A Roadmap to Policy Research Accomplishments and Future Challenges

1
A Roadmap to Policy Research -Accomplishments
and Future Challenges -

• Duminda Wijesekera
• George Mason University

2
Policies a working definition

• Policies define rules governing changes in the
  system behavior
• Emil Lupu, 2005
• Behavior Do we know what they are ?
• Things that we would like to see happening
• Requirements, use cases, promises and obligations
  
• Things that we detest happening
• Faults, exceptions, safety and security violations

3
Policies a working definition - II

• Policies define rules governing changes in the
  system behavior
• Emil Lupu, 2005
• Rules What constitute a rule?
• Norms, laws, treaties, duties and informal
  agreements
• Mostly says what but not always how
• At a higher level, from societal to formal
• Behavioral models that says how and therefore
  what
• Agent based systems, military duty cycles etc.

4
The Context for Policies

• Distributed systems Good things we like
• (Applications Operating systems Networks)
• Started with algorithms on semi-formal models
• EG Producer consumer, termination detection,
  consensus building, voting, sharing, duplication
  etc.
• Infrastructure that made them work
• Tool support, compilers, distributed event
  service, interfaces.
• The need to have guarantees
• Verification and validation, proof systems
• Simulating and testing

5
Where did policies come in ?

• Many choices for behavior some like it one way,
  other want it their way.
• Need to customize from a modular artifact
• Parametrizing the choices within a singular.
• EG cache update policies, scheduling policies,
• Encoding defaults always acknowledge, notify if
  failed, notify if succeed.
• Permit if not prohibited, prohibit if not
  permitted.
• Policy as a means of customizing lean Kernals

6
The Present Applications 1

• Modeling contemporary behaviors
• Identities of requestors
• No one knows everyone
• Can provide introductions to the unknowns through
  a chain of knowns A.K.A. trust management
  systems.
• RT, Cassendra etc.
• Identity management
• Ease of use, single sign on etc.

7
The Present Applications II

• Modeling contemporary behaviors
• Unknowns negotiate to introduce themselves
• Privacy concerns not every request grantor need
  to know all of the requestor attributes
• Not every service is granted for good!
• Revocation cascading / non-cascading
• Revoking the right to revoke
• Constraints in granting
• Mutual exclusion principles
• Preventing insider misbehavior

8
The Present Applications III

• What about before and after?
• Prerequisites
• Need to check if preconditions of the request are
  satisfied.
• Provisions may need to depend on someone else to
  check them
• Post-requisites
• Obligations request, once granted binds the
  requestor to satisfy some conditions
• Under permissible exceptions, these obligations
  can be delegated, renegotiated.
• Need to monitor obligation fulfilment

9
The Present Infrastructure I

• New networking infrastructures are emerging
• Photonics
• (G)MPLS for the network core
• Policies to divide the bandwidth, schedule and
  police it usage
• New application that use them
• Command and control networks that use GMPLS have
  strong security requirements.

10
The Present Infrastructure II

• Co-Habilitation of the Old
• IP-based and PSTN-based Voice service
  interoperation
• Packet switched and circuit switched networks
  operating through gateways OR
• Adaptation layers requiring policies
• Bring the vulnerabilities and fault in one area
  to the other.

11
Heading On Applications

• Applications from the control domain
• SCADA security vs. QoS
• Signal streams that control remote stations
• Signal sequence and content change remote
  application behavior
• Need to screen signal in (near) real time
• Positive Train Control Security and Trust
• Inter operating trains need to ride on each
  others signal controlled tracks.
• Slightly more complex than PSTN because one
  companys train go on the other companys track

12
Heading On Technology

• Applications come over the web
• Need to know each other linguistic conventions
• Ontologies, Resources and usage policies
• Workflow oriented systems
• Need to specify policy pairs (producer, consumer)
  or groups of policies requiring
• Negotiation frameworks
• Ensuring the consistency, completeness, fairness
  and liveliness will become issues

13
Heading On Theory

• Application with richer semantics driving
  researchers to develop more expressive languages
  such as
• Constraint programming with sets, bags etc
• Inductive and abductive proof systems
• More elaborate calculi of dynamism and
  interactions such as p-calculus, y-calculus.
• More elaborate modeling of unacceptable behaviors
  using heavier machinery such as Strand Spaces
  etc.
• Agent-based and hybrid control policies.

14
Summary

• Policies have come a long way in specifying
  accepted and detested behavior.
• Emerging behavior patterns have interactivity,
  constraints, and continuity
• what was considered post usage behaviors (of the
  past) will be negotiated and agreed upon,
  monitored and rewarded/reprimanded.
• Perhaps more non-particle like (hybrid)
  behaviors.
• More detailed semantics of detestable behavior.
• Discrete and continuous?
• Will require new languages, tools and analysis
  techniques to go forward.
• Therein lies the hope and optimism!