Draft EU Project Proposal

EU IST Reference:   http://www.cordis.lu/ist/workprogramme.htm
                                CPA4 – Towards dependable and survivable systems and infrastructures

Project acronym:     Safeguards

Project title:   An Intelligent Agents Organization to Enhance Dependability and Survivability of
                   Large Complex Critical Infrastructures

Document status:Technical note, confidentional , draft 25/01/01


1. Problem Context

Nowadays, the surviving and proper functioning of global industrial and information infrastructures are basic requirements for their users and managers. Modern industrial infrastructures are heterogenious distributed systems. They include different levels of automation, regulation, and control, but, for the safety reason, so called "intelligent" functions are related to their humans operators.

In general, the functioning of Large Complex Critical Infrastructures, LCCI (such as, electricity, gas, telecommunication, transport networks) are supported by, always more and more complex, different networks of computer systems.

The management of survivability of these jointed networked systems requires:
  1. distributed decisional support for their human managers and operators
  2. integrity supervision: capacity of intrusion detection and autorepairing capabilities
  3. capacity of local autonomous interventions and task execution
  4. capacity of the development of modifications and self-adaptation.
The all above tasks should be realizable by spatially distributed organizations of autonomous specialized supervision and intervention units. The application of the human organizations metaphor and generic models of immunological mechanisms of living organisms and epidemiological defence of human organizations, should provide general patterns and ontologiesforthe modeling of an organization of software intelligent agents focused on the supporting of the LCCI functionning.
References: IDIFENSE:http://www.idefense.com/pages/ialertexcl/MidEast_010501.htm
MAFTIA: Malicious- and Accidental-Fault Tolerance for Internet Applications: 
SIRENE:Security in Computer Networks: http://www.semper.org/sirene/
Computer Securrity Institute:http://www.gocsi.com/

2. Problem definition

An intelligent agents organizations composed of tightly networked embedded autonomous software units should enhance the dependability and survivability of LCCIs.
In this proposal we are focused on the generic, task-independent cognitive intelligence of such agents and on the architecture of their role-dependent organizations.
As a test case, an organization of Safeguards(similar to the idea of softbots, http://www.botspot.com/) with the role of an internal system supervisoryincluding an auto-defence againstdifferent intrusions and hackers attacks,is proposed. Here, the main goal is to avoid unexpected widespread network failures and infrastructural disasters. Safeguards should learn role (= competencies+ responsibility+ duties+ data access), accept tasks and evolve and adapt to the changing environment improving, in parallel, their acting capacities.


References:Malicious Mobile Agents:http://www.rdg.ac.uk/~ssr96apg/research.html,
Malicious Mobile Software Agents: A Categorisation of Defence Initiatives:
Information Security, IBM:
SAMSON: Security and Management Services in Open Networks Project:
Intrusion Dedection:

3. Project Objectives  

The project proposal is focused on two main research objectives:
  1. Analysis and development of an architecture and functionality models of a Safeguards as a cooperating intelligent cognitive agents.
  2. Analysis and development of a generic model of Safeguards-agency (a hierarchical organization of safeguards).
  3.      Validation of the generic, domain-independent  solutions by incremental prototyping in a preselected test domain.
The primary objective requires development of semi–intelligent reasoning, goal-modification and learning structures with sensing, communication and intervention possibilities adequate to their environment. Safeguards knowledge should include such rules and procedures as various access controls, recognition of loss of data/information, encryptionkeys checking, diagnostics ofloss of functionality, intrusions and damage propagation, mitigation of losses and protection of integration of the LCCI components.
The second objective also requires the development ofthe agent interactions and communication 
For the both objectives an identification of software platform is necessary. Its choice will depend on the individual experience and preferences of the project partners. Here we should mention that the development of a full-scale multi-agent software infrastructure using one of available MAS (Multi Agent System) technologiesis notthe objective of this proposal. 
We intend to simulate and validate the developed ontologies, models and methods using a relatively simple but incremental demonstrative software. 
A conceptualarchitecture of cognitive models interfacing decision-making models with cooperation
strategies and proper design methodologies of multi-task interactions with the intervention domains,Safeguards-agency, and safeguards environment, are considered as the main item of this proposal. 
The project results will be a mindware (computer implementable methods, models and methodologies) and their benchmark based validation.

Summarizing,the results should provide a contribution tothe foundations ofstandards of the safeguards design, organization, and their mission in Large Complex Critical Infrastructures.

Main Key RTD Aspects
The main research aspects of the project are focused on the following domains:
1.Modeling of Intelligent Cognitive Agents
2.Software Communication Platform for resident and mobile agents
3.Ontologies for Malicious Agents and Intrusions-Detection
4.Risk assessment, risk management and Defense Strategies 
5.Artificial life 
The range and contributions of these research lines should be assessedin the context of the test-case provided by the industrial partner. 

Project Partners’ Profiles

The Project Safeguards has strongly cross-disciplinary character: it requires at least four types of partners:
  1. Domain experts, practitioners which will provide Information, Preferences and Knowledge (IPK) relative to the description and functioning of LCCIs, and benchmark building.
  2. Modellers and Knowledge Engineers with the tasks of knowledge acquisition,hierarchical and flat ontologies development, IPK formalization, and the validation of incremental prototypes.
  3. Agent Software Developers with the implementation task, knowing agent–oriented software engineering tools.
  4. Managers for the Project Management and the Results Exploitation and Dissemination. 
The following specialist contributions are requested: interdisciplinary system researchers, cognitivists, knowledge and software engineers, network experts. At a consequence, the partners should represent industry, software companies and academic and research centers.

Provisional Workpackages

The workpackages of theSafeguards Project include:
Provisional Consortium
ENEL –domain expert, end user
ENEA – modeller and knowledge engineering: 
ECONA -cognitive modelling and human-like factors: http://studens.psi.uniroma1.it/


Main Contact Person

Dr. Adam M.Gadomski

ENEA CR- Casaccia 

Via Anguillarese, 301
00060 Roma
Phone: +39-06-3048 3404
Fax: +39-06-3048 6511

Research Line:

Informed Person

Dr. Sandro Bologna
ENEA CR-Casaccia
Via Anguillarese, 301
00060 Roma
Phone: +39-06-30483708
Fax: +39-06-30486511

Research Line: http://erg4146.casaccia.enea.it/SANDRO/index.html