Agent Grid

1
Agent Grid http//www.objs.com/agility/index.html
PI Craig Thompson Object Services and
Consulting, Inc. (OBJS) thompson_at_objs.com,
http//www.objs.com DARPA coABS Program PM
Jim Hendler
Acknowledgements to Brian Kettler (ISX) and Frank
Manola (OBJS)
2
Agent Grid - System Concept View
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
Agents the Global Grid
3
Agent Grid
http//www.objs.com/agility/tech-reports/990623-ch
aracterizing-the-agent-grid.html

• Our paper Characterizing the Agent Grid
• documents examples of grids
• describes views of the Agent Grid as
• a set of agent mechanisms
• a global registry/system management backplane
  environment for agents and agent systems that
  provides resource, services, and system wide
  properties
• a collection of interacting semantic grids
  representing various kinds of collections
  organizations, teams, ensembles (including ALP)
  all acting like mini-grids to control local
  resources
• all of the above
• lists grid architecture issues
• Is the agent grid itself a kind of agent system?
• Is the agent grid logically centralized?
  hierarchical?
• How can we use existing object services? Must
  they be wrapped as agent services?
• Is there a minimal set of services?
• will be published in Bradshaws agent book
  forthcoming

4
Grid Vision Meta Comments

• The Grid Vision is evolving
• Goals of Grid Vision Effort
• capture shared vision for Grid from CoABS
  community to guide development of Grid prototypes
• communicate vision to potential Grid users
• GITI/ISX has put out a strawman draft for
  discussion
• GITI/ISX Vision Team coordinating the process
• Vision Doc Ver. 2.0 (June 1999) - released
  internally to CoABS
• Current version is not necessarily a proper
  subset of CoABS Program Vision...
• Will capture additional community inputs and
  iterate...
• discussions at the June workshop
• bottom-up inputs from prototype building
  experience
• establish affinity groups to produce interface
  descriptions for specific services
• coordination with ALP, CPOF, CAST, and other
  efforts

5
The Challenge

• Information Systems in the Military (and
  elsewhere) are
• moving from sneaker net to network-centric to
  information-centric systems exploiting
  interconnectivity
• e.g., USAF Global Grid (C2 CONOPS), USAF SAB
  Battlespace Infosphere, ABIS Information Grid,
  etc.
• enabled by faster, cheaper hardware and software
  technologies (telecomm, WWW, Java, etc.)
• The Challenge
• How do we go from isolated applications to
  interconnected and interoperating super
  applications (aka systems of systems) that
  can work together to solve complex tasks in
  dynamic environments and that are built with
  minimum effort and maximum reuse?
• need a new name for these concepts

6
An Example

• Problem
• Rapidly configure a collection of military
  command and control systems in 24 hours to handle
  a new kind of crisis involving a coalition of the
  US and various countries that have never worked
  together before.
• Need a super application tailored to the
  current crisis and battlespace...
• Tasks to achieve
• establish interconnectivity (networks, etc.) over
  a wide area
• establish interoperability among systems (syntax
  and semantics of data exchanged)
• manage computing and data resources and protect
  them from misuse (by enemy and coalition members)
• allow human C2 staff to interact with the super
  application
• reconfigure the super application to handle
  changes in the mission, battlespace, coalition
  membership, collection of IT systems and
  resources (computing, comm, and data) available,
  etc.

7
Target Agent Grid Capabilities

• When your personal assistant connects to the
  Grid, it tells the Gridwhere you are, what you
  are doing, how your resources are
  configured,what supplies you need, and so on.
• Your forces might be dynamically reassigned to a
  new plan your computer equipmentmight briefly
  be recruited to run a meteorologicalsimulation
  by a load-balancing agent due to your personal
  expertise inArabic, you might receive documents
  to translate, or perhaps not if theGrid realizes
  your time is already claimed by other
  responsibilities.
• All resources - mental and material, human and
  non-human, permanentand ephemeral - are balanced
  by the Grid. Goals are reconciled byagents in
  the Grid and priorities are established.
• Whatever kind of agent you are, when you enter
  the Grid, youimmediately become part of a
  larger, coherent system. And when you leave the
  Gridthe Grid prepares for your return by
  generating statusreports, reading and
  summarizing your mail, planning how to use
  yourresources, and so on.

