Constructing the Ubiquitous Intelligence Model based on Frame and High-Level Petri Nets for Elder Healthcare

Transcription

1 Constructing the Ubiquitous Intelligence Model based on Frame and High-Level Petri Nets for Elder Healthcare Jui-Feng Weng, *Shian-Shyong Tseng and Nam-Kek Si Abstract--In general, the design of ubiquitous intelligence application is difficult because of several technical issues. Most of ubiquitous intelligence applications have implicit knowledge but it is hard to represent. Moreover, the dynamic nature of ubiquitous computing environment makes the acquisition of ubiquitous intelligence knowledge hard and limits the system extensibility. In this paper, we proposed the Frame-based Ubiquitous Intelligence Knowledge Model (FUIKM) which consists of two layers. The Ubiquitous Object Frame (UOF) knowledge layer is modeled by the frame representation for better extensibility. The Context-aware Service Process (CSP) knowledge layer is modeled by the High-Level Petri Nets for explicitly context-aware knowledge. To show the expressive power of FUIKM, several scenarios of elder healthcare are proposed and modeled. The FUIKM is applied to model these scenarios. Index Terms-- elder healthcare, High-Level Petri Nets, ubiquitous intelligence, frame. A I. INTRODUCTION s mobile computing and Internet technology become more and more popular in recent years, we come to the era of so-called global information world. From the Weisor s vision [8], he described that a variety of computing and communication capability computers will be ubiquitous around our environment and the associated technology is called the ubiquitous computing. Ubiquitous computing is a new computing paradigm in which computing systems are seamlessly integrated into the users everyday life and provide pervasive services. In more concrete description, the computing and communicating capabilities (such as sensors, robots or RFID tags) can be embedded or attached to ordinary things and let them behave smartly to provide intelligent services. This feature of perception and reasoning mechanism for the user centric service is called the ubiquitous intelligence [5]. Many researches believe that context-awareness is an This work was partially supported by National Science Council of the Republic of China under contracts NSC S MY3, NSC S MY3 Jui-Feng Weng is with Department of Computer Science, National Chiao Tung University, ROC ( roy@cis.nctu.edu.tw). Shian-Shyong Tseng is with Department of Computer Science, National Chiao Tung University, ROC and Department of Information Science and Applications, Asia University, ROC ( sstseng@cis.nctu.edu.tw). Nam-Kek Si is with Department of Computer Science, National Chiao Tung University, ROC ( edwardsi@cs.nctu.edu.tw). important aspect of the ubiquitous intelligence. The context-awareness means that the things of the environment are capable of perceiving user s situation based on the context information of environment and the context-aware system can provide appropriate services. Since the context information may be ubiquitous around us, the dynamic nature of environment makes the design and implementation of ubiquitous intelligence application more difficult. Many researches such as sensor network, ambient intelligence, context awareness middleware and smart space aim to be aware of the smart resources and manage the context information in more abstract way. Although the proposed architectures or applications can assist the management of the distributed smart objects, they still mentioned that the personal behavior cognition is very difficult. It other words, only naive services instead of the smart services can be provided. Although other context-aware systems may be provided for some specific scenario, the knowledge of ubiquitous intelligence is usually implicit and embedded in the system such that it is still hard to manage or reconfigure the knowledge. Therefore, to assist the design of the context-aware ubiquitous intelligence application raises several technical challenges: How do we explicitly represent the implicit ubiquitous intelligence knowledge? How do we manage the dynamic context information of the environment and still keep the system s extensibility? How do we acquire the knowledge of ubiquitous intelligence in systematical way? In this paper, we believed that the acquisition of ubiquitous intelligence is difficult because it lacks a standardized and extensible knowledge representation. With our observation, the knowledge of the ubiquitous intelligence can be divided into low level ubiquitous objects handling and high level context-awareness reasoning for cognition of user s situation. Therefore, based upon the concept of knowledge base approach, we proposed a Frame-based Ubiquitous Intelligence Knowledge Model (FUIKM) consisting of two layers. The lower one, Ubiquitous Object Frame (UOF) knowledge layer, is modeled by the frame representation. The upper one, Context-aware Service Process (CSP) knowledge layer, is modeled by the High-Level Petri Nets. Based on FUIKM, a process of designing a ubiquitous intelligence application, FUIKM Design Process (FUIKM DP) is also proposed help using FUIKM. Finally, to show the expressive power of FUIKM, several scenarios of elder healthcare are proposed and modeled by FUIKM. The experimental results show that the FUIKM is good enough to represent healthcare scenarios. 174

