On using Policies for Managing Service Provisioning in Pervasive Environments

1
On using Policies for Managing Service
Provisioning in Pervasive Environments

• Mohamed Ganna, Eric Horlait
• Laboratoire dInformatique de Paris 6
• Université Pierre Marie Curie, Paris, France
• mohamed.ganna_at_lip6.fr

2
Outline

• Motivations Objectives
• Our approach
• Environment components and associations
• Agents as management entities
• Automatic control using policies
• Evaluation
• Summary and Future Work

3
Motivations objectives

• Service and network changes adding/changing
  service/network components
• Mobility and heterogeneity of users and services
• Automatic service provisioning and control in
  pervasive environments
• Adaptive behaviour to support context changes
• Dynamic policy creation to handle different
  events
• Providing autonomous behavior of pervasive
  environments

4
Environment modeling

• Depict the environments entities
• Understand the relationships binding these
  components
• Building up XML profiles from collected
  environmental information
• Create policy association

5
Agent-based environment

• Splitting the management functions into
  management blocs Divide to Conquer
• Agent policies to automate agent tasks toward
  autonomous behavior

6
Policies for network configuration
Rules governing the behaviour of the environment

• Defining policies of the environment is quiet
  complex managing users, services and configuring
  resources
• New policies are created when needed Meta
  policies to control policy creation and
  adaptation
• XML policy representation
• Mapping of these policies into device dependent
  ones using environmental information (users,
  services, devices, )

7
Policies for agent management

• Policies to control agent behavior
• Agent collect and provide relevant information to
  adapt the network behavior by creating new
  policies to deal with new events

Policy
8
Policy/Profile manipulation

• Policies/Profiles are defined in a XML format
  ease of definition and manipulation
• Grasshopper as an agent definition platform
• Translating XML documents into Java object JDOM

9
Diagram
10
Evaluation (1) Authentication

• Web server Tomcat 4.0.31
• Radius server Freeradius 0.9.0
• Java Radius Client jradius client
• Agent platform Grasshopper 2.2.4b
• Evaluate introducing policies in the
  authentication process more flexibility

11
Evaluation (2) Negotiation

• Introducing policies in the negotiation process
• Put threshold to avoid loops
• Evaluate the behavior of policy based negotiation

12
Evaluation (3) Overall process
13
Summary

• Policy-driven agent-based framework
• Autonomous behavior of agent in pervasive
  environments
• Adaptive behavior is achieved using collected
  context information using agents
• Future work
• Security part is done and has to be included
• We have already divided the response time by 10
  optimizing code and restricting agent mobility
• Adding control techniques to the framework
• Adding policy trigger mechanisms

14
Thank you! Questions?