Context management for ubiquitous applications Denis Conan, Chantal - - PDF document

Context management for ubiquitous applications

Denis Conan, Chantal Taconet, Sophie Chabridon

Ubimob 2010

Institut Télécom; Télécom SudParis; CNRS UMR SAMOVAR; équipe ACMES; projet MARGE juin 2010

Context management for ubiquitous applications

Contents

Context management for ubiquitous applications Denis Conan, Chantal Taconet, Sophie Chabridon, , Institut Télécom; Télécom Sud- Paris; CNRS UMR SAMOVAR; équipe ACMES; projet MARGE, Ubimob 2010 juin 2010 1 Acknowledgements 3 Introduction 4 Outline 4 1 Context-awareness and context managment 5 1.1 Some definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.2 Context management requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.3 Functionalities and architecture of a context manager . . . . . . . . . . . . . . . . . . . . . . . 6 2 COSMOS context manager concepts 8 2.1 What the client application accesses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.2 What the context node designer builds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.3 What the context source provider builds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.4 How the context designer composes context nodes . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.5 Context node parameterisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.6 Performance evaluation of COSMOS-ME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.6.1 Hardware and software platforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.6.2 Time for observing and notifying a context node . . . . . . . . . . . . . . . . . . . . . . 12 2.6.3 Time for instantiating a context node . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.6.4 Maximum number of context nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.7 General information on the COSMOS project . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3 Design of context-aware applications with CA3M 15 3.1 CA3M: Meta Model and Middleware for Context Awareness . . . . . . . . . . . . . . . . . . . . 16 3.2 CA application design process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 3.3 CA3M runtime architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 3.4 Context awareness meta-model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 3.5 From observable model to runtime bridge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 4 Quality of context information 19 4.1 Definition and motivations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 4.2 Quality criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 4.3 QoC-aware context operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 4.4 Architecture of QoC-aware context operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 5 E-commerce flash-sale demonstrator 21 5.1 Flash sale cosmos node . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 5.2 Siafu simulator of the shopping center . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 5.3 Flash sale notifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 6 Complementarity of ontology-based and process-oriented context management 24 6.1 Two abstraction levels for classifying situations . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 6.2 Hybrid architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 6.3 Comparison of OCM and PCM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 7 Concluding remarks 27 Related work/Process-oriented context management . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Related work / Distribution of context data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 References 29 Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 2

Context management for ubiquitous applications # 2 ✬ ✫ ✩ ✪

Acknowledgements

Institut Télécom, Télécom SudParis; CNRS UMR SAMOVAR Sébastien Leriche, Amel Bouzeghoub Zied Abid, Léon Lim, Mohammed El-Amine Matougui Cong Kinh Nguyen, Cao Cuong Ngo INRIA ADAM, LIFL: Romain Rouvoy, Lionel Seinturier INRIA MASCOTTE: Judicaël Ribault Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 3

Context management for ubiquitous applications # 3 ✬ ✫ ✩ ✪

Introduction

Results of the cappucino project Mobile commerce domain Context management on mobile and (relatively) constrained devices # 4 ✬ ✫ ✩ ✪

Outline

1 Context-awareness and context managment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7 2 COSMOS context manager concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 3 Design of context-aware applications with CA3M . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 4 Quality of context information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 5 E-commerce flash-sale demonstrator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 6 Complementarity of ontology-based and process-oriented context management . . . . 40 7 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 4

Context management for ubiquitous applications 1 Context-awareness and context managment # 5 ✬ ✫ ✩ ✪

1 Context-awareness and context managment

Need for adaptation to environment changes Control loop of autonomic computing [Kephart and Chess, 2003]

Execution environnement

Causal link Observation Analysis Adaptation policy Platform

Planning Execution Knowledge base

Context management = 1st class concern Identify/detect situations requiring adaptations/reactions # 6 ✬ ✫ ✩ ✪