8
Requirements View

• Target operational requirements
• Humans and agents connect to the agent grid
  anytime from anywhere and get the information and
  capability they need. Enable teams led by humans
  and staffed by agents.
• Intelligent automation -- easier application
  connectivity where networks of agents
  self-organized at run-time. Reduce the 60 of
  time in command and control systems spent
  manipulating stovepipes incrementally replace
  stovepipes.
• Connect the 40B worth of DoD equipment that
  currently only interoperates with one or two
  other components, permitting better knowledge
  sharing. Another example is a process
  improvement in factory 1 is broadcast immediately
  to factories 2..N.
• Agent-enable object and web applications to
  reconfigure as new data and function is added to
  the system. Add capability modularly. Stable,
  scaleable, evolvable, reliable, secure,
  survivable, ...
• Scale to millions of agents so agents are
  pervasive and information and computation is not
  restricted to machine or organization boundaries.
• Survivable so if one agent goes down, another
  takes its place

9
Super Applications Key Requirements

• Built from heterogeneous software components
• including legacy systems/applications, objects,
  agents
• components built by different people for
  different tasks over long periods of time
• Must operate continuously, with high reliability,
  in dynamic environments
• requirements always changing
• system (re)configuration must be done quickly,
  with minimal programmer effort - ideally at
  runtime
• Must interact with human users
• humans need to understand and influence the
  operation
• Must play nice with other super applications
• share resources, avoid deadlock, etc.

10
Some Challenges in Using Agents to Build Super
Applications

• How do agents from different agent architectures
  interoperate?
• e.g., in CoABS there are agents from RETSINA,
  OAA, TEAMCORE, etc.
• they have different kinds of agents and different
  control strategies and agent communication
  languages
• what interoperability mechanisms are needed for
  agent communication, etc?
• standards, protocols, services
• How do non-agent components play?
• objects, legacy applications, etc.
• e.g., via wrappers, proxies, etc.

11
Sources of Requirements for the Grid

• Technology Integration Experiments (TIEs)
• Integration/OOTW TIEs (initially NEO domain)
• 1-2 dozen agents from different agent
  architectures
• manually assembled for initial demo
• custom-built interoperability agents (e.g.
  RETSINA-OAA)
• Grid will simplify above by providing more
  general interoperability services and reducing
  the services that an agent developer must build
• Scalability TIEs (agent control, mobility, etc.)
• Grid will provide testbed to host a set of agents
  and collect data on those agents for hypothesis
  testing via Logging, Visualization, and other
  services
• Applications
• Other user application domains will supply
  requirements for Grid-enabled super applications

12
Agent Technology Enables Super Apps

• Components that are agents (or have been wrapped
  as agents) can assemble themselves into super
  applications dynamic configurations or teams
  tailored to the problem/situation
• components can discover one another at runtime
• agents can declare their capabilities functions
  performed, interfaces, languages, etc. (These
  capabilities and interfaces can adapt.)
• middle agents can match needs to capabilities and
  provide brokering, facilitation, translation,
  etc.
• components can establish interoperability in
  order to cooperate on the task at hand (exchange
  knowledge and provide services)
• agents can communicate using shared languages,
  ontologies, protocols.
• agent can negotiate subtasking, resources,
  communication protocols
• the organization/configuration and behavior of
  the team can change if the task, situation, or
  computing environment changes
• agents can represent and reason about the goals
  and beliefs of other agents and users. They can
  self-organize into teams that have team goals.
• agents can detect changes and adapt their
  behavior
• agents can work offline, and some have mobility

13
The CoABS Agent Grid is

• An infrastructure (or meta-architecture) that
  supports interoperability among agents from
  heterogeneous architectures
• collection of standards, protocols, services,
  libraries, wrappers, and low-level infrastructure
• augments, but does not replace, services within
  particular infrastructures the Grid is not
  another reference architecture (i.e.,
  heterogeneity is embraced)
• e.g., The Internet provides interoperability (via
  gateways, services, and standards) between
  heterogeneous computer networks.
• A dynamic collection of agents using this
  infrastructure
• e.g., The Internet is both the plumbing and what
  uses it

