QoC-aware meta-model to identify and qualify situations on the internet of things environment

Abstract

.In the internet of things world, sensors continually generate contextual information that is the core stone for understanding and adapting to the environment according to the user situation. However, diversity of contextual data sources, its imperfection nature and the complexity to identify and model the pertinent situation make it more challenging. Consequently, using a technique like the quality of context to ensure the system quality requirement and having a generic model of context and situation is crucial. Recently, many domain-specific models and few generic meta-models have been proposed to cope with these concerns, but rarely are those that handle context, situation, and quality of context (QoC). In this paper, we introduce a general and extensible QoC-aware meta-model describing both situation and context with their relevant quality of context. Furthermore, we propose a process that a developer or a designer should follow to model and infer situations with adequate quality level. We dementated our model with a prototype application for medical alert.

INTRODUCTION

The internet of things has achieved enormous success in the last decades due to the large and rapid development of technologies like mobile phones, and sensors but still, the wiser vision where every device is completely embedded [ 1 ] isn’t completely accomplished. This vision faces different challenges. One of those core challenges is situation and context awareness that is the ability of a system to sense and adapt to the environment surrounding it. To identify a situation, researchers have used multiple techniques and methods like specification-based and learning-based techniques [ 2 ]. Furthermore, researchers also presented meta-models and tools to specify a situation for a certain application through a Model-Driven technique that gives the opportunity to model a situation in a generic way. Most researchers expect the contextual data to be correct and ignore the aspect of the imperfection. Moreover, a technique like the quality of context must be used to handle and ensure an adequate quality level of the contextual data. In this paper, we introduce a generic and extensible QoC-aware situation metamodel helping designers and developer to identify and qualify situations on the internet of things environment. We also propose a process that assists developers to create their own model and describe situations with adequate quality level.

RELATED WORK

Quality of Context

The contextual information is characterized by their imperfection nature which can be imprecise, ambiguous and erroneous [ 3 ]. Accordingly, Quality of Context (QoC) is essential to ensure the worthiness of the collected contextual data. QoC stand for every information that outlines the quality level of context information and it's represented as a set of parameters like accuracy and completeness. QoC was firstly introduced by [ 4 ] as “any information that describes the quality of information that is used as context information”. Researchers in [ 5 ] claimed that this definition ignores the subjective view of the concept and doesn't involve the consumer satisfaction and define it as “Quality of context indicates the degree of conformity of the context collected by sensors to the prevailing situation in the environment and the requirements of a particular context consumer”. The definition of [ 4 ] is considered as the accepted definition of this article. Many works have proposed their own vision of QoC and their own parameters of measuring the quality and modeling QoC. [ 6 ] presented a list of quality parameters (precision, probability of correctness, trust-worthiness, resolution and up-to-dateness). [ 7 ] investigated the quality of context parameters and compared them with the quality concepts proposed by ISO. The age and precision where the adopted parameters. [ 8 ] contributed in solving conflict during the life cycle of context by choosing the contextual object with the highest quality based one four parameters. Authors in [ 9 ] proposed the most completed list of quality of context parameters with mathematical calculation formula for each parameter. In the part of modeling QoC, [ 10 ] proposed QoCIM, an UML meta-model that allows developers to define and create any quality parameter. Recently, [ 11 ] developed a three-layer framework where every layer has its QoC parameters and the context situation layer has credibility as new QoC proposed parameter. 2.2

Generic Models

Researchers in [ 12 ] has presented the first model-driven development of context-aware services based on an UML meta-model .the contextUML model represent context as atomic and composite and context sources as service community , after that [ 13 ] proposed a model driven architecture to create a context-aware service based on a context meta-model. The model was expended with OCL rules to avoid invalid instances dentition and to specify the context situation. [ 14 ] designed a generic and extensible modeldriven process for creating a context and quality-aware application. The proposed process separates the context-aware designer that express the application needs conform to a meta-model CA3M completed with QoC from the context manager which handle the implementation of those needs. [ 15 ] proposed a domain-specific language based on an UML entity-based model for context modeling which provides a high level of abstraction without including implementation detail to generate software artifact. [ 16 ] introduced MLcontext, a language with a quality of information meta-model and a contextual situation meta-model that provide the composition, the parameterization of a situation and the quality level to generate artifact code for context-aware applications. [ 17 ] presented a model-based approach for context aware-application that capture dynamically the context change. The proposed approach is used by programmers to develop a context-aware application. 2.3