1.1 Some definitions

Context = Any information that can be used to characterise the situation of an entity [Dey et al., 2001] Entity = An element representing a physical or logical phenomenon (person, concept, etc.) to which “observables” can be associated [Denning et al., 2005] E.g., person, room Observable = An abstraction which defines something to watch over (observe) [Taconet et al., 2009] E.g., location, luminescence An Observable is associated to an entity Observation = The state of an observable at a given time [Taconet et al., 2009] E.g., (48°52′N,2°19′E), 50 lux Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 5

Context management for ubiquitous applications 1 Context-awareness and context managment # 7 ✬ ✫ ✩ ✪ Situation change = relevant state change in the set of observables [Taconet et al., 2009] Situations inject meaning into the application and are more stable, and easier to define and maintain than basic contextual cues [Ye et al., 2009] E.g., enter a place, ((lumin.<100lux) = indoor) A situation change may trigger adaptations # 8 ✬ ✫ ✩ ✪

1.2 Context management requirements

From [Bouzeghoub et al., 2010] (similar list in [MUSIC, 2007]) Observation/Notification: context data both observed by and notified to applications Handling constantly evolving context Situation detection latency: recommendations shall be presented to end-users with appropriate delays according to end-users’ perception For scalability and efficiency purpose: distributed context management, ==> CM on constrained device Observation preparation: when permitted by the freshness requirement, context data requiring resource intensive processing shall be prepared in order to be consumed rapidly later on Handling unforeseen situation: dynamic context data and application-specific knowledge unforeseeable at design time Expression of situations: Application designers shall have a means to express relevant situations to be detected by the system Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 6

Context management for ubiquitous applications 1 Context-awareness and context managment # 9 ✬ ✫ ✩ ✪

1.3 Functionalities and architecture of a context manager

Functionalities Situation detection = “clients” negotiate observation contracts Interpretation = different context processors Collection = different context sources

Context interpretation Adaptation situation detection Context collection

Sensors User profiles System ressources Remote data

processing nodes Hierarchies of Versus Design pattern "Layer"

Context collection Context interpretation Situation detection

Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 7

Context management for ubiquitous applications 2 COSMOS context manager concepts # 10 ✬ ✫ ✩ ✪

2 COSMOS context manager concepts

2.1 What the client application accesses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11 2.4 What the context node designer builds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.4 What the context source provider builds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.4 How the context designer composes context nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.5 Context node parameterisation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15 2.6 Performance evaluation of COSMOS-ME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.0 General information on the COSMOS project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 # 11 ✬ ✫ ✩ ✪

2.1 What the client application accesses

Cosmos Architecture: Compose context frameworks in a component-oriented architecture Situation detection = ⇒ Context policy = Provides the context data to the client applications ◮ High-level context data

Pull Client applications Push

Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 8

Context management for ubiquitous applications 2 COSMOS context manager concepts # 12 ✬ ✫ ✩ ✪

2.2 What the context node designer builds

Context interpretation = ⇒ Context node = Encapsulates a context processor called a context operator ◮ Basic building block of a context policy

Context Operator Pull Push Pull Push Parent context nodes Child context nodes Parameters

# 13 ✬ ✫ ✩ ✪

2.3 What the context source provider builds

Context collection = ⇒ Context collector = Encapsulates an object plugged to a context source ◮ Wrapper of external context source ◮ Leaf context nodes encapsulate context collectors

Context Collector Pull Push Parameters Parent context nodes Context sources

Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 9

Context management for ubiquitous applications 2 COSMOS context manager concepts # 14 ✬ ✫ ✩ ✪

2.4 How the context designer composes context nodes

Flow of context data Context sources Client applications Flow of control when

• bserving

notifying when Flow of control

Context report = Extensible message structure enclosing context data Hierarchical structure: messages, chunks, sub-messages # 15 ✬ ✫ ✩ ✪

2.5 Context node parameterisation

Naming for finding, configuring, and sharing Observations / Notifications Blocking versus non-blocking At most once Active versus Passive

WiFi adjusted bit rate

System call

WiFi link quality detector WiFi Connectivity Average link quality Average bit rate if variable WiFi bit rate variable? Is bit rate WiFi manager

max 1 max 1 Block notification Block observation At most one obs./notif. Active observer Active observer and notifier

