Context-aware Systems Context-aware Semantic Middleware Conclusions Context-aware Semantic Middleware Solutions for Pervasive Applications Alessandra Toninelli [email protected]Universit` a degli Studi di Bologna Department of Electronics, Information and Systems PhD Course – Infrastructure and Supports for Wireless Systems Universit` a degli Studi di Bologna – Facolt` a di Ingegneria 17 September 2009 A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
35
Embed
Context-aware Semantic Middleware Solutions for Pervasive … · 2012-09-03 · Context-aware Systems Context-aware Semantic Middleware Conclusions Background & Motivations State
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Context is any information that can be used to characterize the state orthe activity of an entity, and the environment where this entity operates.(based on [Dey, Salber & Abowd, 2001])
I Any information in principle could be context, depending on the application.I Location, device status & capabilities, user preferences, environmental
conditions, date/time, past context, ...
Context Awareness is the ability of a system/application to use availablecontext information to change its behavior accordingly.
→ context-aware adaptation
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
Metadata provide a declarative approach to describe:1 the structure/meaning of resources composing a system and their state→ profilese.g., ”this device belongs to Alessandra and its battery status is now 70%”.
2 the specification of management operations expressed at a high level ofabstraction→ policiese.g., ”if the device battery goes under 30% disable application X”.
Metadata allow to specify context and context-aware adaptation strategieswithout affecting the application logic
→ separation of concerns
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
The Semantic WebSemantic MetadataContext-Aware Access ControlSocially Aware Policies
Some History
First article by Berners-Lee, Hendler & Lassila (2001)
Initially a lot of hype and several contributions...I AI (ontology languages, rules, reasoners)I Web-oriented applications (annotations, folksonomy, ...)I DB management (huge ontology KBs)
But after some years things were getting too complicated to be usable :-(
Towards a usable trade-offI Coexisting ontology languages, used for different purposesI Small interconnected ontologies vs. huge agreed ontologiesI Inconsistent knowledge as a matter of fact
... ”A little semantics goes a long way”, Jim Hendler
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
The Semantic WebSemantic MetadataContext-Aware Access ControlSocially Aware Policies
Original Research Directions
PhD Thesis → Semantic technologies (metadata) to buildcontext-aware middleware
1 Context-aware discovery – matureF Context is used to provide mobile users with a personalized view on servicesF Semantic-based profiles allow flexible matching between user requirements and
3 Socially-aware applicationsF Context is used to personalize user experience with social applicationsF Social characterization of context (the other way round)
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
The Semantic WebSemantic MetadataContext-Aware Access ControlSocially Aware Policies
Context-Aware Access Control Policies
Example
In case of emergency, any qualified physician located within the hospitalis allowed to access Alice’s health protected information
Access control policiesI High level directives defining who can access which resource under which
circumstances
Traditional policies based on identities/roles – static
Context-aware access control policiesI Definition of policies based on contextI Use of semantic technologies to represent & reason about policies/contexts
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
The Semantic WebSemantic MetadataContext-Aware Access ControlSocially Aware Policies
Context-Aware Access Control Policies
Example
In case of emergency, any qualified physician located within the hospitalis allowed to access Alice’s health protected information
Access control policiesI High level directives defining who can access which resource under which
circumstances
Traditional policies based on identities/roles – static
Context-aware access control policiesI Definition of policies based on contextI Use of semantic technologies to represent & reason about policies/contexts
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
The Semantic WebSemantic MetadataContext-Aware Access ControlSocially Aware Policies
The Proteus Context-Aware Policy Framework
Proteus policies associate resources to protection contexts
A protection context is a set of attributes & constrained values
The current state is a set of attributes & values measured by ”sensors”I A protection context is active if context assertions describing the current state
match its context elements
Activation of protection contexts (and associated policies) allows accessto a resource
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
The Semantic WebSemantic MetadataContext-Aware Access ControlSocially Aware Policies
Proteus Context & Policy Representation
Example
In case of emergency, any qualified physician located within the hospitalis allowed to access Alice’s health protected information