Domain Dependent Models

CARA [ 18 ] presented a contextual medical model composed of case-based context model representing the user, his physiology, area and the objects in interaction. This last is used to generate a high-level context for a fuzzy model to deduct the medical situation throw KNN, case-based and fuzzy sets reasoning. [ 19 ] contributed with fuzzy ontology model for situation and context in a u-learning domain where they described situations as a set of specific context taxonomy for learners with fuzzy attributes that been used with fuzzy rules and evidence theory. In activities of daily living, [ 20 ] has used an ontological model for situations when those last are activities described by their dependencies as set of observations presented as RDF graphs and extracted with SPARQL CONSTRUCT graph patterns. [ 21 ] has suggested a context model constructed from three ontologies: a user ontology , a physical environment , and a proactive ontology which contains the events and the appropriate action to manipulate a situation.

According to Table 1, we concluded that few are models that take into consideration QoC. Although the generic models proposed gives us the possibility to use it under all domains it's restricted to logic rules-based reasoning which is often not enough. The domain-specific models have proposed multiple techniques that enforce the learning. Facing this situation, we propose a generic model that consider context, situation, and quality of context and give the developer the freedom of using the appropriate reasoning method depending on his environment and his needs. 3

QOC-AWARE SITUATION META-MODEL

CONSTRUCTION

To create our proposed meta-model, we conducted considerable studies about the proposal context taxonomy for generic and domain dependent meta-models ..We also addressed the problem of context source and the quality of context to create our metamodel which is a combination of three meta-models: 1) the context taxonomy metamodel, 2) the context source meta-model and 3) the QoC meta-model. 3.1

Context Taxonomy Meta-model

Many context taxonomies have been suggested for the domain dependent models in [ 18-21 ]. However, they couldn’t be used in a generic way. We had analyzed the classification of context represented in the generic meta-models and we founded that the categorization in [ 15 ] was the one that covered all the types of context.

Approaches Quan et al [ 12 ] Achilleos et al, [ 13 ] chabridon et al, [ 14 ] Hoyos et al , [ 15, 16 ] Jaouadi et al, [ 17 ] Yuan et al, [ 18 ] Souabni et al, [ 19 ] Medistsko s and Kompatisi aris, [ 20 ] [Machado et al, [ 21 ]

Model type Generic UML Model Generic UML Model Generic UML Model Generic UML Model Generic UML Model Health model Ulearning ontology model Ambient ontology model Ambient ontology model

OCL Rules OCL Rules Rules Rules Fuzzy Rules, Case-based Reasoning, and KNN Fuzzy Rules and Dempster’s combination rules SPARQL query, OWL2 Rules, Bayesian Networks

Qo c  X   X X X X

X Reasoning

To construct our QoC meta-model we wanted to give a semantic way to describe QoC parameters due to the naming problem of the quality parameters. Since some parameters are more important than others, we added the weight and the level (QoC PWeight,QoC PLevel) to the parameters. Fig. 3. illustrates the QoC meta-model mainly inspired by [ 10 ], where every ContextInformation is either Historic if does change over time or Static otherwise and it’s qualified by QoCParamtre which contains a definition QoCMetricDefinition that could be a simple or a composite one and could be described with further key words or informal definition in the description class. Every QoCMetricDefinition instantiate a QoCMetricValue representing valuation of the QoCParmetre.

Context Source Meta-model

Context information could have multiple sources. For that, a classification of sources is essential to choose the appropriate method to manage this information. The chosen classification is inspired by [ 17 ] and illustrated in Fig 4.

Profiled are the information constructed from the user profile,Derived are the information deducted from other contextual information, UserDefined are the information introduced directly by the user and Sensed are the information gathered from physical sensors .The ContextAssociation have a method of acquisition from the Provider that has a ProviderMethod and every Provider is located in an Entity.

After defining the context taxonomy and its sources with the quality of context metamodel and adding the notion of Situation which is described using a composition of context concerning an Entity we obtained the global QoC-aware situation meta-model (Fig 5).

