SOFIA - A Smart-M3 lab course: approach and design style to support student projects. UNIBO/NOKIA

A Smart-M3 lab course: approach and
design style to support student projects
L. Roffia, A. D’Elia, F. Vergari, Jukka Honkola
D. Manzaroli, S. Bartolini, G. Zamagni, Nokia Research Center
T. Salmon Cinotti Itämerenkatu 11-13, 00180 Helsinki, Finland
Alma Mater Studiorum – Università di Bologna jukka.honkola@nokia.com
Via Toffano, 2, Italy
lroffia@arces.unibo.it

Abstract

This paper reports on the lesson learned from setting up and running a course on smart space based
ambient intelligence within a “postgraduate” degree in Information & Computer Science Engineering.
The course consists of a lab module where students are asked to propose, develop and demonstrate simple
smart environment applications. Specifically, the course is built around Smart-M3, the open source
interoperability component of an open innovation platform developed within the European Project SOFIA
(2009-11). This paper will illustrate the above summarized course, presenting the design style and the
application development flow specifically devised for the course, the students evaluation approach taken,
the course achievements, the issues that still need to be handled, and it will envisage topics and further
extensions.

Index Terms: Information Interoperability, Smart Space, Application Design Style, Smart-M3, RDF,
Ontology

I. INTRODUCTION

The course was run well ahead the European Project SOFIA’s end [1]. But how could such an
unplanned course, based on the preliminary results of an underway research project be so quickly
included in a postgraduate degree curriculum? This was possible only thanks to a lucky
circumstance: the Graduate Programs Committee of our ICT Department extended by 50% both
credits and teaching time assigned to our course, and gave us “carte blanche” on how and what
to teach in the additional 30 granted hours. In this circumstance, a choice had to be made: should
the current content of the course be smoothly extended, or should we rather go for a rather
disruptive education initiative? The latter choice was risky, but it was considered a win-win one,
with great potential benefits for both research and education. This was our choice, as, on one side
an open source prototype of an interoperability platform for smart environments applications, i.e.
Smart-M3 [2, 3], was available and on the other side the hosting course was named “Information
Processing Systems”, which is a very general identifier widely compatible with the proposed
initiative. Many risks were involved in our choice: particularly, no course material (such as
textbook and exercises) was available; no application design style still existed; no single person
had all the skills required to run the entire course. To deal with these risks, a training team of five
researchers was set up, all of them with complementary and course relevant skills. They jointly
defined a design style, attended all classes, gave oral presentations and provided direct support
to the students. At the beginning of the course the 40 students had been partitioned into groups of
five and each group was asked to propose an own project. After the first month each group was
asked to formalize the description of their project, using the design style and the application
development flow specifically devised for the course. Consequently the students filled out a
project presentation template that will be described in this paper.

This paper is organized in three main sections: in Section II (Course Outline) are presented
the course organization, the evaluation criteria and procedure, the course related research
framework, and an overview of the course program. In Section III (Smart-M3) is presented the
smart space management framework that have been adopted to develop the students projects.
Section III (Design style and tools) reports the methods and tools that have be designed to
support students projects. In Section IV (Conclusion) conclusion and future developments are
drawn.