• Properties of a context node

– Provide a name to be registered into a registry and searched for for configuration ∗ Attribute nodeName – Control propagation of information ∗ Can observe (down to the leafs) and/or notify (up to the root) ∗ May block the context flow (down or up) or not · Attributes observeThrough and notifyThrough + attributes observeOnlyOnce and notifyOnlyOnce – Control the activity of context nodes Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 10

Context management for ubiquitous applications 2 COSMOS context manager concepts ∗ Is passive or active · Attributes periodObserver and periodNotify · Possibility to design context operators that filter notifications Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 11

2 COSMOS context manager concepts 2.6 Performance evaluation of COSMOS-ME # 16 ✬ ✫ ✩ ✪

2.6 Performance evaluation of COSMOS-ME

2.6.2 Hardware and software platforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.6.2 Time for observing and notifying a context node . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.6.4 Time for instantiating a context node. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20 2.6.4 Maximum number of context nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 # 17 ✬ ✫ ✩ ✪

2.6.1 Hardware and software platforms

Nokia N800 Processor TI OMAP2420, 330 Mhz, 256 Mo of ROM, 128 Mo of RAM, Java 1.5 with CacaoVM (0.99.3), Linux; before tests, free memory ≈ 16 Mo Nokia N95 Processor ARM11-based TI OMAP2420, 330 Mhz, 128 Mo of RAM, native JVM, Symbian S60 3rd HTC TyTN II Processor Qualcomm MSM7200, 400 Mhz, 256 Mo of ROM, 128 Mo of RAM, native JVM, Windows Mobile 6; before tests, free memory ≈ 69 Mo HTC Hero Processor Qualcomm MSM7200A, 528 Mhz, 512 Mo of ROM, 288 Mo of RAM, native JVM, Androïd OS 1.5; before tests, free memory ≈ 144 Mo Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 12

2 COSMOS context manager concepts 2.6 Performance evaluation of COSMOS-ME # 18 ✬ ✫ ✩ ✪

2.6.2 Time for observing and notifying a context node

≈ 40 calls before measurements, measurements = 100 times 10000 calls Type Hardware Average Standard deviation 95% confidence (ms) (ms) interval% Pull Nokia N800 1.68 0.032 [1.684..1.715] blocking Nokia N95 0.78 0.041 [0.759..0.775] HTC TyTN II 0.94 0.013 [0.940; 0.945] HTC Hero 1.90 0.017 [1.899..1.906] Push Nokia N800 0.98 0.003 [0.975; 0.976] blocking Nokia N95 0.62 0.038 [0.615..0.630] HTC TyTN II 0.54 0.005 [0.541..0.543] HTC Hero 1.36 0.012 [1.361..1.365] # 19 ✬ ✫ ✩ ✪

2.6.3 Time for instantiating a context node

≈ 10 instant. before measurements, measurements = 10 times 10 instantiations To avoid side effects, “reboot” the device every 10 measured instantiations Hardware Average Standard deviation 95% confidence (ms) (ms) interval Nokia N800 2662 77 ± 48 Nokia N95 8891 80 ± 50 HTC TyTN II 42 2.1 ± 1.3 HTC Hero 91 1.83 ± 1.13 Specific case of Androïd More time for the first instantiation (ignored in the table) ◮ For HTC Hero, 387 ms Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 13

2 COSMOS context manager concepts 2.6 Performance evaluation of COSMOS-ME # 20 ✬ ✫ ✩ ✪

2.6.4 Maximum number of context nodes

Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 14

Context management for ubiquitous applications # 21 ✬ ✫ ✩ ✪

2.7 General information on the COSMOS project

COSMOS stands for COntext entitieS coMpositiOn and Sharing Main publications: [Conan et al., 2007, Conan et al., 2008, Rouvoy et al., 2008, Taconet et al., 2009, Abid et al., 2009a] Web site: http://picoforge.int-evry.fr/projects/cosmos Ready to run examples For J2SE For J2ME: Android, Windows Mobile, and Symbian OS Labs (discovery, developing, J2ME) Subversion forge: https://picoforge.int-evry.fr, guest/guest Maven repositories: http://picoforge.int-evry.fr/projects/svn/cosmos/maven2-release http://picoforge.int-evry.fr/projects/svn/cosmos/maven2-snapshot http://picoforge.int-evry.fr/projects/svn/cosmos/maven2-site # 22 ✬ ✫ ✩ ✪ Tools: Subversion and Maven Framework dependencies: Fractal, Fractal ADL, Fraclet, Julia, Juliac, and Dream Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 15