14
A View of the Grid
15
Super Application Self-assembly
Reconfiguration

• Capabilities
• Component discovery
• Component interoperability (with semantics)
• Component adaptability
• Teams of diverse, distributed components
• Services
• Metadata directory services (white/yellow pages)
• Ontology services (capability/need advertisement)
• Facilitation services (find/recruit components)
• Comm mechanisms/infrastructure (messaging)
• Translation services (interoperability)

16
Smooth-running, Efficient Super
Applications

• Capabilities
• Efficiency
• Adaptability
• Reliability and Security
• Understandability and Taskability
• Services
• Team coordination services
• Mobility services
• Exception management, component lifecycle
  management, security services
• Logging and event services
• Visualization services

17
Easy to Build/Maintain Super
Applications

• Capabilities
• Programmability
• Customizability
• Testability
• Services
• Infrastructure services and adapters
• Exception mgmt., translation, mobility, component
  lifecycle
• Logging, event, visualization, simulation,
  debugging
• Policy and protocol management services
• Grid mgmt. services (start, monitor, manage,
  maintain Grid services and infrastructure)

18
Architectural View of the Agent Grid

• Grid concept implies
• A set of connected resources
• Advanced capabilities for integrating them
• Agents services such as
• Infrastructure (Jini discovery join, messaging,
  )
• Metadata directory (naming, registration,)
• Translation
• Facilitation (matchmakers, brokers,
  facilitators,)
• Team management
• Lifecycle and Grid management
• Mobility
• Ontology management
• Security
• Logging event management
• Visualization

19
Architecture Principle separation of
concerns deconstructionist view - what can you
take away and still have an agent system
Agent Reference Architecture
http//www.objs.com/agility/tech-reports/9808-agen
t-ref-arch-draft3.ppt

• policy, management
• resource dial

ALP, HLA, IA
GRID
federates

• AGENT SYSTEM
• single Vs. multi-agent

• heterogeneous
• computing environ.
• agent systems
• ACLs
• content languages
• ontologies
• policies
• services
• open world assumption

systemic grid features
common services

• ensembles
• of agents
• teams, peers, contracting,
• org. responsibility
• roles, capabilities,
• mutual beliefs
• hierarchy
• conversational policies

• societies
• closed vs.. open, communities of interest

• agent properties kinds
• communication capability
• computation capability
• by role in system
• information agent
• data sources
• interface agent
• NLI, multimodal
• coop response
• task agent
• web/email agent
• middleware agent
• mobile agent, itinerary
• social, personality, motivation, forgetting
• intelligent agent

distribution messaging svcs agent life cycle -
start, stop, checkpoint, name service event
monitoring leasing, compensation catalog
services, registry/repository
register, offer/accept/decline publish,
subscribe trading, matchmaking,
advertising, negotiating, brokering,
yellow pages security authenticate
encrypt access control lists firewall
CIA model agent suspects transactions persistence
query, profile (of metadata) data
fusion replication groups
multicast (scarce) resource mgmt, allocate,
deallocate, monitor, local, global
optimization, load balancing, negotiation for
resources scheduling time, geo-location rules,
constraints planning property list versioning,
config
autonomous decentralized

• control, coordination,
• multi-agent synchronization
• cooperation, competition

I3 BADD AICE
OMG JTF Jini
scalability
adaptation, evolution via market model, ...
licensing cost

• ONTOLOGY
• Ontolingua, OKBC
• metadata representations
• interests, locations, availability, capability,
  price/cost
• XML and web object models

mobility
secure, trust
IA
speech acts ACL - KQML, FIPA ACL, OAA ICL
survivability

• infrastructure
• primitives
• reflection
• serialization
• threads
• interceptors
• proxies
• filters
• multicast
• wrappers
• legacy sys
• data sources

• planning
• reactive
• goal interactions
• discrete vs. continuous
• constraints
• iterative, revision
• workflow

evolvability
EDCS

• missing
• views
• MOP

reliable

• QoS
• accuracy
• priorities

Quorum

• learning
• by example
• ...