2 II. RELATED WORKS There are several researches about ubiquitous computing applications. In research [3], the healthcare system is proposed to prevent patients from getting lost by providing four monitoring schemes consisting of indoor residence monitoring, outdoor activity area monitoring, emergency rescue scheme, and remote monitoring mode. The technologies of RFID, GPS, GSM, and GIS techniques are used to the monitoring system. In research [2], the smart space architecture of project UbicKids for ubiquitous kids care was proposed. In [1], a prototype of a mobile context-aware tour guide using user s location portfolio was presented to assist user in visiting scenic spots and writing tour diary. Dey [2] described a conference assistant, a prototype of mobile, context-aware application, which assists conference attendees using user s current location and the conference s agenda in order to provide conference guiding and valuable information gathering. In general, existing ubiquitous computing application developments focused on how to use and manage context information and the usage of embedded device to provide the context-aware service. However, the knowledge of these applications is implicitly implemented in systems and results in lacks of extensibility. In order to support the development of ubiquitous computing application, various architectures [3][6][7][9] focused on the middleware design. Context Toolkit [3] which is context information architecture consists of context widgets, context interpreters and context servers to support the context transformation and resource discovery. Similarly, Contextual Information Service (CIS) [7], HIVE [6] and smart space [5] provided a multi-layer software engineering process to separate context data from application development. Yang [9] proposed a context model with learner ontology and service ontology, and context acquisition mechanism for collecting contextual information at run time. Although these researches proposed the middleware technique to improve the system extensibility, the acquisition of ubiquitous intelligence knowledge is still difficult. In summary, to assist the design of ubiquitous intelligence application raises several technical challenges: How do we explicitly represent the implicit ubiquitous intelligence knowledge? How do we manage the dynamic context information of the environment and still keep the extensibility? How do we acquire the knowledge of ubiquitous intelligence in systematical way? III. THE FRAME-BASED UBIQUITOUS INTELLIGENCE KNOWLEDGE MODEL To support the issues of the explicitness, extensibility and acquisition of the domain knowledge mentioned above, a knowledge representation is proposed to model the ubiquitous intelligence knowledge. A. The ubiquitous intelligence knowledge representation As shown in Fig. 1, since the physical sensors and actuators usually have stereotyped data fields and APIs, we apply the frame knowledge representation to model the functionality of ubiquitous objects, called Ubiquitous Object Frame (UOF). With the model of UOF, the complicated sensors and actuators can be handled in the unified frame representation. Next, we apply the High-Level Petri Nets, which is a powerful language for process flow modeling and validation, to model the context-aware event reasoning flow, called Context-aware Service Process (CSP). The CSPs can be used as operation sequence building blocks. Next, the modeled sequencing building blocks can be further cooperatively combined to construct the given scenarios of ubiquitous intelligence application. Fig. 1. The idea of ubiquitous intelligence knowledge representation. B. The definition of Frame-based Ubiquitous Intelligence Knowledge Model (FUIKM) Based on the concept of knowledge base approach, we proposed a two-layer Frame-based Ubiquitous Intelligence Knowledge Model (FUIKM) to model the knowledge of ubiquitous intelligence applications. The lower layer is Ubiquitous Object Frame (UOF) and the upper one is Context-aware Service Process (CSP). Definition 1. The Ubiquitous Object Frame (UOF) is a 4-tuple: UOF = (FN, Rel, P, S), where 1. FN is the name of a frame. It can be a service name or context name. 2. Rel = {rel 1, rel 2,,rel h } and rel k = <relation, FN> which is the relation with other frame specified by frame name FN. There are two types of relation where a kind of denotes the inheritance relation and is a denotes instantiation relation. 3. P = {p 1, p 2,, p m } is a finite set of procedures which can be triggered by some event of a frame s procedural attachment. 4. S = {s 1, s 2,..., s n }is a finite set of slots, and s i = <SN i, V i, R i >, where SN i is the name of the i-th slot. 175