Context management for ubiquitous applications 3 Design of context-aware applications with CA3M # 23 ✬ ✫ ✩ ✪

3 Design of context-aware applications with CA3M

3.2 CA3M: Meta Model and Middleware for Context Awareness . . . . . . . . . . . . . . . . . . . . 25 3.2 CA application design process. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25 3.4 CA3M runtime architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 3.4 Context awareness meta-model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .27 3.5 From observable model to runtime bridge. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28 # 24 ✬ ✫ ✩ ✪

3.1 CA3M: Meta Model and Middleware for Context Awareness

[Seyler et al., 2007, Taconet et al., 2009] Model Driven Engineering approach Meta-model, model of context contracts + Generation of context management code Meta-model, model of context sensitivity + Generation of context adaptation Design process of context aware applications Introduction of two new roles ◮ Context specifier ◮ Context-awareness designer CA3M Middleware takes in charge context management aspects According to context contracts defined in context-awareness models Runtime models for runtime reconfiguration of monitoring aspect Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 16

Context management for ubiquitous applications 3 Design of context-aware applications with CA3M # 25 ✬ ✫ ✩ ✪

3.2 CA application design process

Collectors models

* *

Application design

Collector provider applied to select implemented by Application models and classes

Collector and context operator designer Application designer

− defines and implement collectors and operators − models and implements application − describes collector and operators semantics

Context specification and design

Application designer Context−awareness designer

Context−Awareness designer (CA3M)

− defines entities − for each entity, selects observables and situations − defines notification and observation contracts Context− awareness models − defines catalog of generic situations

Context specifier

Observable Context specifier − describes situations semantics

# 26 ✬ ✫ ✩ ✪

3.3 CA3M runtime architecture

Collector Bridges

notification

• bservation
• bservation

notification

CACONTROLLER

bridge creation callback updating browsing, Context Awareness Model CA3M Meta−Model conforms to

Model Manager

model updating

CA3M

Distributed context providers (context managers)

Distributed context sources (including sensors)

several technologies and providers

• bservation

notification

Context−aware Application

Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 17

Context management for ubiquitous applications 3 Design of context-aware applications with CA3M # 27 ✬ ✫ ✩ ✪

3.4 Context awareness meta-model

entities

• bservables
• bservable

InterpretedObservable upperCardinality: Integer lowerCardinality: Integer contracts ContextAwareness Contract ContextAwareSystem Entity EntityRelation Observable ObservationContract NotificationContract derivedFrom

0..* 0..* 0..*

entityRelations linkedEntities

0..* 0..* 0..* 1..1 1..*

callbackClass: String triggerCond: TriggerCond Eric Eric peers Eric computer Battery level Location Network QoC

# 28 ✬ ✫ ✩ ✪

3.5 From observable model to runtime bridge

Bridge pattern Several context management frameworks for a given observable /situation No standard API for context managers ◮ The bridge decouples application code from the collector API Runtime binding

Adapter ICollector ICollectorListenerRegister

MessageManager

Adapter CosmosBridgeComponent Cosmos Node ICollectorNotification

Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 18

Context management for ubiquitous applications 4 Quality of context information # 29 ✬ ✫ ✩ ✪

4 Quality of context information

4.2 Definition and motivations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31 4.2 Quality criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 4.4 QoC-aware context operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.4 Architecture of QoC-aware context operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 # 30 ✬ ✫ ✩ ✪

4.1 Definition and motivations

Quality of Context = Any inherent information that describes context information and can be used to determine the worth of the information for a specific application [Krause and Hochstatter, 2005] Context data are imperfect [Henricksen et al., 2002]: Incorrect: if it fails to reflect the true state of the world it models Inconsistent: if it contains contradictory information Incomplete: if some aspects of the context are not known Decision based on erroneous context data [Chen and Kotz, 2000, Dey et al., 2001] Cost of reasoning about uncertain contexts [Ranganathan et al., 2004, Dargie and Hamann, 2006] Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 19

