1. Universal Top SYNTEX Functional Architecture Building Working Document Adam Maria Gadomski, Vittorio Rosato ENEA    A Contribution to the SYNTEX Development for the IRRIIS WP 1.3 + WP 1.4 Meeting, Sankt Augustin, l7 May.06 ( a proposal for discussion) Wp. 1.3 IRRIIS Project Document, Wp. 1.317 May 2006

2. Universal Top SYNTEX Functional Architecture Building © Adam M.Gadomski at al., ENEA, 15 May 2006  Contents 1.Methodological Base 2.Top-down Problem Decopmposition 3.Tools-Agents 4.SYNTEX Core Scenario 5.Universal Top SYNTEX Functional Architecture 6.Summary of SYNTEX Agents 7.Description of Interpretation 8. Final Remarks    

3. Universal Top SYNTEX Functional Architecture © Adam M.Gadomski at al., ENEA, 15 May 2006  Methodological Base We assume the TOGA (Top-down Object-based Goal- oriented Approach) based universal approach to the specification of the Functional Architecture of SYNTEX. The TOGA methodology starts from the recognition in a problem the following top elements: Intelligent Agent (IA) its Domain of Activity connected with IA by Interaction relation their Environment, what is illustrated on the fig.1. Intel. Agent Domains of Activity Environment Interaction Fig.1. An elementary Intelligent Agent’s World (IAW). Its basic couple is graphically represented as: In the SYNTEX system a main IA is its human user. The details of the domain of activity have to be recognized yet. Its functions and configuration depends on the SYNTEXT-User Goal. User/ Users SYNTEXT System SYNTEXT System Environment Interaction LCCI operators LCCI System/(s) LCCI System/(s) Simulated Environment Interaction Fig.2 : Top view on SYNTEX-User World Fig.3 The kernel of SYNTEX

4. Universal Top SYNTEX Functional Architecture © Adam M.Gadomski, ENEA, 15 May 2006  Goal-oriented Approach SYNTEXT-User Goal The important goal of the SYNTEXT-User couple is a demonstration of the preselected scenarios of an Emergency Management (EM) and enabling their modifications in such way that some previously accepted indicators of Efficacy and Quality of EM (EEM, QEM) could be improved. EEM and QEM are referred to the GEM (Goal of EM), and can also be called Quality of Service (QoS). Let us assume that the Goal of EM is to minimize total human, economical and environmental losses during the emergency. EEM and QEM depends on the specific properties {p} of the Scenario of EM, and their values are assessed by the user (as a human expert). The evaluation of QoS is done by human experts. SYNTEX Goal - The main goal of SYNTEXT is to simulate a certain large class of emergency events and to enable their interpretations by human experts. - The simulation results have to be usable by the MIT system (according to the MIT goals). The interpretation of the result has to provide data for the recognition of a Efficacy of the Emergency Management (EEM), its Quality (QEM) and modification of initial state of the simulated behavior of the Critical Infrastructure under analysis.

5. Universal Top SYNTEX Functional Architecture © Adam M.Gadomski at al., ENEA, 15 May 2006  The problem decomposition is composed of the decomposition of Generic Intelligent Agent World (IAW) which relies on the identification of its topology, i.e. invariant components/objects and interactions/interrelations. In parallel, the dynamics of the interaction is presented in form of a generic scenario of the SYNTEX application. This scenario is decomposed/specialised according to the requirements resulting from the Goal of SYNTEXT-User system and, the goal of the SYNTEX simulator results from it. In consequence, we have Initial Top SYNTEX Application Scenario (fig.4) and the fig.2 can be specialized as presented on the fig.5.  Preparatory activity Simulated event Interpretation activity Attention: Every next decomposition step requires a consensus on the previous one. Fig. 4 First top-scenario executed by the couple: aSYNTEX manager agent (SM) and the human User of SYNTEX. User/ Users SYNTEXT System Environment Interaction SM SYNTEX Functional Domain Fig. 5 User interacts directly with SM. Top-down Problem Decopmposition

6. TOGA based Universal Top SYNTEX Functional Architecture © Adam M.Gadomski, ENEA, 15 May 2006 Fig. 6 First decomposition of the Domain of Activity of the SYNTEXT Tasks Manager agent. It receives tasks from h-user and presents information on request. H-user Active Simulator A-Tools and interpretation Domain Data/ Information Simulation Model components Autonomous SYNTEX management tools Interaction: tasks, information SYNTEX Manager SYNTEX System Interaction: tasks, information Scenario is a time sequences of events and tasks/actions connected by consequence relations. Events are distinguished arbitrarily selected sequences of changes in the Domain of Activity of an Intelligent Agent or Agent. Event specification is a decomposition of an event using events and actions. Actions are specifications of changes caused by Agent/(Intelligent Agent) in its Domain of activity Task is a specification of the request from IA to an agent which can be frequently realized by different actions. Definitions Agent – a functional unit which realize actions according to obtained tasks and possessed: domain information. knowledge (algorithms) and domain preferences.(IPK frame).

