Using Datalog to Describe Simulation Capabilities

Size: px

Start display at page:

Download "Using Datalog to Describe Simulation Capabilities"

Kerry Harrell
6 years ago
Views:

1 U.S. Army Research, Development and Engineering Command Executable Scenario Definition Using Datalog to Describe Simulation Capabilities US Army Research Laboratory (ARL) Human Research and Engineering g Directorate (HRED) Simulation & Training Technology Center (STTC)

Next Generation Architectures (NGA) for ing &Simulation The Next Generation Architectures (NGA) for ing & Simulation (M&S) research project refines and demonstrates advances in computer science that

2 Next Generation Architectures (NGA) for ing &Simulation The Next Generation Architectures (NGA) for ing & Simulation (M&S) research project refines and demonstrates advances in computer science that support the development of M&S architectures required for future training, experimentation and acquisition decisions This presentation ti discusses an initial iti effort under this project area to identify scenario primitives that can link analytical data elements to simulation execution objects in a way that supports automated model and simulation selection based on the content of the scenario *Unclassified For Public Release* 2

3 The Situation Simulation models with fixed interfaces Explicit it integration ti Failing events due to models and simulation not meeting expectations *Unclassified For Public Release* 3

4 What we have I provide Stress Levels I provide Munitions I need Stress Levels and Munition ti Detonation Simulation models with fixed interfaces *Unclassified For Public Release* 4

5 What we have I provide Stress Levels I provide Munitions I ll take Munitions And request Stress Levels Detonation Explicit integration *Unclassified For Public Release* 5

6 What we have I provide Stress Levels I provide Munitions I ll take Munitions And request Stress Levels Detonation Invalid Stress Levels Detonation Event(s) Failing events *Unclassified For Public Release* 6

7 What we want Fluid, discoverable simulation models sandboxing Integration via composition *Unclassified For Public Release* 7

8 What we want I need Stress Levels and Munitions ti Fluid, discoverable simulation models *Unclassified For Public Release* 8

9 What we want OK. I ll Ill check I need Stress Levels and Munitions ti Broker Fluid, discoverable simulation models *Unclassified For Public Release* 9

10 What we want Anyone provide Stress and ti? Broker Detonation Fluid, discoverable simulation models *Unclassified For Public Release* 10

11 What we want I provide Stress Levels I provide Munitions Broker Detonation Fluid, discoverable simulation models *Unclassified For Public Release* 11

12 What we want Do these Work for You? Broker Detonation Fluid, discoverable simulation models *Unclassified For Public Release* 12

13 What we want I don t know. I need a finer-grained Stress Level Broker Detonation Fluid, discoverable simulation models *Unclassified For Public Release* 13

14 What we want OK. Why don t you send me a few tests Broker Detonation sandboxing *Unclassified For Public Release* 14

15 What we want Broker Test detonations Stress requests Detonation sandboxing *Unclassified For Public Release* 15

16 What we want Stress requests Test detonations Stress levels Broker Stress levels Detonation sandboxing *Unclassified For Public Release* 16

17 What we want That doesn t look right. I need casualty stress too. Broker Detonation sandboxing *Unclassified For Public Release* 17

18 What we want I know about another one, give me a second Broker Detonation Integration via mix-ins *Unclassified For Public Release* 18

19 What we want Start Casualties Broker Detonation Integration via mix-ins *Unclassified For Public Release* 19

20 What we want Yes Can you two compose*? *Always a tricky question! Yes Casualty Broker Detonation Integration via mix-ins *Unclassified For Public Release* 20

21 What we want Well, try this out. Test detonations Casualty Stress requests Stress levels Broker Stress levels Detonation sandboxing *Unclassified For Public Release* 21

22 What we want How did that look? Casualty Broker Detonation Integration via mix-ins *Unclassified For Public Release* 22

23 What we want Looks great! Let s begin. Casualty Broker Detonation Integration via mix-ins *Unclassified For Public Release* 23

24 What we want Stress requests Casualty Detonation Event(s) Stress levels Broker Stress levels Detonation Successful events *Unclassified For Public Release* 24

