applications.

1 PECES PROJECT

The objective of the PECES project (PECES project,

2009) is the creation of a comprehensive software

layer to enable the seamless cooperation of

embedded devices across various smart spaces on a

global scale in a context-dependent, secure and

trustworthy manner. The increasing number of

devices that are invisibly embedded into our

surrounding environment as well as the proliferation

of wireless communication and sensing technologies

are the basis for visions like ambient intelligence,

ubiquitous and pervasive computing. The benefits of

these visions and their undeniable impact on the

economy and society have led to a number of

research and development efforts. These include

various European projects such as EMMA (EMMA,

2006) that develop specialized middleware

abstractions for different application areas such as

automotive and traffic control systems or home

automation. These efforts have enabled smart spaces

that integrate embedded devices in such a way that

they interact with a user as a coherent system.

However, they fall short of addressing the

cooperation of devices across different

environments. This results in isolated “islands of

integration” with clearly defined boundaries such as

the smart home or office. For many future

applications, the integration of embedded systems

from multiple smart spaces is a primary key to

providing a truly seamless user experience. Nomadic

users that move through different environments will

need to access information provided by systems

embedded in their surroundings as well as systems

embedded in other smart spaces.

The PECES project is committed to developing

the technological basis to enable the global

cooperation of embedded devices residing in

different smart spaces in a context-dependent,

secure, and trustworthy manner. The most

innovative features of the PECES middleware are to

enable the communication among heterogeneous

devices across the different smart spaces using

dynamic addressing, security and context ontologies.

The PECES project has the several scientific and

technical objectives such as development of a

flexible ontology, development of a middleware,

development of set of application development tools

and validation of the abstractions using lab tests and

prototype applications.

In this paper, the PECES development tools are

discussed. Section 2 describes the PECES project

and its novel features such as role specification,

dynamic addressing, and gateway concepts. Section

3 provides an introduction to the development

Zhao R., Selvarajah K. and Speirs N..

DEVELOPMENT TOOLS FOR PERVASIVE COMPUTING IN EMBEDDED SYSTEMS (PECES) MIDDLEWARE.

DOI: 10.5220/0003406900210026

In Proceedings of the International Conference on Wireless Information Networks and Systems (WINSYS-2011), pages 21-26

ISBN: 978-989-8425-73-7

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

environments for pervasive computing. Section 4

describes the PECES development tools that have

been developed by the PECES consortium for

PECES middleware based application development.

An additional set of tools that are currently being

under development to provide features to test and

analyse the smart space based application is given in

Section 5. Conclusions are presented in Section 6.

2 PECES MIDDLEWARE

The PECES project consortium has built the targeted

cooperation layer on top of BASE middleware

(Becker, Schiele, Gubbels and Rothermel, 2003).

The BASE layered architecture is shown in Figure 1.

This enables the project consortium to focus the

development efforts on the novel and innovative

features of the PECES middleware. BASE is freely

available as open source under BSD license which

facilitates the necessary modifications and

extensions and enables the free reuse – even for

commercial exploitation. The BASE middleware

enables the communication between devices that are

within communication range. Yet, in order to

achieve the goal of providing a cooperation layer

that enables the seamless interaction within and

across the boundaries of a single smart space, it is

necessary to extend the BASE middleware concepts.

The extension of the BASE middleware focused on

communication gateways, dynamic addressing and

smart space and internet registry.

Figure 1: BASE Architecture.

2.1 Communication Gateway

Due to the heterogeneity of devices and

communication technologies, it is not safe to assume

all future devices will be equipped with the same set

of communication technologies. As an example

consider that a sensor node might only be equipped

with ZigBee but not with Bluetooth in order to

enable energy efficient communication. Thus, in

order to enable a Bluetooth device to communicate

with such sensor nodes, it is necessary to use a

device that is equipped with both technologies as a

local gateway. Similarly, due to the associated costs

and other factors, not all devices will have a direct

connection to a global interconnection network like

the Internet. In order to enable the communication

between devices that are not directly connected to

the Internet, it is necessary to enable some devices to

act as remote gateways for others.