II. COURSE OUTLINE

The course was implemented through the following steps: first of all, the students were
introduced to the smart environments scenario by presenting candidate application domains.
Then the need to connect physical world elements, such as places, users, objects and devices, to
the information world was discussed, together with the interoperability requirements. A
significant part of the course was focused on the semantic data representation. Students were
introduced to RDF and OWL and they were guided to the use of the Protégé ontology editor [4].
Moreover, students were introduced to Smart-M3 (see Section III). At the end, a design style was
presented as reference model for projects development (see Section IV).
A. Lab setup
The course has been supported by the flexible Moodle e-learning platform [5]. The “Smart-
M3 @UNIBO Moodle” [6] includes a WiKi which has been used to share student projects and
still informal course material, e.g. slides, templates, APIs and links. Each group had its own
dedicated workstation (and WiKi page) and, in order to do not interfere with other groups,
Smart-M3 has been installed on each single workstation, i.e. 10 interconnected workstations.
Each workstation runs Microsoft Windows as host and Ubuntu as guest, i.e. using VirtualBox
[7]. The workstation configuration allows also to use a shared networked instance of Smart-M3,
i.e. each workstation has two IP address, one for the host and one for the guest. This
configuration allows also to develop applications on two different operating systems, i.e. Ubuntu
and Microsoft Windows.
B. Student projects
In this subsection, we report the proposed projects on which students have been working on.
1. Smart Multi-robot Environmental Mapping: Lego Mind Storm® robots move within an
environment aiming to create a map of it. The map creation is displayed in real-time on a
smart phone. Topics: Space representation; SIB based coordination; Lego Mind Storm® KP
adapter; smart phone
2. Smart Traffic Light (evaluated): collision avoidance between two Lego Mind Storm® cars
moving on two tracks that have one intersection (simulated solution). Topics: SIB based
coordination; Lego Mind Storm® KP adapter
3. Smart Domus: monitoring power consumption at home and applying power saving policies.
Topics: Presence monitor; appliances power monitor; power saving algorithms
4. Smart Cupboard (evaluated): monitoring products store to notify the user smartphone when
quantities go under thresholds. Topics: Alarm generation; barcode products identification;
smart phone
5. Smart Speechy: send vocal commands to devices, e.g. TV, Hi-fi, lights. Topics: JAVA
Speech Library; commands & actions
6. Smart Cart: creating a shop list at home via Web. At the supermarket the cart identifies
the user and it helps her to find products. Topics: User and product RFID identification;
RFID reader KP adapter

7. Smart Coffee Maker: setting coffee preferences and have it ready in your hotel room when
you wake up. Topics: Customer preferences; smart devices emulation
8. Smart Secure Chat: a secure chat system supporting also room based chats. Topics: Secure
login; messages dispatching; messages garbage collection
All the above projects have been described and implemented using the design style and tools
presented in Section IV.
C. Course introduction, basic concepts on smart environments and ontologies
As first step, students were introduced to smart environments. The first part of the course has
been fundamental also to share a “common language”. As shown in Figure 1, we started from
presenting the relationships between physical entities and their digital representation, i.e. Smart
space. Within this framework, students understood the role of sensors, i.e. extracting data from
the physical word. With respect to this, some sensing technologies were presented, like RFID,
inertial sensors, magnetic sensors and location systems (see Where? in Figure 1). After that, we
discussed the need to unique identify each single physical entity, i.e. devices, users, objects,
places (see Who? in Figure 1), within the physical space. As next step, the concept of data model
was introduced: each physical entity and its relationships should be described in a proper way
into the smart space (see What? in Figure 1). As introduction to the semantic data model, the
Semantic Web project was introduced [8, 9]. A bottom up approach to ontologies was followed.
First of all, the XML (Extensible Mark-up Language) [10] was presented as a language for
representing data and their corresponding metadata as a tree structure. After that, the Resource
Description Framework (RDF) [11] was discussed and the Universal Resource Identifier (URI)
[12] was introduced. Moreover, the definition of an RDF statement as a triple <subject,
predicate, object> where the subject and the predicate are univocally identified through URI,
while the object may be a URI or a literal, i.e. a string or a number, was given. Also, it was
shown as a set of RDF statements correspond to a graph: the subject and the object are
represented by nodes, while each predicate is represented by a labelled edge connecting two
nodes. Then we moved to the concept of knowledge base by presenting the RDFS (RDF Schema)
[13] and the OWL (Ontology Web Language) [14, 15]. As conclusion to the semantic data
representation, some examples of OWL ontology modelling were presented using the Protégé
ontology editor [4].

Figure 1. Connecting the physical word to the information word

The course introduction ended presenting the role of a smart space as a “semantic interface”
between the physical and the digital world (see Figure 1) and introducing Smart-M3 as the
reference smart space of the course (see Section III for more details on Smart-M3).