Context management for ubiquitous applications 4 Quality of context information # 31 ✬ ✫ ✩ ✪

4.2 Quality criteria

[Buchholz et al., 2003, Preuveneers and Berbers, 2007]:

• 1. Accuracy: how exactly the provided context data mirrors the reality
• 2. Precision: how detailed a measurement is stated
• 3. Probability of correctness: probability that a piece of context data is correct;

unintentional error

• 4. Trust-worthiness: how likely it is that the provided data is correct; rated by client
• 5. Resolution: granularity of information
• 6. Up-to-dateness/freshness: age of context information

# 32 ✬ ✫ ✩ ✪

4.3 QoC-aware context operators

[Abid et al., 2009a, Abid et al., 2009b]

Operator QoCParameter

0..n

Arithmetic Comparison

... ...

Fuzzy Logic Threshold

... ... ...

compute QoC

Precision Accuracy Trustworthiness UpToDateness

...

Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 20

Context management for ubiquitous applications # 33 ✬ ✫ ✩ ✪

4.4 Architecture of QoC-aware context operators

Weaving of QoC aspect

Pull Push

Context Operator QoC Operator ChildFaçade

Pull Push Data QoC Get

QoCAwareContextOperator ParentFaçade

QoC Compute Push Pull Push Pull

# 34 ✬ ✫ ✩ ✪

5 E-commerce flash-sale demonstrator

5.3 Flash sale cosmos node . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 5.3 Siafu simulator of the shopping center. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37 5.3 Flash sale notifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 21

Context management for ubiquitous applications 5 E-commerce flash-sale demonstrator # 35 ✬ ✫ ✩ ✪

5.1 Flash sale cosmos node

Notification of flash sales in a shopping center (Cappucino project) Identification of situations (qualified by QoC)

Active observer and notifier System call

Data interpretation Adaptation situation detection Data gathering

System call System call System call

Legend:

Blocking observation Active notifier Active observer Blocking notification

A A A A A A T T T T T T F F F F F F F

WiFi manager Bluetooth

location choice manager User profile manager Shop manager flash−sale filter Privacy rules Flash sale offer Matching information ? Transmission decision ac.t. freshness, accuracy, trust−worthiness and privacy Adjust Location Speed Stabilized GPRS manager A T A T F

# 36 ✬ ✫ ✩ ✪

5.2 Siafu simulator of the shopping center

COSMOS collectors wrap access to the siafu simulator through web services Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 22

Context management for ubiquitous applications 5 E-commerce flash-sale demonstrator # 37 ✬ ✫ ✩ ✪

5.3 Flash sale notifications

Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 23

Context management for ubiquitous applications 6 Complementarity of ontology-based and process-oriented context managemen # 38 ✬ ✫ ✩ ✪

6 Complementarity of ontology-based and process-oriented context management

Ontology-based = data orientation CoBrA [Chen et al., 2004], SOCAM [Gu et al., 2004], GAIA [Román et al., 2002] and MUSE [Bouzeghoub and Do Ngoc, 2008] Process-oriented context management Context Toolkit [Dey et al., 2001], Contextor [Coutaz and Rey, 2002], MoCA [da Rocha and Endler, 2006], MUSIC [Paspallis et al., 2008] # 39 ✬ ✫ ✩ ✪

6.1 Two abstraction levels for classifying situations

An illustrative scenario Marie enters the campus at 08:00am. An application displays the necessary documents for her next class as well as the room number and location. Another application finds the nearest available printers which are in unengaged rooms. At lunchtime, John is close to the self-service restaurant. A lot of people are waiting in the line. Since John has a meeting scheduled within one hour, an application proposes him the delivery of a tray lunch in his office adapted to his culinary habits.

• 1. Computed from the context data collected from sensors, etc.