Our meta-model is characterized by its domain independency and promote universal modeling of context and situation. Furthermore, it integrates the quality of context in a way to ensure well definition of quality parameters and does not consider that the only way to infer. 4

PROCESS OF QOC–AWARE SITUATION

IDENTIFICATION SYSTEM

To build a reactive system, we must be aware of the environment, the quality of the data we are using, and be able to identify the situation the user is currently having it. To deal with those problems, we propose a complete process of identifying situation base on our QoC-aware situation meta-model.

Our building process shown in Fig 6 starts with context acquisition and labeling where the developer chooses the sensors of the system and adopted techniques of acquisition to capture the data. The next phase is the modeling were the developer create his own domain dependent model conform to the QoC-aware situation meta-model to represent the situation with the chosen quality parameters and the context taxonomy and sources. The last phase is the learning and reasoning were the appropriate method of learning is selected according to the type of data and the specificity of the domain to be able to recognize the pertinent situation. 5

Case study: medical alert application

In hospital, every doctor carry a smartphone equipped with WIFI ,GPS,3G, and Bluetooth and where he receive important alert and information .In the other side the location and the medical device monitoring the patient health condition are connected to the hospital server. In case of emergency situation, the system should identify the closest specialized doctor relaying on the technologies cited earlier and the quality of context parameters like UptoDateness and accurency to choose the most reliable location and provide him with the case and shortest path to the patient. Fig 8 present an excerpt of medical alert model created from QoC and situation meta-model by defining the situation, entities, context taxonomy and type and the QoC parameter required for medical alert application. For the reasoning and due to the simplicity of the example rule based technique is enough. We use OCL rules.This OCL rules illustrate the logical expression determinate the occurrence of the situation in particular contextual condition. To define a cardiology emergency situation a patient must have a pulse upper then 120 beat per minute Context s: Situation s.name=’emergency situation’-> drive inv: if any(e:entity|e.name=’Julia’).EntityContext->(h:physical| |name=pulse| pulse>120)then true else false endif. An addition we need to identify the nearest cardiology doctor relaying on the QoC parameter Context si:Situation si.name=’emergency alert’->drive inv: if any (e.entity| e.name=’sami’).EntityContext>(e:envirement|e.name=’cardiology’)and (en:envirement |en.name=’location’) .From> (sp1:sensed provider |sp.name=’wifi’)or(sp2:sensed provider |sp2.name=’3G cell id’)or( sp3:sensed provider |sp3.name=’bluetooth’)or( sp4:sensed provider |sp.name=’GPS’)anden.QualifiedBy(p1:QoCParameter|p1.name=’UptoDateness’).has (ql=QoCLevel|ql.value=’0.8’) and(p1:QoCParameter |p1.name=’Accuracy ’).has(ql=QoCLevel|ql.value=’0.8’) then true else false endif.

CONCLUSION AND PERSPECTIVES

In this paper, we have presented a generic and extensible QoC-aware situation metamodel helping designers and developer to identify and qualify situations on the internet of things environment. The proposed meta-model is constructed by merging three metamodels: 1) context taxonomy meta-model, 2) context sources meta-model, and 3) quality of context meta-model. We have also described our process of identifying the pertinent situation for a specific domain based on our meta-model and the labeled contextual information and the learning and the reasoning methods depending on the chosen environment. In addition, we have presented a medical alert use case to show the feasibility of our approach. For the reasoning part we have proposed a set of OCL rules due to the simplicity of the example.

Although, this work can be continued is several ways. As ongoing work, we plane to evaluate the effectiveness and usability of our QoC-aware situation meta-model for smart health environment monitoring where we will a more powerful reasoning technique like Bayesian networks, Hidden Markov models, Decision trees, etc. 7

References

1. Al-Shargabi , A. A. , Siewe , F. and Zahary , A. ( 2017 ) Quality of Context in Context-Aware Systems . EAI Endorsed Transactions on Context-aware Systems and Applications,

2. Ye , Juan, Dobson, Simon , Et Mckeever , Susan. Situation identification techniques in pervasive computing: A review . Pervasive and mobile computing , 2012 , vol. 8 , no 1, p. 36 - 66 .

3. Häussermann , Kai, Hubig, Christoph, Levi, Paul, Et Al. Understanding and designing situation-aware mobile and ubiquitous computing systems . In : Proc. of intern. Conf. on Mobile, Ubiquitous and Pervasive Computing . 2010 . p. 329 - 339 .