PersonalEmergencyContext ≡ProtectionContext u ∃ owner.Alice u ∃ requestor.InHospitalQualifiedPhysician u∃ resource.AliceHPI u ∃ environment.PersonalEmergency
< Dr .Green, located , EmergencyRoom >< CurrentState, environment, PersonalEmergency >
Context & policy representation based on Description Logic
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
The Semantic WebSemantic MetadataContext-Aware Access ControlSocially Aware Policies
Proteus Middleware Architecture
OWL-DL to represent contexts and policies
Java prototype v1.0I Pellet 1.5 DL reasoner via OWL API & SPARQLI PEM implementation via Java Security Manager extension (JAAS)I CM implemented on top of Contory context provisioning platform
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
The Semantic WebSemantic MetadataContext-Aware Access ControlSocially Aware Policies
Usable Security on Mobile Phones
Relevant use cases in mobile phone usage modeled as access control issues:I access to user status information (status sharing)I access to the user’s attention (call filtering)
Access control policiesI High level directives defining who can access which resource under which
circumstances
Why is it difficult for mobile phone users to express their policies?I Users are not able to handle technical details – only details?I Users don’t understand policy models they are supposed to useI User strategies are strongly dependent on their social context –
software systems are not
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
The Semantic WebSemantic MetadataContext-Aware Access ControlSocially Aware Policies
Access Control Policies - State of the Art
Powerful representation model for access control decisions
Research on policy specification, management & enforcementI Different logical approaches (LP, ILP, DL, abduction, ...)I Semantic policies (e.g. KAoS, Rei)I Conflict analysis & consistency checkI Policy refinement
Limitations of existing policy languages/modelsI Little support for social vocabulary in policy definitionI Social data remain fragmented (no semantic interconnection)I Inflexible policy definition and retrieval (fixed place, time & application)
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
The Semantic WebSemantic MetadataContext-Aware Access ControlSocially Aware Policies
A Socially Aware Policy Model
A policy is modeled as a set of attributes and constrained values defining:I what (the resource)I who (the requestor - a socially aware description)I the context (when, where, why)I how (the access modality)
The current state is modeled as a set of attribute-value pairs
A policy is in effect if the current state values matchthe definition of the policy constraints.
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
The Semantic WebSemantic MetadataContext-Aware Access ControlSocially Aware Policies
Why a Graph-based Policy Model
RDF graphs allow to create and semantically interconnect social datafrom different sources
I Different applications running on the user’s behalf (e.g., calendar, addressbook, facebook)
I Phones belonging to different users (e.g., colleagues, family)
The policy graph can be browsed in multiple directions andstarting from different applications
I From the address book - which status info can this person access?I From the calendar - who can call (access) me during this event?I From the document manager - who can see this document in which situation?I And from the policy manager, too
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
Context-aware systems at the state of the artI Context representation modelsI Representation & enforcement of context-aware adaptation strategiesI Context management support services
Semantic metadata to build context-aware middlewareI Meaningful exchange of context information → interoperabilityI Reasoning to infer new context knowledge → flexibility and adaptation
Semantic approaches require a tradeoff between complexity and expressivityI Performance (reasoning takes time and resources)I Scalability (when dealing with thousands of triples)
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
Focus on information in distributed systems designI Wide-scale integration of devices/services based on interoperable
information exchange (including context)I Interoperability is moving at the information level – Semantic Web as key
technology?
Mobile phones as truly ubiquitous devices
Emerging trends in ubiquitous applicationsI Social characterization of context – Social Semantic WebI Network-based approaches rising up to the application level – Web 2.0 and
othersI Cloud computing – infrastructure/middleware as a service
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009
Managing the ”web of data”I Provenance tracking – where does this data graph come from?I Graph partitioning – how can we split this graph without loss of information?I Link metadata to data (e.g., link access control policies to controlled
resources)I Information overload – usability and performance issues
Novel approaches to (semantic) context-aware middleware should considerthe network and socially aware nature of pervasive applications
A. Toninelli - Context-aware Semantic Middleware Infrastructure and Supports for Wireless Systems – Bologna, 17 Sep 2009