The situation “a visitor enters the restaurant” is detected from dynamic context data The same for the estimation of the waiting time in the line

• 2. Requiring the use of both context collectors’ data and a knowledge base

The recommendation “delivery of a tray lunch in the visitor’s office” requires the use of both context collectors’ data and a knowledge base: the visitor’s profile, agenda and habits Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 24

Context management for ubiquitous applications 6 Complementarity of ontology-based and process-oriented context managemen # 40 ✬ ✫ ✩ ✪

6.2 Hybrid architecture

callback infer/read/update

• bserve

Context Proxy

recommend search

Business Applications

Inference Engine Context Ontology Business Application Ontology

MUSE OCM

• bserve

infer/read callback

COSMOS PCM

notify update

Context sources

notify

(e.g. sensors, user profiles, platform)

• bserve

(forest of context nodes)

# 41 ✬ ✫ ✩ ✪

6.3 Comparison of OCM and PCM

Observation/notification: OCM and PCM provide both Handling constantly evolving context PCM: filter of situations (less heavy-weight and more efficient for detecting low-level situations) Distributed context management, scalability, and constrained device On mobile devices by PCM and on a central server by the OCM Observation preparation PCM: control of the flow (e.g., “blocking node” and “active node”) Situation identification latency PCM: full control of activities performing context inference Handling unforeseen situations OCM: reasoning process able to deal with unforeseen situations (find the most similar situation) Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 25

Context management for ubiquitous applications 6 Complementarity of ontology-based and process-oriented context managemen # 42 ✬ ✫ ✩ ✪ Expression of situations OCM: several languages for context modelling and context reasoning Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 26

Context management for ubiquitous applications # 43 ✬ ✫ ✩ ✪

7 Concluding remarks

Context management = 1st class concern for adaptation in ubiquitous environments COSMOS: process orientation, component, architecture, fine-grained configuration and management of OS resources CA3M: meta-model and model of context-awareness at runtime, and design process QoC: expression of quality into observation and notification contracts Complementarity of process-oriented and ontology-based context managers Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 27

Context management for ubiquitous applications 7 Concluding remarks # 44 ✬ ✫ ✩ ✪

Related work/Process-oriented context management

Context Toolkit [Dey et al., 2001] Event programming and widget concepts Interpretation and aggregation functionalities in monolithic blocks No thread management MoCA Context Service architecture [da Rocha and Endler, 2006] Context data are typed objects and described using an XML-based model Partition of the context information space into views for improving the performance The Contextor [Coutaz and Rey, 2002] Implicit component model No sharing of contextors, no management of OS resources # 45 ✬ ✫ ✩ ✪

Related work / Distribution of context data

Message passing systems for distributed system monitoring Phoenix [Boutros Saab et al., 2002] and LeWYS [Cecchet et al., 2005] Object request broker GAIA [Román et al., 2002]: CORBA-based RESTful approach: context data as a resource SPACES [Romero et al., 2009, Romero et al., 2007] Tuple-spaces EgoSpaces [Julien and Roman, 2006]: views are sets of tuples, are owned by agents, are located on hosts in a network Peer-to-peer systems [Nguyen and Rouvrais, 2007]: every context data is shared in a dynamic overlay MUSIC context manager [Paspallis et al., 2008]: sensor peers, disseminator peers, consumer peers, protocol for group formation using JXTA Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 28

Context management for ubiquitous applications 7 Concluding remarks

References

[Abid et al., 2009a] Abid, Z., Chabridon, S., and Conan, D. (2009a). A Framework for Quality of Context

• Management. In Proc. First International Workshop on Quality of Context, volume 5786 of Lecture Notes