7. ENEA’s Contribution to the SYNTEX Development © Adam M.Gadomski,at al. ENEA, 15 May 2006 Actions require : action algorithms, tools, data/materials and domains which are subjects of modification/transformation/processing, more formally: Action (agent, algorithm, tools, data, domains) Definitions System Specialization Specific Case Setup & Initial Date Insertion Simulation Event 1 34 ComputationaI Interpretation of the Results: What-If Expert Interpretation of the Results: What-If 56     Link to MIT Scenario Specialization 2  Fig. 7 Second decomposition level scenario User + SM User Decomposition of the (SYNTEX – User) Interaction in the form of a scenario Next specialization step requires an introduction of new concepts and their univocal definitions. Carriers: Agents - managers

8. Universal Top SYNTEX Functional Architecture Building © Adam M.Gadomski at al. i, ENEA, 15 May 2006  Fig. 8 Generic Agent Activity Frame Agent performs actions: Agent Domains of Activity Results Domain Task Domains of Data Domains of Tools Action (agent, algorithm, tools, data, act-domain) Tools in SYNTEX: We have 3 principal types of tools: 1.Actors tools (a-tool), into the simulator: tools of intelligent agents which are actors in the Simulator 2.SYNTEX management tools/agents( s-agents): autonomous tools/agents. They execute tasks of the User and SM. 3.User interpretation tools (i-tools): programs which transform simulation results to the forms most significative and useful for the interpretation of the final state of the variables represented the simulated domain. Action algorithms: Action algorithms are specification how tasks can be executed. In the case of simple Agents, action algorithms are pre-prepared. In case of IA they can be also elaborated on the base of his IPK ( information, preferences, knowledge) during decisional-processes. TOOLS - AGENTS

9. Contribution to the SYNTEX Development SYNTEX Core Scenario: Generic Top-Down decomposition and incremental development framework Fig. 8 Event/Action Module (EAM) Fig. 9 Event/Action Scenario (EAS) EAM1 EAM2 EAMn … EAMi EAMm … © Adam M.Gadomski at al., ENEA, 15 May 2006 F a, F b, … are system functions necessary for the realization of an event. They depend on the chosen LCCI or their network. T a, T b, … are actors tools for realization of an intervention/action, they depend on the chosen LCCI or their network Domain Event FaFa FbFb Data Setup Results Visualization … … Action TaTa TbTb Data Setup Results Visualization … … In the top-down approach, a scenario is initially represented by one most general EAM, after, this EAM is successively decomposed.

10. Universal Top SYNTEX Functional Architecture (TOGA based proposal) Functions activated in the Event-Action Scenario (EAS) A function & tool under development Tools used in an action in EAS … Agent An autonomous SYNTEXT management tool Org. Layer OL Cyber Layer CL Phys. Layer PL Org. Layer OL Cyber Layer CL Phys. Layer PL I -tool : a tool for the interpretation of SYNTEXT simulation results. SYNTEX User-Tasks Manager (agent) & Cognitive Interfaces MIT Interface Simulation Agent A-Tool Spec. Agent Functions Base Core Scenario Simulator Function Factory Event Factory A-Tools Factory OLCL PL Configuration Agent F. Specializati on Agent E.Specialization Agent Interpretation- Tools Factory I-Tool Application Agent OL CLPL  © Adam M.Gadomski at al., ENEA, 15 May 2006 Fig. 10

11. Universal Top SYNTEX Functional Architecture: AGENTS © Adam M.Gadomski, ENEA, 15 May 2006 SYNTEX Manager – it communicate with h-User and transform his tasks to the tasks for other agents. Configuration Agent – its task is to configure a Event-Action Scenario according to the h-operator/user indications/tasks Simulation Agent – It manages user requirements/tasks during simulation process. A-Tools Specialization Agent – it decomposes existing actor-tools and specialize them according user requirements/tasks. I-Tools – It manages application of instruments for the interpretation and new simulation planning. Functions Specialization Agent - it decomposes existing events in the Scenario and specialize them according user requirements/tasks. Event Specialization Agent - it decomposes existing events in the Scenario and specialize them according user requirements/tasks. … Other support agents are possible yet.  Fig. 11 Generic Frame of SYNTEX management agents/(tools) Basic Agents Support Agents SYNTEX Agent Domains of Activity Results Domains Task Domains of Data Summary of SYNTEX Agents