3 value value value value V i is the value of the i-th slot. R i ={<r 1, p i >, <r 2, p j >,, <r n, p k >} is a set of guarding rules r i which trigger the procedural attachments p j based on different events. There are four types of guarding rules in frame which are if needed, if added, if changed and if removed. The procedural call of a guarding rule is referred to one of P where the input parameter can be slot value or a variable holder. Fig. 2. The diagram of UOF variables in G(t) must belong to Σ. The firing rule G(t) of a transition (t T) is defined as if-else form. The guard function can generate specific conditional behaviors. There are three guard functions: G(T C ): specify whether the context information is ready to deliver. G(T S ): specify whether the prerequisite parameter is ready to apply the process. G(T A ): specify whether the prerequisite data is ready to apply the sub-csp process. 7. E: A Expression is an arc expression function. It maps arc into an expression which must have the type of its places color. It denotes the information that how many and which kinds of token colors should be removed from the input places or added to the output places. 8. I init : P Initialization Expression is an initialization function. The initialization expression is defined as Color=default slot value of frame. The context-aware service means that the system is capable of referring the context events for specific purpose and triggering the service based on the context information. In other words, the reasoning process is actually the context events processing flow among UOFs. The UOF can model the context information as context frame and it can model the parameters of service API as service frame. Therefore, we apply High-Level Petri Nets (HLPN) to model the Context-aware Service Process (CSP) cooperated with UOFs. The definition of CSP is as follows. Definition 2. The HLPN of Context-aware Service Process (CSP) is a 8-tuple: CSP = (P, T, Σ, C, A, G, E, I init ), where 1. P = {p 1, p 2,, p m } is a finite set of places. P includes two types of places: P C denotes the data obtained from the context frame that need to pass to other frames. P S denotes the parameters obtained from the service frame that activates an action. 2. T = {t 1, t 2,, t n } is a finite set of transitions (P T=0). T includes three types of transitions: T C represents the context frame actor, which performs a designated procedure call of the associated frame to deliver the context information from one place to another. T S represents the service frame which submits the event parameters to trigger the service procedure. T A represents the sub-csp component. 3. Σ = <SN i : v i > is the non-empty finite color sets of tokens. Each token represents the specified slot value of the associated frame. 4. C is color function, where C: P Σ. 5. A (P T) (T P) is the union of Get Value arcs and Set Value arcs. PT is the arc from a place to a transition which gets value from frame slot. TP is the arc from a transition to a place which sets value to frame slot. 6. G: T Boolean expression is a guard function. All Fig. 3. The diagram of CSP and its relation with UOF. IV. THE FUIKM DESIGN PROCESS With the FUIKM model defined above, we can apply the FUIKM design process to construct the ubiquitous intelligence application by knowledge base approach. As shown in Fig. 4, there are three steps in the FUIKM Design Process. value value value Fig. 4. The FUIKM Design Process In the first step, new frame creation, the ubiquitous objects such as sensors and service systems can be represented by UOF. The second step, context-aware operation sequencing, the 176

4 context-awareness reasoning process of UOF can be represented by CSP. In the third step, with the modeled UOFs and CSPs, the ubiquitous intelligence application can be easily constructed to fit the scenarios we need. A. New frame creation With the UOF model, the ubiquitous objects can be managed as the ubiquitous object ontology as shown in Fig. 5. The ubiquitous object are classified into Context frame which is for context information gathering and Service frame which is for service providing. Tab. 1. Context-awareness in CSP representation Context Awareness in CSP Representation Context Reference Context Aggregation Description Refer the context P C to trigger the service P S Aggregate context source P C1,P C2, P Cn to P Cx State Transition The context status transits between state P C1 and P C2 Device Selector Refer the context P C1 to select a service from P S1 to P Sn Fig. 5. Ubiquitous objects ontology As shown in Fig 6, to model the context information with UOF, we can set the context data fields as frame slots and context information process as attached rule and procedure calls. To model the service system with UOF, we can set the API parameters as frame slots and attach the rule with the service procedure call. Adaptive Configuration Refer the context P C1 to trigger the service P S with an adaptive configuration Fig. 6. UOF examples B. Context-aware operation sequencing The context-awareness is the main capability of ubiquitous intelligence. Therefore we apply the CSP to model several context-awareness components. V. THE FUIKM MODELING FOR HEALTHCARE SCENARIOS In order to show how do we model the knowledge of ubiquitous intelligence applications, several elder healthcare service scenarios are proposed. Suppose the elder Bob lives alone in his house and he needs the healthcare service. A. Fall prevention The elders often fall down in the house because of their presbyopia. When Bob is walking and approaching the stairs, the fall prevention service detects his position. Next, it light up his position to prevent Bob from falling. 177