in Computer Science, Stuttgart, Germany. Springer-Verlag. [Abid et al., 2009b] Abid, Z., Chabridon, S., and Conan, D. (2009b). Cohérence et qualité des informations de contexte en environnement pervasif. In Proc. 3ème Workshop sur la Cohérence des Données en Univers Réparti - CDUR’09, Toulouse, France. [Boutros Saab et al., 2002] Boutros Saab, C., Bonnaire, X., and Folliot, B. (2002). PHOENIX: A Self Adapt- able Monitoring Platform for Cluster Management. Cluster Computing, 5(1):75–85. [Bouzeghoub and Do Ngoc, 2008] Bouzeghoub, A. and Do Ngoc, K. (2008). A situation based metadata for describing pervasive learning objects. In Proc. of 1st International Conference on Mobile Learning, University of Wolverhampton, Ironbridge, UK. [Bouzeghoub et al., 2010] Bouzeghoub, A., Taconet, C., Jarraya, A., Do, N., and Conan, D. (2010). Comple- mentarity of Process-oriented and Ontology-based Context Managers to Identify Situations. In Submitted to CMMSE’2010. [Buchholz et al., 2003] Buchholz, T., Kupper, A., and Schiffers, M. (2003). Quality of context information: What it is and why we need it. In 10th International Workshop of the HP OpenView University Association (HPOVUA), ACM, Geneva, Switzerland, Hewlett-Packard OpenView University Association. [Cecchet et al., 2005] Cecchet, E., Elmeleegy, H., Layaïda, O., and Quéma, V. (2005). Implementing Probes for J2EE Cluster Monitoring. Studia Informatica. [Chen and Kotz, 2000] Chen, G. and Kotz, D. (2000). A Survey of Context-Aware Mobile Computing Re-

• search. Technical Report TR2000-381, Dartmouth College, Hanover, NH, (USA).

[Chen et al., 2004] Chen, H., Finin, T., and Joshi, A. (2004). An Ontology for Context-Aware Pervasive Computing Environments. Special Issue on Ontologies for Distributed Systems, Knowledge Engineering Review, 18(3):197–207. [Conan et al., 2007] Conan, D., Rouvoy, R., and Seinturier, L. (2007). Scalable Processing of Context Information with COSMOS. In Indulska, J. and Raymonds, K., editors, Proc. 6th, volume 4531 of Lecture Notes in Computer Science, pages 210–224, Paphos, Cyprus. Springer-Verlag. [Conan et al., 2008] Conan, D., Rouvoy, R., and Seinturier, L. (2008). COSMOS: composition de nœuds de

• contexte. Technique et Science Informatiques, 27(9–10):1189–1224.

[Coutaz and Rey, 2002] Coutaz, J. and Rey, G. (2002). Foundations for a Theory of Contextors. In Kolski,

• C. and Vanderdonckt, J., editors, Proc. 4th International Conference on Computer-Aided Design of User

Interfaces, pages 13–34, Valenciennes (France). Kluwer. [da Rocha and Endler, 2006] da Rocha, R. and Endler, M. (2006). Context Management in Heterogeneous, Evolving Ubiquitous Envrionments. IEEE Distributed Systems Online, 7(4). [Dargie and Hamann, 2006] Dargie, W. and Hamann, T. (2006). A distributed architecture for reasoning about a higher-level context. In IEEE International Conference on Wireless and Mobile Computing, Networking and Communications, 2006. (WiMob’2006), pages 268– 275. [Denning et al., 2005] Denning, P., Horning, J., Parnas, D., and Weinstein, L. (2005). Wikipedia Risks. Communications of the ACM, 48(12):152. [Dey et al., 2001] Dey, A., Salber, D., and Abowd, G. (2001). A conceptual framework and a toolkit for supporting the rapid prototyping of context-aware applications. Special issue on context-aware computing in the Human-Computer Interaction Journal, 16(2–4):97–166. [Gu et al., 2004] Gu, T., Wang, X. H., Pung, H. K., and Zhang, D. Q. (January 2004). An Ontology-based Context Model in Intelligent Environments. In Proceedings of Communication Networks and Distributed Systems Modeling and Simulation Conference, pages 270–275, San Diego, California, USA. Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 29

Context management for ubiquitous applications 7 Concluding remarks [Henricksen et al., 2002] Henricksen, K., Indulska, J., and Rakotonirainy, A. (2002). Modeling context in- formation in pervasive computing systems. In In: Mattern, F. anf Naghshineh, M. e., editor, PERVASIVE 2002, volume 2414 of Lecture Notes in Computer Science, pages 79–117, Heidelberg. Springer Verlag. [Julien and Roman, 2006] Julien, C. and Roman, G.-C. (2006). EgoSpaces: Facilitating Rapid Development

• f Context-Aware Mobile Applications. IEEE Transactions on Software Engineering, 32(5):281–298.