PECES middleware support local gateways as

well as remote gateways. The main difference

between these two types of gateways is that the local

gateway locally shares the required knowledge. In

the remote case, the knowledge sharing should be

restricted to a minimum in order to avoid the costly

distribution of frequently changing information. The

remote gateways need to be realized differently in

that they require an external entity to distribute the

information that is distributed by means of device

discovery in the local case. This information is

distributed by means of the registry that is specified

in the PECES Communication Mechanism and

Registry Interface Specification deliverable (PECES

project, 2009).

2.2 Generic Role Assignment

Due to the continuous changes in context, the

mobility of devices and the network topology can be

highly dynamic in the pervasive computing

environments. So that it is vital to enable pervasive

computing applications such as PECES prototype

applications to adapt to the continuous changes in

context and device availability. The responsibility

for adaptation can be shifted between different

entities. In cases where changes are infrequent, a

user may manually configure and adapt the system.

However, if changes are frequent, manual

configuration and adaptation are clearly not a viable

approach as they conflict with the goal of distraction

free support for tasks. In order to mitigate this, the

adaptation can be automated through the application.

This approach relieves the user from performing

manual adaptation but it complicates the

development of applications and it may result in

inefficiencies in cases where multiple applications

implement and use similar adaptation mechanisms.

As a result, the PECES middleware is aiming to

automate the initial configuration and the continuous

adaptation to changes in order to shield the user and

the application developer from the accompanied

complications.

In order to be suitable for a broad range of

different systems and in order to minimize the

utilization of resources that are required for

automation, PECES provides configuration and

Application

Layer

Micro‐

broker

Layer

Plug‐in

Layer

Invocation

Broker

Plug‐inManager

Object

Registry

Device

Registry

Plug‐ins

Invocations

Device&Plug‐

inDescriptions

Invocation

Handlers

Proxies/Skeletons Proxies/Skeletons

ApplicationObjects Services

WINSYS 2011 - International Conference on Wireless Information Networks and Systems

adaptation support by means of a uniform

abstraction. To create a uniform abstraction that is

suitable for a broad range of different configuration

tasks, it is necessary to introduce a clear separation

between the result of a configuration, the

computations that need to be done to produce it and

the utilization of this result. This enables the reuse of

the same basic mechanisms for different tasks.

Generic role assignment provides such a uniform

abstraction. More detailed information about the role

assignment concepts can be found in PECES

Addressing Scheme Specification deliverable

(PECES project, 2009).

2.3 Smart Space

A smart space can be defined as a group of

networked devices that cooperate to support their

users. The boundaries of a smart space are typically

defined on the basis of a geographic location, e.g. a

room or a building. However, such narrow

definitions are not flexible enough to support the

application prototypes in the PECES project.

Obviously, these smart spaces cannot be defined on

the basis of a single location. For example in

applications based upon a car, the whole car, i.e. the

smart space itself, is mobile.

To support the smart space concept, the PECES

middleware introduces three additional components

which are coordinator, member and gateway. These

components can be easily motivated by looking at

the anatomy of the smart spaces that are identified in

the PECES Use-Case Specification deliverable

(PECES project, 2009). The coordinator is

responsible for identifying the members of the smart

space defined by the role specification. The Gateway

functionality provides connectivity for other devices

(for the coordinator and device) in the smart space.

3 DEVELOPMENT

ENVIRONMENTS

The PECES project provides a set of development

tools that can be used by the application developers

to develop, test, and analyse their applications.

Development tools also provide an environment to

simulate/emulate applications. These types of

development tools are economical because

developers can carry out experiments without the

actual hardware and it is a feasible way to test

scalability of any proposed applications. For

wired/wireless networked based application and

protocol development, there are many simulators

(NS 2, OPNET) have been proposed.

A Middleware based application framework for

Active Space applications was proposed in (Roman

and Campbell, 2003). The Active Space consists of

the Gaia middleware OS (Roman and Campbell,

2000) managing a distributed system composed of

several systems. The framework focuses on

providing an application framework that leverages

the functionality provided by the Gaia middleware

OS to assist developers in the construction of Active

Space application.

The Distributed Trust Toolkit (DTT) (Lagesse,

Kumar, Paluska and Wright, 2009) proposed a

framework for implementing and evaluating trust

mechanisms in pervasive computing systems and

introduced two new abstractions: trust groups and

trust blocks. Trust groups allow associated

application devices to share recorded trust data and

trust computations. Trust blocks makes policy

decisions based on data gathered by the computation

component which implements network based trust

protocols and allows the DTT to interoperate with

legacy trust systems. The DTT facilitates the

extension and adaptation of trust mechanisms by

abstracting trust mechanisms into interchangeable

components.