4. Henricksen , Karen Et Indulska, Jadwiga. Modelling and using imperfect context information . In : Pervasive Computing and Communications Workshops , 2004 . p. 33 - 37 .

5. Buchholz , Kupper , et Schiffers. Quality of context: What it is and why we need it . In Proceedings of the 10th Int. Workshop of the HP Open View University Association, 2003 .

6. Manzoor , Atif, Truong, Hong-Linh , Et Dustdar, Schahram. Quality of context: models and applications for context-aware systems in pervasive environments . The Knowledge Engineering Review , 2014 , vol. 29 , no 2, p. 154 - 170

7. Neisse, ricardo, wegdam, maarten, et van sinderen, marten. Trustworthiness and quality of context information . In : Young Computer Scientists , 2008 . ICYCS 2008 . The 9th International Conference for . IEEE, 2008 . p. 1925 - 1931

8. Manzoor , Atif, Truong, Hong-Linh , Et Dustdar, Schahram. On the evaluation of quality of context . Smart Sensing and Context , 2008 , p. 140 - 153 .

9. Manzoor , Atif, Truong, Hong-Linh , Et Dustdar, Schahram. Quality of context: models and applications for context-aware systems in pervasive environments . The Knowledge Engineering Review , 2014 , vol. 29 , no 2, p. 154 - 170 .

10. Marie , Pierrick, Desprats, Thierry, Chabridon, Sophie , Et Al. QoCIM: a meta-model for quality of context . In : International and Interdisciplinary Conference on Modeling and Using Context . Springer, Berlin, Heidelberg, 2013 . p. 302 - 315 .

11. Al-Shargabi , Asma Abdul Ghani. A multi-layer framework for quality of context in ubiquitous context-aware systems . International Journal of Pervasive Computing and Communications , 2018 .

12. Sheng , quan z. Et Benatallah, boualem. ContextUML: a UML-based modeling language for model-driven development of context-aware web services . In : Mobile Business , 2005 . ICMB 2005. International Conference on. IEEE, 2005 . p. 206 - 212 .

13. Achilleos , Achilleas, Yang, Kun , Et Georgalas, Nektarios. Context modelling and a context-aware framework for pervasive service creation: A model-driven approach . Pervasive and Mobile Computing , 2010 , vol. 6 , no 2, p. 281 - 296 .

14. Chabridon, s., Conan, d., Abid , Building ubiquitous QoC-aware applications through model-driven software engineering . Science of Computer Programming , 2013 , 78 ( 10 ), p. 1912 - 1929 .

15. Hoyos , josé r., García-molina, Jesús , et Botía, juan a. A domain-specific language for context modeling in context-aware systems . Journal of Systems and Software , 2013 , vol. 86 , no 11, p. 2890 - 2905 .

16. Hoyos , jose r., garcía -molina, jesus, botía, juan a ., et al. A model-driven approach for quality of context in pervasive systems . Computers &Electrical Engineering , 2016 , vol. 55 , p. 39 - 58 . 2010 ).

17. Jaouadi , imen, Djemaa, Raoudha ben, et Ben-abdallah, hanene. A model-driven development approach for context-aware systems . Software & Systems Modeling , 2016 , p. 1 - 27 .

18. YUAN , B. et HERBERT , J. Context-aware hybrid reasoning framework for pervasive healthcare . Personal and ubiquitous computing , 2014 , vol. 18 , no 4, p. 865 - 881 .

19. R. Souabni , I. Bayoudh Saadi , N. Ben-Salah , M. C. Kinshuk and H. Ben-Ghezala , Approach Based on fuzzy ontology for situation identification in situation-aware ubiquitouslearning environment , in IEEE Int. Conf. Fuzzy Systems , 2016 ,pp. 1805 { 1812 .

20. Meditskos , georgios et kompatsiaris, Ioannis. Iknow: Ontology-driven situational awareness for the recognition of activities of daily living . Pervasive and Mobile Computing , 2017 , vol. 40 , p. 17 - 41

21. Machado , Alencar, Maran, vinicius, augustin, iara, et al. Reactive, proactive, and extensible situation-awareness in ambient assisted living . Expert Systems with Applications , 2017 , vol. 76 , p. 21 - 35 .