25 What we want Fluid, discoverable simulation models s need to be able to describe what they think that they can do sandboxing Consumers and models need a way to test their interactions in a safe way to determine viability Integration via composition composability needs to be expressible *Unclassified For Public Release* 25

26 We need a primitive language To express wants and needs To express capabilities To express composition of capabilities That can host existing systems *Unclassified For Public Release* 26

27 We need a primitive primitive language Simple Syntactically ti composable Syntactically inferential Simple solution: Datalog *Unclassified For Public Release* 27

28 Datalog Hi, I m a tuple! [<entity> <attribute> <value>] *Unclassified For Public Release* 28

29 Datalog The entity ypart is a referent (think identity) [<entity> <attribute> <value>] *Unclassified For Public Release* 29

30 Datalog The attribute part is a tagged property [<entity> <attribute> <value>] *Unclassified For Public Release* 30

31 Datalog The value part is a property s value for an entity [<entity> <attribute> <value>] *Unclassified For Public Release* 31

32 Datalog capabilities [ :provides Stress] *Unclassified For Public Release* 32

33 Datalog discoverability [ :provides Stress] [User :needs Stress] *Unclassified For Public Release* 33

34 Datalog composability [ :provides Stress] [ :stress/component ] + Casualty [Casualty :provides Stress] [Casualty :stress/component Casualty ] *Unclassified For Public Release* 34

35 Datalog Allows inference and querying Hosts primitive iti model AND Object s Simple Self-describing / self contained *Unclassified For Public Release* 35

36 Proof-of-Concept Experiment Executable scenarios This is a Scenario (has desired capabilities ) Simulation Functions (has provided capabilities ) Win! *Unclassified For Public Release* 36

37 Petri-dish scenario Red and Blue start with an obstruction between them. They move forward until they see each other and engage. *Unclassified For Public Release* 37

38 System overview *Unclassified For Public Release* 38

39 Capabilities and primitives Capability Primitives Notes Run-time infrastructure n/a This is a special category of capability Plan View Display n/a This is a special category of capability Creation Time, Location, Heading, Units, Platforms Side, Orientation Movement Latitude, Longitude, Velocity, Orientation Damage Health Fires Weapons Control Vision Sensation, Orientation *Unclassified For Public Release* 39

40 Broker architecture *Unclassified For Public Release* 40

41 Broker architecture Persistent Peer architecture Inferential database Browser-based user interface Hyper Text Transfer Protocol (HTTP)-based system-to-system interface Messages containing collections of tuples I.E. [<entity> <attribute> <value>] *Unclassified For Public Release* 41

42 Peer architecture EASE: Executable Architecture Systems Engineering contact authors for more information *Unclassified For Public Release* 42

43 Peer architecture Persistent connections Capabilities declaration Launch logic *Unclassified For Public Release* 43

44 Lessons learned We can use the Broker/Peer technologies now, and in the future Coordinator (persistent connections, launch) EASE (meta-model model and inference) We need to think harder about simulation primitives We need to talk to some analysts We need to explore sandboxing in depth We need to explore auto-extraction of desired capabilities from scenarios *Unclassified For Public Release* 44

45 Authors Chris Gaughan Christopher J. Metevier Simulation and Training Technology Center, Human Research and dengineering i Directorate, t Army Research Laboratory, Orlando, Florida Mike Fogus Joseph S. McDonnell, Ph.D. presenter joe.mcdonnell@d-a-s.com Dynamic Animation Systems, Inc. Fairfax, Virginia Scott Gallant Effective Applications Corp., Orlando, Florida *Unclassified For Public Release* 45

46 Questions/Comments? ARL-HRED-STTC Public Affairs Office (407) *Unclassified For Public Release* 46

47 Additional Reference *Unclassified For Public Release* 47

Introduction to Information Systems

Introduction to Information Systems Table of Contents 1... 2 1.1 Introduction... 2 1.2 Architecture of Information systems... 2 1.3 Classification of Data Models... 4 1.4 Relational Data Model (Overview)... 8 1.5 Conclusion... 12 1 1.1 Introduction