[Kephart and Chess, 2003] Kephart, J. and Chess, D. (2003). The Vision of Autonomic Computing. IEEE Computer, 36(1). [Krause and Hochstatter, 2005] Krause, M. and Hochstatter, I. (2005). Challenges in Modelling and Using Quality of Context (QoC). In et al., T. M., editor, Mobility Aware Technologies and Applications (MATA), volume 3744 of Lecture Notes in Computer Science, pages 324–333. Springer-Verlag. [MUSIC, 2007] MUSIC (2007). Requirements of methods, languages, algorithms, and tools to modeling and management of context. Technical report, IST MUSIC Project, Deliverable D2.1. [Nguyen and Rouvrais, 2007] Nguyen, T. and Rouvrais, S. (2007). Towards a Peer-to-peer Middleware for Context Provisioning in Spontaneous Networks. In Proc. 5th MiNEMA Workshop on Middleware for Network Ecentric and Mobile Applications, Magdeburg, Germany. [Paspallis et al., 2008] Paspallis, N., Rouvoy, R., Barone, P., Papadopoulos, G., Eliassen, F., and Mamelli, A. (2008). A Pluggable and Reconfigurable Architecture for a Context-aware Enabling Middleware System. In Proc. 10th International Symposium on Distributed Objects and Applications, volume 5331 of Lecture Notes in Computer Science, pages 553–570, Monterrey, Mexico. Springer-Verlag. [Preuveneers and Berbers, 2007] Preuveneers, D. and Berbers, Y. (2007). Architectural backpropagation support for managing ambiguous context in smart environments. In Proc. UAHCI Universal Access in Human-Computer Interaction, Ambient Interaction, volume 4555 of Lecture Notes in Computer Science, pages 178–187. Springer-Verlag. [Ranganathan et al., 2004] Ranganathan, A., Al-Muhtadi, J., and Campbell, R. (2004). Reasoning About Uncertain Contexts in Pervasive Computing Environments. IEEE Pervasive Computing, 3(2):10–18. [Román et al., 2002] Román, M., Hess, C., Cerqueira, R., Ranganathan, A., Campbell, R., and Nahrstedt,

• K. (2002). Gaia: A Middleware Infrastructure to Enable Active Spaces. IEEE Pervasive Computing,

1(4):74–83. [Romero et al., 2007] Romero, D., Hermosillo, G., Taherkordi, A., Nzekwa, R., Rouvoy, R., and Eliassen, F. (2007). RESTful Integration of Heterogeneous Devices in Pervasive Environments. In Proc. 10th, Lecture Notes in Computer Science, Amsterdam, Netherlands. Springer-Verlag. [Romero et al., 2009] Romero, D., Rouvoy, R., Chabridon, S., Conan, D., Pessemier, N., and Seinturier, N. (2009). Enabling Context-Aware Web Services: A Middleware Approach for Ubiquitous Environments. Chapman and Hall/CRC. [Rouvoy et al., 2008] Rouvoy, R., Conan, D., and Seinturier, L. (2008). Software Architecture Patterns for a Context Processing Middleware Framework. IEEE Distributed Systems Online, 9(6). [Seyler et al., 2007] Seyler, F., Taconet, C., and Bernard, G. (2007). Context Adaptation of Web Service

• Orchestrations. In 16th IEEE International Workshops on Enabling Technologies: Infrastructure for Col-

laborative Enterprises. WETICE’07, Paris. [Taconet et al., 2009] Taconet, C., Kazi-Aoul, Z., Zaier, M., and Conan, D. (2009). CA3M: A runtime model and a middleware for dynamic context management. In Proc. 11th International Symposium on Distributed Objects and Applications, volume 5870 of Lecture Notes in Computer Science, pages 513–530, Algarve, Portugal. Springer-Verlag. [Ye et al., 2009] Ye, J., Coyle, L., Dobson, S., and Nixon, P. (2009). Using Situation Lattices in Sensor Analysis. In Proc. of IEEE International Conference on Pervasive Computing and Communications, pages 1–11, Galveston, TX, USA. Télécom SudParis — Denis Conan, Chantal Taconet, Sophie Chabridon — juin 2010 — Ubimob 2010 30