Figure 1. Proprietary and domain specific approach As long as health and energy communities keep living in Figure 2. Interoperable approachseparate islands, relying on their own standards andinfrastructures that lack interoperability (Fig. 1), the multi- II. APPLICATION FUNCTIONAL DESCRIPTIONdomain nature of the above applications makes them We claim that if environmental and health data are madeunsustainable due to their high complexity and cost. interoperable, then they may be abstracted to generate new In this paper we imagine a different scenario (Fig. 2) where knowledge, and thus effectively reused in innovative applicationsinformation consumers and producers are decoupled and relevant to the benefit of multiple institutions and users. The scenarioinformation is stored on a shared information search domain devised in our research is depicted in Fig. 3. We partitioned theaccessible by all the relevant actors. With this approach, a wide physical space into “rooms” (room1 and room2 in Fig. 3).range of cross-domain applications are possible, as long as the Temperature and relative humidity are sensed by Intel® iMote2data model remains consistent and domain specific knowledge is sensor nodes placed in each room. In our current implementationrepresented in an interoperable way. this data is transmitted to the shared information space through a room PC. Users wear a Zephyr Bioharness BT  and a The shared memory approach has been adopted by SOFIA, smartphone. The Bioharness senses skin temperature, heart rate,an ongoing European Project lead by NOKIA®* which is respiration frequency, posture angle and an activity index, andproposing a platform for sharing interoperable information in transmits this to the smartphone which, in turn, feeds the sharedsmart environment applications. The platform is called the Open information store over a Wi-Fi connection. As we need user’sInnovation Platform (OIP) and its goal is to make "information" location information, each room is equipped with an RFID readerin the physical world available for smart services in embedded . An RFID tag  is attached to each user’sand ubiquitous systems. There are no a-priori restrictions on the smartphone. When a person enters a room, their tag needs to beapplication domain - cross-domain as well as domain specific read by the reader that is located by the room entrance. Thisapplications are equally supported. action is recognized and fed to the shared information space This paper describes the design and the structure of a cross- through the room-PC. This rather primitive but effective RFIDdomain application that relies on the OIP to provide innovative based location system can be swapped out with more viableservices based on the concurrent, dynamic handling of people’s solutions in future deployments. Fig. 3 also shows how dataphysiological parameters and their day to day surrounding gathered from the users and from the environment may be used.environmental conditions. The temperature and relative humidity data are abstracted into the thermohygrometric index, i.e. a bioclimatic index that is measure The rest of the paper is organized as follows: Section 2 is the of the perceived discomfort level due to the combined effect offunctional description of the addressed application; Section 3 humidity and temperature conditions. Health monitoring anddescribes the OIP, it’s terminology and main features; in Section Alarm management are still rather rudimentary. The first tracks4 the ontology is designed, the implementation is modelled, and all of a user’s properties, i.e. her health parameters together withthe main software modules having access to the OIP’s shared the thermohygrometric index and the environmental conditionsinformation domain are detailed; in section 5 usual conclusions of the place where she is located. The alarm generator is meant toare drawn. detect alarm conditions under specified policies and to publish them to the benefit of dedicated alarm handlers. Currently the alarm detection is threshold based and alarms are communicated visually *SOFIA (2009-11) is funded through the European JTI Artemis programmeunder the subprogramme SP3 “Smart Environments and Scalable DigitalServices”, see http://www.sofia-project.eu
Smart Space Access Protocol (SSAP): Physical distribution of a Smart Space Smart Space SIB KP KP Knowledge Processor (KP): SIB SIB Semantic Information Broker (SIB) Figure 4. OIP logical architecture will take a further look at the SIB and its interacting entity, i.e. the KP. The SIB acts as the shared information store for the OIP. It utilizes the Resource Description Framework (RDF), a triple based Semantic Web standard for expressing complex data as directed labelled graphs in combination with an ontology. An ontology contains all the definitions of the entities used within the SS and their properties which are also used to relate the entities with one another. The SIB provides an interface whose Figure 3. Our scenario fundamental components are: join, leave, insert, remove, query Our claim is that along with the shown approach, valuable and subscribe. The protocol used to communicate with the SIB isapplications could be devised. Thanks to the information level entitled the Smart Space Application Protocol, an applicationinteroperability enabled by the platform (see section 3), the layer protocol based on XML. For a KP to interact with the SIB,applications are agnostic with respect to device vendors and it must first join the SS then it can insert or query for informationtypes, as