More common services instrumenting,
logging caching queuing routing,
rerouting pedigree, drill down translation ...
xxx Agility addresses these Architecture WG
in Pittsburg Control WG in Pittsburg
Interoperability WG in Pittsburg red Sun Jini
green other DARPA programs
time-constrained

• content languages
• KIF, FOL, IDL, RDF

DDB
http//www.objs.com/agility/tech-reports/9810-agen
t-comparison.html http//www.objs.com/agility/tech
-reports/9809-best-of-class-capabilities.htm
20
Migration foils goes here
21
Grid Services

• Provide Grid-wide (global) functionality
• Augment services provided by local agent
  architectures
• Grid limited in its visibility control of
  indiv. agents
• Need an architecture/infrastructure that allows
  plugging in of services easily
• various service access mechanisms (ACLs, APIs,
  etc.)
• Distinction between services in and on Grid is
  blurry
• e.g., route planning agent could be used by
  military planning agent or by Grid mobility
  service
• Goal is to encourage market for Grid services
• standard interfaces with multiple implementations
• Leverage existing CoABS/external technology

22
Grid Operation Services

• Infrastructure Services (for interconnectivity)
• message delivery, bandwidth adaptability, etc.
• leverage Internet, CORBA, HTTP, Java RMI, JINI,
  etc.
• Grid Management (for administration of Grid)
• enable human/agent control of Grid services,
  resources
• leverage visualization services, policy
  services, etc.

23
Component Interoperability Services

• Metadata Directory (for component discovery)
• white pages, yellow pages
• leverage lightweight vocabularies, XML, RDF,
  LDAP, JINI, CORBA Naming/Trader, OODB/DBMS, etc.
• Ontology Management (compon. discovery, interop.)
• store provide access to ontologies,
  inferencing
• leverage KB work, XML, etc.
• Translation (for interoperability)
• between agent comm languages (primitives,
  content, ontologies)
• Facilitation (for interoperability, task
  achievement)
• matchmakers, brokers, etc.
• leverage intelligent matching, blackboards, etc.

24
Super Application Operation Services
(1)

• Lifecycle Management (for component mgmt.)
• agent/component instantiation/birth, death,
  cloning, status
• Mobility (for dynamic adaptability)
• agents move while running to other computing
  nodes via docks, itineraries, etc.
• leverage Java, security services
• Security (for resource protection)
• user/agent authentication, access ctrl,
  encryption, etc.
• leverage DARPA IA/IS work, CORBA/Java Security
• Team Coordination (for teams to achieve tasks)
• dynamic team formation and tasking, team
  monitoring
• leverage team-oriented programming

25
Super Application Operation Services
(2)

• Security (for resource protection)
• user/agent authentication, access ctrl,
  encryption, etc.
• leverage DARPA IA/IS work, CORBA/Java Security,
  etc
• Exception Mgmt. (for fault tolerance/recovery)
• detect and handle common agent exceptions
• leverage ontology of exception types
  strategies
• Logging/Event Mgmt. (for debugging and analysis)
• capture agent activities, messages, etc. - can
  mine this data
• enable sharing of common events (via triggers,
  etc.)
• leverage XML, CORBA Events, JINI Events, etc.

26
Super Application Operation Services (3)

• Visualization (for human understanding/control)
• show agent activities/messages (for debugging,
  etc.)
• show problem-solving activity/results (for
  end-users)
• leverage GUI technology (2D/3D, VR), HTML, etc.
• Policy/Protocol Mgmt. (for customization)
• library of policies and protocols for security,
  comm, etc.
• support building super applications on the Grid
  tailored to particular domains/tasks
• supports user admin of Grid and dynamic selection
  negotiation on protocols by agents at runtime
• leverage protocol/policy representation work
• Testing/Debugging
• includes instrumentation, simulation,
  visualization, etc.