12. Universal Top SYNTEX Functional Architecture Building © Adam M.Gadomski at.al., ENEA, 15 May 2006  Let us assume that a complex infrastructure ( or infrastructure system)  is described by a number of the following technological measurable and/or observed reciprocally, independent attributes {x(t)} on t  [0, T].  Some of x i (t) can be constant.  Every time dependent x i (t) can be considered as a continuous or discreet function of time.  These attributes allow to describe the physical structure S, possible internal processes P and a  response on external changes in its environment, using a model M  : M  ⊢ A  : {x 0 }  {x T }. where: {x 0 } describe an a given  initial state, and {x T } denotes a result of the simulation using an algorithm A  obtained from the generic model M .  We assume that the accuracy of M  is sufficient for the simulation purposes/goals, it means, |{x T } -{x T } Real | < {k}, where set {k} results from the definition of the general goal G of the simulation. Or in the other form, in a normed state space, we may write: || M  -  || < K.  The goal GI of the infrastructure  is to provide a set of services for industrial systems and for the society in its Environment, under certain conditions. These service utility can be assessed according to their utility for the  Environment by a.generalized Quality of Service factor QoS. Description of the Interpretation

13. Universal Top SYNTEX Functional Architecture © Adam M.Gadomski at.al., ENEA, 15 May 2006 Description of the Interpretation  Let us introduce a given perturbation set {  } to the normal initial state of the  system (a network). In practice, it changes only a specific subset of {x 0 } but for the generality, we may write that the new state of  will be {x 0 (  )} = {x 0 } + {  }, and A  : {x 0 (  )}  {x T (  )}.  Let us assume that exists such {y} obtainable by known operators set  that  : [0, T].  {y} and  Quality of Service (QoS) be defined by a not formally known functional F such that: F: {y}  QoS and QoS  [0,1] for y n  [- ,+  ] and n=1,..N.  The functional F is weakly defined but using an implicit expert knowledge  Q we may assess QoS knowing {y} and assuming that:  Q  F, hence:  Q : {y}  QoS, for QoS  [0,1]. This mental transformation we call the interpretation of {y} and the set  we call i-tools. The interpretation of simulations enables identification and modification of assumed models, decisions & data.

14. Universal Top SYNTEX Functional Architecture Building © Adam M.Gadomski at.al., ENEA, 15 May 2006 Description of Interpretation The cyclic application of simulation sessions of SYNTEX, so called “what-if simulations”, enable to identify the vulnerability of infrastructure , and to reduce it. Starting from a given configuration: ( {x}, A  ) for every simulation cycle realized by the composed transformation  we obtain QoS, where  = .  A .  Assuming now that {  } describes an unexpected jeopardize modification of the previous initial state of , and we may change {x 0 } then we have:  j : {x 0 j ({  } )}  QoS j for j =1, 2,… In other words we may write: QoS j ({x 0 } j,{  }, A , ,  ).  The main improving robustness(*) procedure of  relies on the modification of {x 0 } j in such way that in every cycle:  QoS j+1 ({x 0 } j+1,{  },. ) - QoS j ({x 0 } j,{  },. ) =  j+1 > , where  is a reasonable difference assessed by human expert (User), and from the Expert Knowledge   we have :   : (  j+1, {x0} j+1 )  {x0} j+2 (*) Robustness is an indicator opposite to vulnerability, its antonym.

15. Universal Top SYNTEX Functional Architecture © Adam M.Gadomski at.al., ENEA, 15 May 2006 The Universal Top Functional Architecture of SYNTEX presented on the fig.10, enables to modify all attributes of QoS j i.e: {x 0 } j, {  }, A , , . This property should be also useful from the perspective of MIT development and testing. Final Remarks This is a Working Document for Discussion yet and the functional range of SYNTEX should be confronted with the realization possibilities dependent on the available software implementation platform and the requirements resulting from the MIT requirements and its functional specification. For the references see : IRRIIS Tech. Annex, 2005. TOGA Meta-theory:, http:// erg4146.casaccia.enea.it/