A DDS-based Distributed Simulation for Anti-air Missile Systems

Dohyung Kim

, Hyun-Shik Oh

and Seong Wook Hwang

Modeling and Simulation Technology Division, Agency for Defense Development,

Bugyuseong daero 488 beon gil, Yuseong, Daejeon, 305-152, Republic of Korea

M&S Department I, SIMNET Co., Ltd., 892-9 Jijok-dong, Yuseong, Daejeon, 305-330, Republic of Korea

Keywords: Air Defense Engagement Simulation, Data Distribution Service (DDS), AddSIM, AddSIM-DDS,

Component-based Distributed Simulation Environment, High-level Architecture (HLA).

Abstract: This paper introduces the development of a distributed air-defense engagement simulation model based on

data distribution service (DDS). To design and develop effectively, system developers need a high-resolution

engagement simulation including complex engineering-level models and operational scenario models.

Increasing the resolution of the model results in the growing model’s complexity which requires greater

resources than that of a single computer. We tried to build a distributed engagement model using AddSIM-

DDS which combines the Advanced distributed simulation environment (AddSIM) and DDS. We describe an

air-defense scenario, overall structure of the model, and simulation construction on distributed nodes. We also

define several DDS topic types for interoperation. Finally, we provide the results that show the validity and

effectiveness of the DDS-based distributed simulation.

1 INTRODUCTION

In the defense acquisition domain, modeling and

simulation is playing increasingly important roles in

entire acquisition life-cycle phases. Especially an

engagement simulation model gives a very useful tool

for system development engineers by helping to derive

detailed specification from high-level requirements. It

also supports system verification and validation

(V&V) through the model-based virtual experiments

which evaluate system performance and effectiveness.

For the effective development of a new anti-air

missile system, an air-defense engagement simulation is

required, which can cover engagement and engineering

levels in terms of the abstraction levels of defense

modeling and simulation (M&S). It should capture

scenarios including strategy and tactics at the

engagement-level which are related with functions such

as surveillance, threat evaluation and weapon allocation

(TEWA), firing and tracking, and flying formation and

evasion (Choi and Wijesekera, 2000). Likewise, it

should also consider the dynamics of individual weapon

systems at the engineering-level. Therefore it can be

developed using a hybrid modeling approach of

continuous-time and discrete event models.

For the analysis of the system developer’s

perspective, system developers require a higher

resolution engagement simulation model than a

conventional generic engagement model which stays

at the system level of granularity. So the hybrid model

for an air-defense engagement simulation should be

able to utilize high-resolution engineering-level

models directly. However, there is a problem that

increased model resolution results in the growing

complexity. For example, the missile model of the

air-defense engagement simulation may consist of

many different sub-system models which consist of

multiple component models again. Each component

model calculates its dynamics, update state variables,

and interacts with other models on the time-scale

varying from millisecond to microsecond.

Accordingly, it becomes more complex and difficult

to simulate than the engagement-level model alone.

On a single computer resource, it is difficult to run

the high-resolution engagement simulation model

which contains a number of high-resolution

engineering models. It may fail to meet user

requirements for execution time or it may not be

executed at all due to insufficient memory.

Addressing these issues, parallel and distributed

simulation method has been used to simulate large

and complex models (Fujimoto, 2000).

Some studies have focused on High Level

Architecture (HLA)-based interoperation of

270

Kim, D., Oh, H-S. and Hwang, S.

A DDS-based Distributed Simulation for Anti-air Missile Systems.

DOI: 10.5220/0005980402700276

In Proceedings of the 6th International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2016), pages 270-276

ISBN: 978-989-758-199-1

engineering models to build a high-fidelity

engagement simulation (Sung, Hong, and Kim, 2009;

Hong et al., 2011). HLA was developed to unify

various distributed simulation approaches and to

define a general purpose architecture for distributed

computer simulation systems. Now it has emerged as

a widely adopted middleware standard for

interoperation simulation. In the HLA-based

interoperation approaches, distributed engineering-

level models and discrete event models representing

such things as command and control (C2), join the

federation for the upper level engagement simulation.

In addition, there have been several studies related

HLA-based interoperation between high-resolution

models such as virtual simulators.