UbiWise, a simulator for ubiquitous computing

system was proposed in (Barton and

Vijayaraghavan, 2002). UbiWise concentrates on

computation and communication devices situated

within their physical environments. Multiple users

can attach to the same server to create interactive

ubiquitous computing scenarios. The devices are

specified through a combination of a device-

description file in XML and Java. UbiREAL

simulator (Nishikawa, et al., 2006) was proposed for

realistic smart space systematic testing. UbiREAL

facilitates reliable and inexpensive development of

ubiquitous applications where application software

controls a lot of information appliances based on the

state of external environment, user’s contexts

information.

A suite of tools which used to construct domain

models and knowledge-based application with

ontology was provided by Protégé (Protégé, 2004).

Protégé implements a rich set of knowledge-

modelling structures and actions that support the

creation, visualization, and manipulation of

ontologies in many different representation formats.

Protégé also provide a Java API for other developer

to build their own tools and applications.

DiaSuite (Cassou, Bruneau and Consel, 2010)

framework includes a domain-specific design

DEVELOPMENT TOOLS FOR PERVASIVE COMPUTING IN EMBEDDED SYSTEMS (PECES) MIDDLEWARE

language, a compiler for this language, which

produces a Java programming framework, an editor

to define simulation scenarios, and a 2D-renderer to

simulate pervasive computing applications. DiaSim

(Jouve, Bruneau and Consel, 2009) is a part of the

DiaSuite which provides a parameterized simulator

for pervasive computing applications. A modelling

and simulation framework is proposed in (O’Neil,

Conlan and Lewis, 2010) to assist context aware

system design for pervasive computing applications.

The simulator provides analysis tool with

information relating to the actual state of the

environment and not just the sensed state that the

prototype system receives. The framework also

presented number of validation case studies.

The tools discussed here provide limited support

for application developers, as they have been

designed for different goals and concepts. Although

these tools are available to support pervasive

computing environment application development,

they only offer little methodological support for role

assignment, context-awareness and security, which

are the core features of the PECES middleware. For

example, Protégé may provide support for context

ontology instantiation for the devices but this

information may be difficult to integrate with other

PECES middleware services without the use of any

specific tools.

4 PECES DEVELOPMENT

TOOLS

PECES development tools focus on configuring

devices, modelling smart spaces and context

dynamics and testing the novel concepts provided by

the PECES middleware. The tools provide support

for application developers to build PECES

middleware application and simulate and analyse the

smart space behaviours with respect to the context

dynamics and network changes. The PECES project

provides a set of tools which are integrated into the

Eclipse development environment. This way, the

usual development assistance provided by the

Eclipse IDE can also be used for PECES focused

development support.

The following sections explain the different tools

which are the Peces Project, Peces Device

Definition, Peces Ontology Instantiation, Peces

Service Definition and Peces Role Specification

Definition. The following subsections are arranged

according to the development sequence that the

developers would typically follow during the

middleware application development.

4.1 Peces Device Definition Tool

As the first part of the development process,

developers should use Peces Project tool to generate

a PECES general project in the Eclipse workspace

(for example, PrototypeDemo project can be created

by the Peces Project tool with ConfigurationTool,

ModellingTool and TestingTool folders). The Peces

Device Definition tool now be used to define

communication plugins such as IP, Bluetooth,

ZigBee (e.g. MxIPBroadcastTransceiver,

MxIPMuticastTransceiver, MxSpotTransceiver), and

device functionalities (Coordinator, Gateway,

Coordinator&Gateway, Member) and also device

name. All smart space applications should define

one device with coordinator functionality which is

responsible of the coordination of the smart space.

The devices can be placed using drag and drop

method. Different colours will be shown according

to the selected device functionality (e.g.,

Coordinator is red).

Figure 2 shows an example in which three

devices are defined (Consumer, PrintTextProvider

and PrintNumberProvider). The screenshot also

shows three different Java projects namely

Consumer, PrintTextProvider and

PrintNumberProvider in the workspace. These Java

projects contain PECES middleware libraries (peces-

1.0.jar) and necessary Java files under the src folder.

