Usage Scenario: Coalition Search & Rescue Policy Service
Andrzej Uszok, Institute for Human and Machine Cognition, Pensacola, FL;
Austin Tate, Artificial Intelligence Applications Institute, University of Edinburgh
URL(s) or other references:
Domain: Integration of policy services with service discovery, matchmaking and planning
This is an example of a service enabling coordination of operation using coalition resources, described using DAML-S, constrained by heterogeneous distributed set of policies.
Most organizations possess policies controlling their organizational procedures, access to information, use of resources, behavior of actors, obligations in relation to audit, logging, etc. These policies differ from organization to organization. When organizations create a coalition to perform common tasks it is necessary to distribute, coordinate and enforce these policies in the new virtual organization.
An organization interested in preserving its policies within the scope of a coalition.
Coalition partners interested in being informed about constraints put on resources before planning.
Resource owners who want to control and protect the use of their resources.
DAML-S service matchmaking
I-X (semi-)automated plan refinement and execution support
KAoS Policy and Domain Services
To plan and perform a search and rescue operation dynamically discovering coalition resources while at the same time assuring conformity to policies constraining visibility and use of these resources, as well as arbitrary actions of distributed clients, as dictated by the partners.
Inputs and Output formats agreed.
The rescue resources offered by the coalition partners are described using DAML-S and registered in Matchmaker:
When the US pilot is shot down the request to organize the search and rescue
operations is forwarded to the Coalition SAR coordinator. It tries to obtain
available rescue pick-up resources and synchronize them with medical facility
to be used during the operation. Arabello policy forbids helicopters from Gao
to land in Arabello. Thus, when a suitable hospital in Arabello is
chosen by the SAR coordinator, the KAoS Policy and Domain Service is aware of
this policy block of the Gao offer when it is returned by the matchmaker. It
is important to note here that the matchmaker is a generic tool that should
not be aware aware of the various policies defined by the distributed organizations.
These policies are enforced at the boundaries of the coalition.
Other Use Cases will show richer planning scenarios and more complex set of policies, with reasoning to resolve issues of potentially conflicting policies.
Ontology describing actors and policies - KAoS ontology
SONAT Elements of National Power (ENP) ontologies of countries, medical resources, military capabilities, etc.:
DAML.org ontologies Location Ontology
Ontology describing rescue resources - Rescue Resources ontology
Description logic reasoning for policy analyses and disclosure,
DAML relation reasoning to discover actor types
Domain relation reasoning to find peers, superiors and subordinates between actors.
An additional mechanism to implement is a functionality to return agents which
will be authorized to perform this task based on the policies.
We also plan to work on better integration between DAML-S (OWL-S) description of services and KAoS policy service implementing automatic mapping functionality.
Policy negotiation mechanism between domains is also considered.