E. Evaluation criteria and procedure
The formal examination requirement is to evaluate the team participants individually, and the
evaluation criteria - specified during the course - were identified as follows:
o Level of understanding of the big picture and of the role in the Information Society of
information interoperability, context aware computing and pervasive computing
o Level of understanding of the course goals and required engineering practices
o Motivation of the proposed project
o Individual contribution to the same
To this end the examination procedure was defined as follows:
o Each project team was evaluated in a single comprehensive session partitioned into five
sequential and interactive presentations, each one assigned to an individual student:
o The first student motivates and illustrates the project, supported by a Power Point
presentation previously prepared, which includes the scenario and the KP templates (see
Section IV)
o The second student discusses the project ontology, motivating the choices and describing
the ontology definition process (see Section IV)
o The third student runs the demo
o The fourth student describes the design process, the tools used and the design itself
o The fifth student describes the code
The session has been very interactive and during each session useful feedback is gained to
improve next year course and possibly the platform and the design process themselves. In the
end, these examination sessions turn out to be interesting, fun and a potential opportunity to be
run in large classrooms to demonstrate their value to a larger public. So far we did not have any
public to the examination sessions and we evaluated just 2 groups on 8 (see Section II.B).

III. SMART-M3

Smart-M3 is an open source implementation of an RDF [16] based information
interoperability system [2, 17]. The system consists of Knowledge Processors (KPs) connected
to one or more Smart Spaces. The KPs share information by modifying information in a smart
space by means of insert, remove and update operations. The KPs can also query for information
using supported query languages, and set up persistent queries, subscriptions, which provide
notifications to KPs when the query results change. In addition, there are operations for KPs to
join and leave a smart space, which can be used to implement access control.
A domain model of the system is presented in Figure 2 describing the concepts of M3 smart
spaces, and Table 1 shows an overview of the operations available to modify and access the
information.

Figure 2. Domain Model of M3 Smart Space

Table 1. Basic Smart Space operations

Operation Description of operation
name
Join Joins a KP to a named M3 space if the credentials match what is required.
Leave Leaves a named M3 space. No more operations may be performed after a
leave until a join.
Insert Atomically inserts a graph into a smart space.
Remove Atomically removes a graph from a smart space.
Update Atomically updates a graph in a smart space. This is an atomic
combination of Delete and Insert operations, where Delete is performed
first.
Query Queries for information in the smart space using one of the supported
query mechanisms.
Subscribe Sets up a subscription (persistent query) in smart space. The KP is notified
when the subscription results change.
Unsubscribe Cancel a subscription.

The smart space is realized by one or more1 Semantic Information Brokers (SIBs) which
together form a combined search extent of information. The KPs connect to the SIB by using
Smart Space Access Protocol (SSAP). SSAP describes the protocol messages needed to perform
the operations described in Table 1, and can be used over multiple connectivity mechanisms,
such as TCP/IP.

1
The current implementation does not support multiple SIBs

Smart-M3 has implemented SSAP protocol KP side front-ends (KP Interface, KPI) for three
different programming environments: GLib/C, Qt/C++ and Python. In addition, there are
independent KP side SSAP implementations for Java [18], ANSI C [19] and C#.
In addition to the SSAP libraries, there are ontology code generators for GLib/C, Python, and
ANSI C [20] which aim to ease the development work by automatically generating class
definitions from ontology, allowing developers to program using familiar domain concepts
instead of using SSAP operations and RDF directly. The Smart-M3 implementation architecture
is described in Figure 3.

Figure 3. Smart-M3 Implementation Architecture

In the course, the SIB and python parts were utilized, as well as the Java and C# KPI.
Ontology code generators were not used in the course. Students have been free to choose on
which programming language(s) develop their projects.

IV. DESIGN STYLE AND TOOLS

This section reports on the method that has been design to support student projects. It is
composed by four main parts: a scenario template, a Power Point presentation, a KP taxonomy
and the KP templates. All of them need to be implemented to properly describe a SSA. In the
following subsections, each part of the method, i.e. design style, is presented with reference to a
simple SSA example.
A. Scenario template
This template consists of three tables and it provides a scenario description along with the
list of the involved actors, i.e. users and artifacts (see Table 2, Table 3, Table 4).