Figure 2: Screenshot of the Peces Device Definition Tool.

4.2 Peces Ontology Instantiation Tool

The PECES context ontologies are composed by the

SmartSpace, Measurement, Device profile, User

Profile and Event ontologies and these ontologies

are freely available to download at the project

website (PECES project, 2009). The PECES Context

WINSYS 2011 - International Conference on Wireless Information Networks and Systems

Ontology and Query Specification deliverable

(PECES project, 2009) clearly explains the

dependencies among them, as well as the external

ontologies which provide a basis for the PECES

concepts and properties. The core ontology for

representing contextual information of a smart space

is the SmartSpace ontology. The basic concepts to

model the contextual information of a smart space

are Device, Context, Location and Service.

The Device profile ontology provides

vocabularies to model specification of devices inside

smart space. There are three categories of devices

defined in the PECES prototype application which

are PECESEmbeddedDevice, Accessory and

SensorDevice. PECESEmbeddedDevice represents

those embedded devices that deploy the PECES

middleware. There are three categories of embedded

devices according to their role inside a smart space,

namely gateway, coordinator and member. In order

to specify which kind of accessories an embedded

device has, the property hasAccessory can be used to

link a PECESEmbeddedDevice instance to an

Accessory instance. Accessory instances are

Keyboard, Touch Screen, Speaker, Screen and

Microphone. In addition to this, SensorDevice has

two sub-concepts: Detector and MeasuringSensor.

A MeasuringSensor instance represents a sensor

which can measure a measurement such as light,

noise, temperature, etc.

Figure 3: Screenshot of the Peces Ontology Instantiation

Tool.

Peces Ontology Instantiation tool enables

application developer to instantiate the devices. The

tool automatically loads the participating device

name and its functionality information from the

project.xml file which is generated by the Peces

Device Definition tool. The Peces Ontology

Instantiation tool provides GUI where application

developers can add instances and link properties.

When the instantiation process is completed, the tool

creates a RDF file (project.owl) in the

PrototypeDemo project ConfigurationTool folder

and also creates *.peces.ctx files which contains the

device context information for each devices. Those

device *.peces.ctx files are placed in the appropriate

device Java project and once the file is placed in the

Java project, the tool will automatically create

necessary Java files for the middleware from the

*.peces.ctx files.

Figure 3 shows an example in which three

devices defined (Consumer, PrintTextProvider and

PrintNumberProvider) in the Device Definition tool

are automatically loaded by the Ontology

Instantiation tool. Two new services

(printTextService and printNumberService) are

defined here with Ontology Instantiation tool. The

printTextService linked (via provides) with

PrintTextProvider device and printNumberService

linked (via provides) with PrintNumberProvider

device. Also the Consumer device is linked (via

consumes) with both printTextService and

printNumberService.

In addition to the tools that have been described

here, the PECES development tools also include

other tools such as Peces Service Definition tool and

Peces Role Specification Definition tool. The

Service Definition tool provides features to generate

necessary Proxy and Skeleton for the application

services defined in the Ontology Instantiation tool.

The “Scope” option in the Peces Service Definition

tool determines at which scope (Internet, Space, and

Device) the service will be published. The Peces

Role Specification Definition tool provides an

interface where developers can define the different

rules that the application will use to dynamically

form groups (smart spaces) of collaborative devices.

The Role Specification Definition tool automatically

generates all necessary Java code in the required

device Java projects to define and instantiate it using

the middleware. Due to space limitations we are

unable to describe these tools in more details here

but they may be provided upon request.

5 FUTURE WORK

Section 4 only presented the tools that have been

developed to provide support for application

development using PECES middleware. The next

phase of the development tools will provide support

for modelling (Peces Event Tool), emulating and

visualising (Peces Testing Tool) the PECES

middleware based smart space application.

The Peces Event tool will be used to generate

several events such as device switch ON, device

DEVELOPMENT TOOLS FOR PERVASIVE COMPUTING IN EMBEDDED SYSTEMS (PECES) MIDDLEWARE

switch OFF, context changes and network topology

changes. This tool will also provide features (Event

Diagram Editor) to define the generated events in a

sequence. The defined event sequence information

will be stored in a XML file (events.xml) and placed

in the ModellingTool folder. The events.xml file will

have different tags for different events with delay

time information.

