Policies - Wiki LES PUC-Rio

Policy-Based Context-Aware
Applications for Mobile
Computing
José Viterbo Filho
[email protected]
Laboratory for Advanced Collaboration
Norms and policies
• Multi-agent systems usually assume some form
of structure, or set of norms or conventions that
articulate or restrain interactions in order to
make them more effective…
• Mobile computing represents a typical scenario
where these norms or policies are constantly
changing
• Inside this scenario, agents must be able to adapt
to each new set of rules and policies accordingly
Gustavo R. Carvalho.
Governance in Open Multi-Agent Systems.
Seminar’s Opening Session, 2005.
Mobile computing
• Using agent
computing:
technology
to
support
mobile
– Is motivated by agent’s intrinsic properties such as
autonomy, mobility, proactivity
• In order to exploit these properties, policies may
be attached to agents’ behaviors
• These policies define
– the actions that agents can perform
– resources that they can use
– information that they can access…
H. Harroud., M. Khedr and A. Karmouch - University of Ottawa.
Building Policy-Based Context-Aware Applications for Mobile Environments.
Mobility Aware Technologies and Applications, MATA 2004.
Context-aware
• In addition these policies are tightly
bound to the context in which they are
evaluated and used
• A policy may be triggered based on the
agent’s state, tasks and/or operating
environment conditions
• Applications must be able to efficiently
manage, exchange and interpret context
information
H. Harroud., M. Khedr and A. Karmouch - University of Ottawa.
Building Policy-Based Context-Aware Applications for Mobile Environments.
Mobility Aware Technologies and Applications, MATA 2004.
Example 1: Smart Environments
• Mobile and wearable computational devices that
interact with the environment are becoming
widespread…
– NOT ONLY: laptops and PDAs
– BUT ALSO: watches, active badges, smartphones, mp3
players, digital cameras
• Devices should be able to automatically
reconfigure themselves based on current context
and user preferences
• Behavior rules may be imposed to devices that
enter a given space
K. Connelly and A. Khalil - Indiana University.
On Negotiating Automatic Device Configuration in Smart Environments.
2nd IEEE Annual Conf. on Pervasive Computing and Communications Workshops, 2004.
Example 1: Smart Environments
• A user may want his cell phone to vibrate,
instead of ringing, whenever he enters in a
theater or concert room IMPROVE USABILITY
• Following legislation that outlaws the use of video
phones in locker rooms, a gym may want the
ability to temporarily turn off the camera
capabilities of a video phone PREVENT ABUSE
• A university may wish to disable all cell phone
calls (except for an outgoing 911) in a lecture
hall when an exam is taking place PREVENT ABUSE
K. Connelly and A. Khalil - Indiana University.
On Negotiating Automatic Device Configuration in Smart Environments.
2nd IEEE Annual Conf. on Pervasive Computing and Communications Workshops, 2004.
Example 1: Smart Environments
• Aside from the user preferences (policies), the
smart environment may have its own policies
when it comes to devices brought into that space
• Conflict may arise when the device policy differs
from the space policy
• Policy resolution techniques may be implemented
automatically adopting an agent-based negotiation
approach
K. Connelly and A. Khalil - Indiana University.
On Negotiating Automatic Device Configuration in Smart Environments.
2nd IEEE Annual Conf. on Pervasive Computing and Communications Workshops, 2004.
Example 2: Travel aide
• A university professor flies to a new city and stays the night
for two project meetings, one in the morning, one in the
afternoon
• With a MAS, the professor’s PDA has a connection with the
local weather and traffic network, his agent matches this
information with his calendar and wakes him up 15 minutes
earlier, preventing him to be delayed by a traffic jam
• In the first meeting, his agent receives the information that
the second meeting was postponed, and arranges for another
night’s stay
M. Khedr and A. Karmouch - University of Ottawa.
ACAI: Agent-Based Context-Aware Infrastructure for Spontaneous Applications.
Journal of Network and Computer Applications, 2005.
Example 2: Travel aide
• To evolve from a passive state to an active
pervasive state, the infrastructure must be
able to support uniform context
representation, to reason about context,
to offer context-based service discovery,
and to support a context management and
communication protocol
M. Khedr and A. Karmouch - University of Ottawa.
ACAI: Agent-Based Context-Aware Infrastructure for Spontaneous Applications.
Journal of Network and Computer Applications, 2005.
Example 2: Travel aide
Context provider
Context consumer
Common context
representation
Must be able to reason
about context
M. Khedr and A. Karmouch - University of Ottawa.
ACAI: Agent-Based Context-Aware Infrastructure for Spontaneous Applications.
Journal of Network and Computer Applications, 2005.
Policies and context again
• The effective contextual information that
is used by mobile applications is translated
into a set of policies
• Generated context policies are
represented semantically to help achieve
the common understanding across
different domains
H. Harroud., M. Khedr and A. Karmouch - University of Ottawa.
Building Policy-Based Context-Aware Applications for Mobile Environments.
Mobility Aware Technologies and Applications, MATA 2004.
Policy-based approach
• Agents are provided with a set of policies that
govern their overall behavior in the environment
and specify their privileges and constraints
• Each policy includes conditions that permit or
prohibit an agent to perform actions on target
components
• Conditions may concern the subject, the target or
a particular state or event of the operating
environment
H. Harroud., M. Khedr and A. Karmouch - University of Ottawa.
Building Policy-Based Context-Aware Applications for Mobile Environments.
Mobility Aware Technologies and Applications, MATA 2004.
Policy-based approach
UA
CPM
UA
SA
CPM
SA
PSA
SAT
User moves
PSA
SAT
Network
Network
H. Harroud., M. Khedr and A. Karmouch - University of Ottawa.
Building Policy-Based Context-Aware Applications for Mobile Environments.
Mobility Aware Technologies and Applications, MATA 2004.
Policy-based approach
CPM
PSA
SAT
• CONTEXT POLICY MANAGER
Responsible for monitoring the context information
and managing the environment resources based on
this context
• POLICY SERVICE AGENT
Manages policies of the domain under its
administration to control the behavior and
decision-making of the system agents
• SITE ASSISTANT
Is in charge of preparing and setting up a
temporary working environment to a user at a
visited site
H. Harroud., M. Khedr and A. Karmouch - University of Ottawa.
Building Policy-Based Context-Aware Applications for Mobile Environments.
Mobility Aware Technologies and Applications, MATA 2004.
Modeling Context and Policies
• Policies have to be defined and
understood by both machines and humans
and to be domain interoperable
• Modeled using ontology (OWL)
• The CLNP is a multi-attribute negotiation
protocol that allows automated context
identification and agreement
H. Harroud., M. Khedr and A. Karmouch - University of Ottawa.
Building Policy-Based Context-Aware Applications for Mobile Environments.
Mobility Aware Technologies and Applications, MATA 2004.
Next step
• Extending the context policy manager
with a fuzzy inference engine in
generating policies
• This will provide a necessary robustness
against uncertain situations in the
environment and maintain the
performance of the system even in vague
corrupted context information
H. Harroud., M. Khedr and A. Karmouch - University of Ottawa.
Building Policy-Based Context-Aware Applications for Mobile Environments.
Mobility Aware Technologies and Applications, MATA 2004.