Table 2. Executive summary

Table 3. End users and their role

Table 4. Artifacts and their users

B. Project presentation
The project presentation consists of several sections: the application layout, the software
architecture, i.e. in terms of KPs and their relationships, the hardware-software infrastructure,
i.e. on which device each KP runs, the run-time data flow, the application ontology, and the
ontology awareness level of each KP, i.e. which are the instances that each KP is aware of.
As shown in Figure 4, the application layout aims to give a general graphical overview of the
SSA. As reported in Table 2, the described SSA example is related to the localization of a
general user (U1) who is carrying a nomadic device equipped with an RFID reader (A1). In this
example there are two rooms, each of them identified by an RFID tag. The current user position,
i.e. room1 or room2, is monitored by a monitor operator (U2) on a remote computer (A2).

Figure 4. Application layout

Once the application layout is presented, students are asked to describe the SSA in terms of
KPs and their interactions with the smart space, i.e. software architecture (see Figure 5).
The next step consists in describing on which device each KP runs, in which programming
language it has been developed and where the smart space is located, i.e. hardware-software
infrastructure (see Figure 6).
In order to have a general overview of the data flow at run-time, i.e. abstracting from the
presence of the smart space, another description has been designed: as shown in Figure 7, here
are represented the KPs, connected through labelled arrows, i.e. on each arrow is reported the
“concepts flow” extracted from the software architecture (see Figure 5).

Figure 5. Software architecture

Figure 6. Hardware-software infrastructure

Figure 7. Run-time data flow

The following steps aim to present the data model in terms of the reference ontology, the
instances and the ontology awareness of each KP. First of all, the ontology class tree is presented
(see Figure 8).

Figure 8. Ontology class tree

As Smart-M3 stores the ontology instances as RDF triples, the main RDF instances are
represented showing the relevant predicates (see Figure 9).

Figure 9. Instances

The project presentation ends by presenting the ontology awareness for each KP. The focus
is on showing which are the RDF triples a KP must be aware of. In Figure 10 is shown an
example related to the “E2_User Location Detector” KP that acts as aggregator within the SSA
example. Note that, as shown in Figure 10, at this step is important to underline if the KP modify
some predicates, i.e. semantic connections.

Figure 10. An example of ontology awareness for the E2_User Location Detector KP

C. KP Taxonomy

The purpose of the KP taxonomy is to facilitate the smooth growth of SSAs and maximize
KP reuse by assisting the developer in searching a suitable KP from the KP repository and
storing new ones therein. Usually a KP originates as an application-specific KP (it is produced
because there is a specific need in a specific scenario), and it has then the opportunity to be
promoted through the entire KP category-chain through a generalization process. The work of
building a KP taxonomy (and also a more general KP ontology) is taken in change within the
SOFIA project. An example of a KP taxonomy from a SOFIA public deliverable is shown in
Figure 11. With reference to the SSA example presented in this section (see Figure 11), the three
KPs have been classified according to this taxonomy as shown in Figure 11.

Figure 11. Kp taxonomy: an example from a SOFIA public deliverable

D. KP templates
In order to have a quick reference for each KP, a template has been design. The template
consists in two tables: the first is used to give a short description and a name to the KP; the
second one is used to summarize how the KP interacts with the smart space in terms of
exchanged RDF triples. The second table includes also some implementation details, like the

programming language used to implement the KP, if the KP exposes a GUI and the SIB version
for which the KP has been developed for (see Table 5).
Table 5. The KP template

V. CONCLUSION

As conclusion, a new education approach was experimented. The real value of the course
within the addressed curriculum may not be assessed yet. Certainly, a community of Smart-M3
developers was created. All of these students will have to do their six months postgraduate thesis
next year. All of them are therefore candidates for research contribution to projects based on
SOFIA Interoperability Platform. Therefore we may claim that a fruitful closed loop research →
education → research was set up which could turn out to be quite interesting in the coming
years. The student background was mostly software, and no hardware lab was available.
Therefore, during this first course edition no hardware adapters to connect legacy devices to the
smart space could be developed, i.e. no real device was connected to the Semantic Information
Broker. Devices and sensors were mostly simulated. This was a course limitation, which could
be overcome in the future by admitting to the course not only information engineering but also
electrical engineering students. We would like also to make project on device and service level
interoperability, experimenting different solutions like the one proposed by NoTA.
This experience allowed us also to create a new team of Smart-M3 trainers that, along with
the implement design style, should hopefully give a contribute to research and teaching on smart
space and more in general on smart environments. There are still some needs to be handled and
we envisage topics and further extensions for the future. First of all, one of the aims will be to
build a multi-domain smart environment by fusing together all the ontologies (and KPs) derived
from each single project. This will bring to have a “Smart-M3 Lab Ontology” to support both
current and future projects. We would like also to spend more effort on KP taxonomy modelling
in order to have a large set of reusable and meaningful KPs. Trying to find a synergy with other
courses in the same term will be also a challenge: introducing agent systems technologies, AI
algorithms and advanced reasoning mechanisms will be a promising way to enhance the
smartness of future SSAs.