long as the information is available in the OIP. as needed. The interoperability between KPs is provided when each KP is imbued with the knowledge from the relevant portion III. LOGICAL ARCHITECTURE OF THE OPEN INNOVATION of the application’s domain ontology. PLATFORM IV. APPLICATION DESIGN AND IMPLEMENTATION The OIP being developed within Sofia aims to make Designing SS applications is a two-step process: the entities"information" in the physical world available for smart services involved in the application and the relationships between themin embedded and ubiquitous systems. This implies a shift away must be modelled in an ontology, then the application must befrom the classical focus on interoperability in the physical/service partitioned into distinct KPs. When approaching the design of anlevel towards interoperability at the information level. The ontology, the designer must take into account the informationinformation level can be seen as a set of information producers hierarchy in their application. The information interoperability isand consumers, shared information and its semantic data model. directly impacted by how expressive the ontology is. One distinctThe OIP architecture is simplicity driven and it is made up of advantage of the OIP is that the ontology used can be sub-classedthree distinct entities (Fig. 4): after design time and new information can still be interpreted thanks to the deductive closure calculated by the query engine 1) Smart Space (SS) is a named search extent of . information; When designing our ontology we took an iterative approach: 2) Semantic Information Broker (SIB) is an entity (at we first performed a top-down analysis of our system functional the information level) for storing, sharing, and requirements. After finding the abstract concepts, we took a governing the information of one SS; bottom-up view of the system to understand what kinds of sensor information our hardware would be providing. The resulting 3) Knowledge Processors (KP) is an entity interacting ontology is shown in Fig. 5. Our main entities, i.e. classes, are with the SIB and contributing and/or consuming Person, Environment, Alarm, Device, and Data. The Person and content according to a relevant ontology. Environment entities are self-explanatory. Alarms are entities characterized by an AlarmType, e.g. HeartRateAlarm, and in this case are related to Environments or to Persons. Devices are Using this viewpoint, a SS can be made up of a number of objects that can produce data or run KPs and are described bySIBs which provide information interoperability and at least two their characteristics, e.g. resolution, communication channels,KPs, one a producer and the other a consumer. In order to better MAC addresses. Data read by a sensor is represented by a literalunderstand the application presented in Section 2, the following value, i.e. the reading, a timestamp, the type of measurement, e.g. heart rate, temperature, humidity, the unit of measure and its privacy level. By modelling the data class in this way, we
ensured that any KP consuming sensor data would be able to takeadvantage of new sensor types without having to rethink the KP.Once our application’s ontology had been defined, the KPs wereidentified and modelled as shown in Fig. 6. The followingsubsections describe the functionality of the separate KPs in thesystem.A. Environmental Sensing KP The environmental sensing KP publishes environmentalsensor data to the system. It has a simple user interface to registerthe environment, the sensor platform and to associate them.When humidity and temperature data are inserted into the SIBthey are associated with the room that is being monitored. In atypical SS application the registration of theenvironment and platforms would be performed separately,however, for simplicities sake the two have been integrated.B. Physiological Sensors KP Figure 6. KP-SIB interaction This runs on a Windows Mobile smartphone and has a GUIwhich is used to register the device, its data and alarm thresholds.After configuration, data is captured from a Zephyr BioHarness:heart rate, skin temperature, respiration rate, posture angle and an D. Location KPactivity index are associated with a user and the information is This KP interfaces with our RFID readers so that when ainserted into the OIP. person enters a room, they are associated with the new location. This KP also updates the location of any devices the person mayC. Thom Index KP have in their possession. This KP calculates the thermohygrometric index for everyenvironment in the system. As discussed in section 2, this index - E. Alarm Generator and Announcer KPalso called Thom Index - is a derivation of the humidity and Performs a search for all entities that have an associatedtemperature in the room and adjusts it to a new temperature that safety threshold and subscribes to the relevant data. When theis more akin to that of what a person “feels”. The KP subscribes data falls outside of the threshold, an alarm is raised and placedto any temperature data, performs a check on the unit of measure, in the SIB. This way any other KP wanting to perform someand inserts the new Thom index data. The information is action is capable of doing so. Furthermore, in this particularassociated with a given environment based on the location of the instance, the same KP visualizes the alarm (Fig. 6).sensors. F. Health Care Monitoring KP Allows the health care service to monitor a patient in real time. The KP allows the viewer to select from all the people available