The Peces Testing tool will provide support to

execute the applications (each device application

considered as separate JVM) built by the other tools.

The Testing tool will load necessary device related

information from the project.xml file and event

related information from the events.xml file and

display in the Testing tool Multi-page editor page.

Application developers will be able to define

necessary time for the test. All middleware and

application related information will be appended

into a single log file with the specific JVM (with

device name) information and the log data will be

used to visualise and analyse the smart space

features.

6 CONCLUSIONS

One of the main objectives of the PECES

middleware is to provide a cooperation layer that

enables seamless interaction and coordination

among devices in and across smart spaces in a

secure manner. This paper presented a set of tools

which provide support for PECES middleware based

application development. The tools provided support

for device configuration, instantiation, role

specification and service definition. This paper also

outlined a new set of tools which are currently under

development to provide features to model and

simulate the smart space applications. The new set

of tools will provide features for dynamics modeling

testing visualizing the smart spaces. It is the authors’

intention to present at the conference that the tools

already developed as well as the new set of tools

which are currently under development.

ACKNOWLEDGEMENTS

The work presented here is sponsored by EC under

FP7 programme (FP7-224342-ICT-2007-2) and

authors also would like to thank all the project

partners for their contributions.

REFERENCES

Barton, J., and Vijayaraghavan, V., 2002. UBIWISE, A

Ubiquitous Wireless Infrastructure Simulation

Environment, [online] Available at:

<http://www.hpl.hp.com/techreports/2002/HPL-2002-

303.html> [Accessed December 2010].

Becker, C., Schiele, G., Gubbels, H., and Rothermel, K.,

2003. BASE - A Micro-broker based Middleware For

Pervasive Computing, In IEEE, 1st International

Conference on Pervasive Computing and

Communications, pp. 443-451, Fort Worth, USA,

March 2003.

Cassou, N., Bruneau, J., and Consel, C.. A tool suite to

prototype pervasive computing applications, In: IEEE.

8th IEEE International Conference on Pervasive

Computing and Communications Workshops

(PERCOM Workshops), Mannheim, Germany, March

2010.

EMMA Project, (2006). [online] Available at:

<http://www.emmaproject.eu> [Accessed May 2011].

Jouve, W., Bruneau, J., and Consel, C., DiaSim: A

parameterized simulator for pervasive computing

applications, In: IEEE, the Seventh Annual IEEE

International Conference on Pervasive Computing and

Communications, Galveston, Texas, March 2009.

Lagesse, B., Kumar, M., Paluska, J., and Wright, M., 2009

DTT: A Distributed Trust Toolkit for Pervasive

Systems, In: IEEE, Pervasive Computing and

Communications Conference. Galveston, Texas, USA,

March 9-13, 2009.

Nishikawa, H., Yamamoto, S., Tamai, M., Nishigaki, K.,

Kitani, T., Shibata, N., Yasumoto, K., and Ito, M.,

UbiREAL: Realistic Smartspace Simulator for

Systematic Testing, In: UbiComp. 8th International

Conference on Ubiquitous Computing, LNCS4206,

pp. 459-476, Irvine CA, USA, September. 2006.

O’Neil, E., Conlan, O., and Lewis, D., 2010. Modelling

and simulation to assist context aware system design,

[online] Available at: <http://sim.sagepub.com/

content/87/1-2/149>, [accessed May 2011]

PECES Project, (2009). [online] Available at:

<http://www.ict-peces.eu> [Accessed May 2011].

Protégé. (2010). [online] Available at:

<http://protege.stanford.edu/> [Accessed May 2011].

Roman, M., and Campbell, R., 2000. Gaia: Enabling

Active Spaces, In: SIGOPS EW'00, 9th ACM SIGOPS

European Workshop, pp.229-234, Kolding, Denmark,

September 2000. New York: USA.

Roman, M. and Campbell, R., 2003. A Middleware-based

Application Framework for Active Space

Applications, In: ACM/IFIP/USENIX, International

Conference on Middleware. Rio de Janeiro, Brazil

June 2003.

WINSYS 2011 - International Conference on Wireless Information Networks and Systems