ACKNOWLEDGMENT

This paper is part of the dissemination activity carried out within the European Research Project
SOFIA (http://sofia-project.org/) coordinated by NOKIA and funded by the European
Commission through the ARTEMIS JU, contract n. 100017.
The course presented could not be held without the support from the SOFIA research team from
NOKIA which greatly contributed with seminars, material, support and, encouragement.
A special thank goes to the 40 Cesena students who took part to the first run of the course,
providing precious feedback for further education innovation.

REFERENCES

[1] "SOFIA Project," ARTEMIS - VII EU FP, 2009-2012, pp. http://www.sofia-project.eu/.
[2] J. Honkola, L. Hannu, R. Brown, and O. Tyrkko, "Smart-M3 information sharing
platform," IEEE Symposium on Computers and Communications (ISCC), pp. 1041-1046,
2010.
[3] "Smart-M3 on Wikipedia," http://en.wikipedia.org/wiki/Smart-M3.
[4] "Protégé Ontology Editor," Stanford University School of Medicine,
http://protege.stanford.edu/.
[5] "Moodle," http://moodle.org/.
[6] "Smart-M3 Lab@Unibo," University of Bologna (Cesena) - Italy,
http://www.moodle.unibo.it/mod/wiki/view.php?id=10624.
[7] "VirtualBox," http://www.virtualbox.org/.
[8] T. Berns-Lee, J. Hendler, and O. Lassila, "The semantic web," Scientific American, vol.
284, pp. 34-43, 2001.
[9] C. Bizer, T. Heath, and T. Berners-Lee, "Linked Data - The Story So Far," International
Journal on Semantic Web and Information Systems (IJSWIS), vol. 5, pp. 1-22, 2009.
[10] "Extensible Markup Language (XML)," W3C, http://www.w3.org/XML/.
[11] "Resource Description Framework (RDF)," W3C, http://www.w3.org/RDF/.
[12] T. Berners-Lee, R. Fielding, and L. Masinter, "Uniform Resource Identifier (URI):
Generic Syntax," RFC 3986, http://labs.apache.org/webarch/uri/rfc/rfc3986.html January
2005.
[13] "RDF Vocabulary Description Language 1.0: RDF Schema," http://www.w3.org/TR/rdf-
schema/.
[14] "OWL Web Ontology Language Overview ", http://www.w3.org/TR/owl-features/.
[15] H. Ian, "Ontologies and the semantic web," Commun. ACM, vol. 51, pp. 58-67, 2008.
[16] "Resource Description Framework," Accessed October 7th 2010,
http://www.w3.org/RDF/.
[17] I. Oliver and J. Honkola, "Personal Semantic Web Through A Space Based Computing
Environment," presented at Second IEEE International Conference on Semantic
Computing, Santa Clara, CA, USA, 2008.
[18] "Smart-M3 Java KPI Library," Accessed October 7th 2010,
http://sourceforge.net/projects/smartm3-javakpi/.
[19] "KPI low," Accessed October 7th 2010, http://sourceforge.net/projects/kpilow/.
[20] D. Korzun, A. Lomov, P. Vanag, S. Balandin, and J. Honkola, "Generating Modest High-
Level Ontology Libraries for Smart-M3," presented at UBICOMM, Florence, Italy, 2010.

SOFIA - A Smart-M3 lab course: approach and design style to support student projects. UNIBO/NOKIA

More Related Content

What's hot (20)

Viewers also liked (6)

Similar to SOFIA - A Smart-M3 lab course: approach and design style to support student projects. UNIBO/NOKIA (20)

More from Sofia Eu (20)

SOFIA - A Smart-M3 lab course: approach and design style to support student projects. UNIBO/NOKIA