and then creates a subscription to all of their relevant data. It uses all of the data on the SIB, so not just physiological data is available, but also the person’s environmental information as well. The KP visualizes instantaneous heart rate, skin temperature, respiration rate, posture angle and an activity index in addition to the user’s location, its Thom Index and environmental data (temperature and humidity). Fig. 7 shows the UI of this KP. The user’s information is also logged to a file. Fig. 8 shows some data collected from one such session; during this example the patient is in a home environment walking between a cellar and a living room. The seemingly low skin temperature is due to the Bioharness infrared temperature sensor position on the user’s chest. This KP runs unchanged on a Ubuntu laptop and on a Maemo tablet (Nokia N810). Figure 5. Ontology class tree
offered by the OIP. Multiple KPs are run on heterogeneous devices, while each one takes advantage of the shared information. This type of system can be extended by new producer KPs as long as the new information produced follows the ontology used in the original system. As long as this condition is met, a consumer KP can take advantage of the new information with little or no modification of its codebase. This feature ensures the flexibility of the system to future expansion or modification of the sensor platforms used. We are continuing the development of this application by integrating it into a larger system that will offer us the ability to effect the user’s environment given their alarms and current state. We also hope to continue this work by analyzing the medical benefits of having a more holistic view of a patient by performing field tests with biomedical engineers and clinicians. The continued development of the OIP will lead to a number of varying new services and applications. As our everyday devices evolve towards the vision of the Internet of Things, data interoperability at the information level will become ever more important. Figure 7. Screenshot of health care monitoring KP ACKNOWLEDGMENT This work was funded by the European Commission, within the framework of the ARTEMIS JU SP3 SOFIA project (http://sofia-project.org/) . The authors would like to thank all project partners who contributed to the definition and implementation of SOFIA Open Innovation Platform. Section 3 and Fig. 3 are largely based on SOFIA work-in-progress. The iMote2 devices used in the research were donated to the University of Bologna by Intel Lab during a previous research in 2006. NOKIA is a trademark of Nokia® Corporation. Intel and iMote2 are trademarks of Intel® Corporation. Other names and brands may be claimed as the property of their respective owners. REFERENCES  Zephyr, 1 Sept. 2009 <http://www.zephyr-technology.com/>  Telcomed Advanced Industries Ltd., 1 Sept. 2009 <http://www.telcomed.ie/>  Alive Technologies , 1 Sept. 2009 <http://www.alivetec.com/>  F. Spadini, F. Vergari, L. Nachman, C. Lamberti, T.S. Cinotti, "A wireless and context-aware ECG monitor : an iMote2 based portable system", Computers in Cardiology 2008, 14-17 Sept. 2008 pp. 997-1000  Elena Meli “Più sereni col controllo telematico” Interview to Prof. Figure 8. Health care monitoring session: Thom index, respiration rate, Massimo Santini, Head of Cardiology Division of San Filippo Neri heart rate and skin temperature Hospital in Rome, Corriere della Sera, Sept. 6, 2009  S. Wagner, "Towards an open and easily extendible home care system infrastructure", Pervasive Computing Technologies for Healthcare, 2008. PervasiveHealth 2008. Second International Conference on, Jan. 30 2008- Feb. 1 2008 pp. 42-45 V. CONCLUSIONS  Siemens-Desigo. Siemens AG, 1 Sept. 2009 In this paper we demonstrated an application where <http://www.buildingtechnologies.siemens.com/bt/global/en/products_systinformation consumers and producers are decoupled and relevant ems/building_comfort_hvac/home_and_building_automation/desigo/Pages /desigo_home.aspx>information is stored on the shared information search domain
 N. Noury, G. Virone, P. Barralon, J. Ye, V. Rialle, J. Demongeot, "New 2008, edited by David Arnold, Franco Niccolucci, Daniel Pletinckx, Luc trends in health smart homes" Enterprise Networking and Computing in Van Gool Healthcare Industry, 2003. Healthcom 2003. Proceedings. 5th International  S. Zauli Sajani, G. Garaffoni, C. A. Goldoni, A. Ranzi, S. Tibaldi, P. Workshop on , 6-7 June 2003 pp. 118-127 Lauriola, "Mortality and bioclimatic discomfort in Emilia-Romagna, Italy" I. Roussaki, I. Papaioannou, D. Tsesmetzis, J. Kantorovitch, J. Kalaoja, and J Epidemiol Community Health - 2002, vol 56 pp 536–537 R. Poortinga, "Ontology Based Service Modelling for Composability in  NanoUHF RFID Reader, 1 Sept. 2009 <http://www.tagsense.com/> Smart Home Environments", Constructing Ambient Intelligence, 2008, vol. 11 pp. 411-420.  UHF RFID TAGS, 1 Sept. 2009 <http://www.power-id.com/>  Ora Lassila, "Taking the RDF model Theory out for a spin" In Ian N. Ryan, P. Mohr, D. Manzaroli, G. Mantovani, S. Bartolini , A. D’Elia, Horrocks and James Hendler, editors, Proceedings of the First International M. Pettinari, L. Roffia, L. Sklenar, F. Garzotto, T. Salmon Cinotti (2008) Semantic Web Conference, ISWC 2002, Sardinia, Italy, number 2342 in "Interoperable multimedia mobile services in cultural heritage site" in Lecture Notes in Computer Science, pp. 307–317. Springer-Verlag, EPOCH Conference on Open Digital Cultural Heritage Systems, Rome, Heidelberg, Germany, June 9 – 12, 2002.