5 Fig. 7. The FUIKM diagram for scenario fall prevention B. Adaptive treadmill Bob needs to exercise everyday. When he is jogging on the treadmill, the adaptive service reconfigures the speed and period of treadmill to adapt his healthy status. Fig. 9. The FUIKM diagram for scenario task monitoring service VI. DISCUSSION After the experiments of three scenarios modeling, the complex ubiquitous intelligence application can be explicitly constructed by the proposed five CSP components. The UOF can be easily created to represent the detail context information and service process. This experimental result shows that the FUIKM has enough expressive power and extensibility to explicitly represent the ubiquitous intelligence knowledge. Fig. 8. The FUIKM diagram for scenario adaptive treadmill VII. CONCLUSION In this paper, we proposed the knowledge base approach to model the ubiquitous intelligence applications. The FUIKM model with UMF layer and CSP layer are defined as the knowledge representation model. Five basic context-awareness components are modeled by CSP. With the FUIKM, we also proposed the FUIKM design process to assist the application design. Three scenarios are modeled by FUIKM and it shows that the FUIKM has better maintainability and extensibility. In the near future, the authoring tool based to FUIKM will be implemented and experimented. In addition, more ubiquitous intelligence applications for elder healthcare will be investigated for improving the elder s life. C. Task monitoring service When Bob is cooking with hot plate, if he leaves the kitchen without switching off the fire, then the warning service rings Bob s phone to remind him of the hot plate. ACKNOWLEDGMENT This work was partially supported by National Science Council of the Republic of China under contracts NSC S MY3, NSC S MY3 178 REFERENCES [1] G. D. Abowd, et al., "CyberGuide: A Mobile Context-Aware Tour Guide". ACM Wireless Networks, [2] A. K. Dey, et al., "The Conference Assistant: Combining context-awareness with wearable computing". Proceeding of the 3rd International Symposium on Wearable Computers, 1999

6 [3] A. K. Dey, D. Salber, and G.D. Abowd, A Context-Based Infrastructure for Smart Environments, Proceedings of the 1st International Workshop on Managing Interaction in Smart Environments, 1999 [4] C.C Lin Chung-Chih ; Chiu Ming-Jang ; Hsiao Chun-Chieh ; Lee Ren-Guey ; Tsai Yuh-Show ; A Wireless Healthcare Service System for Elderly with Dementia, IEEE Transactions on Information Technology in Biomedicine : Accepted for future publication, Volume PP, Issue 99, 2006 Page(s):1 1 [5] J. Ma, L. T. Yang, B. O. Apduhan, R. Huang, L. Barolli and M. Takizawa, Towards a Smart World and Ubiquitous Intelligence: A Walkthrough from Smart Things to Smart Hyperspaces and UbicKids, International Journal of Pervasive Comp. and Comm., 1(1), March [6] M. Nelson, et al., Hive: Distributed agents for networking things. IEEE Concurrency, vol. 8, no. 2, pp , 2000 [7] J. Pascoe, Adding generic Contextual Capabilities to Wearable Computers, Proceedings of the 2nd International Symposium on Wearable Computers, 1998 [8] M. Weiser, The Computer for the Twenty-First Century. Scientific American, Scientific American, 1991 [9] S. J. H. Yang, Context Aware Ubiquitous Learning Environments for Peer-to-Peer Collaborative Learning, Educational Technology & Society, vol. 9, no. 1, pp ,