27
Directions

• Increased development of standards (technical and
  domain)
• Further integration with Internet/Web
  technologies, e.g.,
• Use of XML in agent technologies
• Internet as agent communication mechanism
• Further integration with components / service
  architectures, e.g.,
• Increasing use of Java technology (e.g., Jini)
• Agent Grid (CoABS) agentized Object Services
  Architecture

28
Layered Grid Perspectives

• Layered functional grids
• Information, Sensor, and Engagement grids
• e.g., ABIS, Network Centric Warfare
• Layered technical grids
• Computational grid
• Data grid
• Object grid
• adds behavior to data grid and links data and
  links them
• Object enhancements to the Web an illustration
• Agent grid (object grid plus smarter objects)

29
Layered Technical Grids

• Each technical grid layer provides advanced
  compositional mechanisms for things at that
  level, e.g.
• A computational grid allows formation of larger
  virtual computers from combinations of
  physical computers
• A data grid allows formation of federated data
  collections from combinations of existing data
• An agent grid allows formation of new agents
  (teams) fromcombinations of existing agents
• Ideally, the compositional mechanisms will
  supporta closure property
• The resulting compositions can be treated as
  individualentities at that level for further
  composition

30
Agent/Grid Architecture Issues

• What are agents? - code and data packets that are
  autonomous, adaptive, cooperative, mobile,
  interoperable We want all these properties in
  future agent-based systems. We need experience
  building systems with these properties.
• Pervasiveness - How do we insure that the
  architecture stays lite-weight for wide-spread
  adoption.
• Embracing heterogeneity - We must piggyback agent
  systems on already pervasive infrastructure like
  ORBs, the Web, email, and DBMS systems. We must
  identify the specific kinds of heterogeneity we
  want agent system architectures to support.
• Separation of concerns
• agent-agent separation - can agents access each
  others state directly
• agent-service separation - do agents implement
  the long list of services that the grid provides
  or is that done via underlying component-based
  middleware?
• grid-agent separation - agents are autonomous but
  they cooperate and compete for resources within
  the software grid. The grid provides some global
  systemic properties and some basic shared
  services. Is there an explicit grid or is it
  implicit in the way agents interact with each
  other? Are some services (like planning)
  optionally distributed into agents or are they
  available from the grids planing service? Can
  new services be autoloaded into a grid that does
  not have them?
• Semantic interoperability, ontology - do
  ontologies scale? How do they extend class
  libraries?
• Licensing - Agents, data sources, and component
  software need an economic model so broad
  communities can get value from them. A model of
  licensing might be critical to success in the
  large.
• Agent communication language (ACL) - Is the ACL
  compositional and extensible so one can define
  new speech acts from existing ones? How many
  speech acts is enough? 20 or 5000?
• Control points - where are the control points
  where different control algorithms might be
  substituted into the architecture
• Grid federation issues - How are software grids
  federated - flat versus hierarchical models? If
  different grids contain different policy choices
  or different services, how does that affect
  agents communicating across grid boundaries? Can
  we add new services and -ilities to a grid once
  it is deployed? how transparent is addition or
  subtraction of services and ilities
• Coordination - Insure Agent Reference
  Architecture augments DARPA ISO ATAIS
  architecture. Provide template for next
  generation unified OMG, FIPA, and W3C agent
  standards. Insure that reference implementations
  (toolkits) exist and are widely available.

31
Some Referencessee http//www.objs.com/reports.ht
ml

• Characterizing the Agent Grid
• http//www.objs.com/agility/tech-reports/990623-ch
  aracterizing-the-agent-grid.html
• Systemic Properties
• http//www.objs.com/aits/9901-iquos.html

32
What is an Agent? deconstructionist view agents
augment objects with additional capabilities
Object ? Component ? Agent
? ?

• ACL
• process inside
• agent framework
• planning
• mobility
• rules
• goal/task-oriented
• autonomous
• ontologies
• collaborative/teams

• state
• behavior
• encapsulation
• inheritance

• reflection
• packaging
• serialization
• repository

• TBD

33
What is Interoperability?

• Definition the ability of two or more systems
  or components to exchange information and to use
  the information that has been exchanged IEEE
• Information includes anything exchanged, e.g.
• Data (control or domain-related)
• Operation invocations on objects
• Error notifications
• Interoperability is based on various agreements
  (shared assumptions) among the interacting
  objects about the information exchanged
• Disagreements may limit the possible
  interoperability,(partial interoperability is
  possible) or deny it entirely