However, HLA was not designed to support the

low latency data sharing which is required as the

simulation complexity and scale increases (Andrew,

2014; Zheng et al., 2009) The HLA Runtime

Infrastructure (RTI), a software implementation of

the Interface Specification of HLA, are not enough

for massive distributed simulation federations with

frequently exchanging data (Lopez, 2011). In order

to address these challenges, AddSIM-DDS was

developed which is the engagement simulation

environment based on data distribution service (DDS)

for distributed systems (Kim et al., 2014). The Object

Management Group (OMG) DDS is a standard

specification for data-centric publish and subscribe

communications. DDS provides high performance

with low latency and various qualities of service

capabilities (OMG, 2007).

This paper introduces a DDS-based distributed

simulation approach for anti-air missile systems. We

made a high-resolution air-defense engagement

model which can be utilized in performance

prediction and evaluation of the missile system. We

developed the hybrid model for the systems involved

in the scenario and constructed a DDS-based

distributed simulation using AddSIM-DDS. Our

experiments and results show the validity and

effectiveness of the DDS-based distributed

simulation. Especially, we compared the results of a

distributed simulation based on two different types of

middleware-DDS and HLA/RTI.

2 BACKGROUND

In this study, we developed a DDS-based distributed

engagement model using AddSIM-DDS. Before

explaining the air-defense simulation model, this

section describes DDS, AddSIM, and AddSIM-DDS,

which are the background of our study.

2.1 Data Distribution Service

DDS is a functional specification to efficiently

delivery data across distributed systems in publish-

subscribe manner. OMG approved DDS as a

machine-to-machine middleware standard since 2003.

DDS aims to enable scalable, real-time, dependable,

high-performance and interoperable data exchanges

between applications. By these advantages, DDS has

widely used in military and commercial area.

To send and receive data, events, and commands

among the nodes, publisher nodes create "topics" and

publish the data. DDS delivers the data to subscribers

that declare an interest in that topic. The subscriber

catches and uses the data. With the key benefit, the

application and the DDS communication part can be

decoupled; the application can be developed without

determining who should receive the messages, where

recipients are located, what happens if messages

cannot be delivered.

Additionally, DDS allows the user to specify

Quality of Service (QoS) parameters to configure

discovery and behavior mechanisms. DDS simplifies

distributed applications and encourages modular,

well-structured programs.

In spite of its benefits, DDS has some limitations

to apply to distributed simulation area which needs

more requirements like federation save/restore and

synchronization. Joshi et al. tried to overcome those

limitations by making an equivalent to HLA-like

federation or time management service (Joshi and

Castellote, 2006). Nextel Aerospace Defense &

Security (NADS) also focused on HLA architecture

migrating to new architecture by fusing DDS

middleware (Lopez and Martin 2011). They used the

DDS standard as default for messaging, while the

middleware object model was based on HLA metadata.

However, previous studies have some limitations.

They did not fully consider building a distributed

high-resolution engagement model using a reusable

component-based simulation environment. As

mentioned above, system developers want the

execution of a high resolution engagement model

within reasonable time limits. So it has been required

to develop a component-based distributed simulation

infrastructure which can simulate a large and complex

engagement model effectively.

2.2 AddSIM and AddSIM-DDS

AddSIM is an engagement simulation environment

for composing and reconfiguring weapon system

models, in plug-and-play way (Oh et al., 2014).

AddSIM aims to integrate the models which were

A DDS-based Distributed Simulation for Anti-air Missile Systems

271

developed and used during each weapon system

development phase. AddSIM users can make and

simulate their models in synthetic battle-fields for

weapon system effectiveness analysis.

Figure 1 shows the operational concept of

AddSIM. AddSIM is installed on a local computer.

Users can develop models, setup simulation scenarios,

execute, and get the results with GUI. Developed

models are saved in the repository. AddSIM can

accommodate models on remote computer. AddSIM

provides environmental services (terrain, atmosphere,

and maritime), spatial service, journaling/logging

service.

Figure 1: Operational concept of AddSIM.

A top-level component in AddSIM is called a

player and it describes the behavior of a single

weapon system. About distributed case, AddSIM

supports integrating remote players (weapon systems)

into the master simulation. The Adaptive

Communication Environment (ACE) Object Request

Broker (ORB), in short TAO is used for

communicating data in Figure 2. AddSIM performs

the simulation using distributed objects rather than

executing a distributed simulation in a parallel

manner.

However, the distributed object simulation of

such schemes is relatively slow to communicate data

because of its tightly coupled character. Also, in

simulation cases of many loosely coupled participants,

it seems to be needed that distributed simulation

techniques, which have distributed simulation

engines and communicate essential data for

interoperation.

Figure 2: Distributed simulation concept of AddSIM.

Based on these ideas, AddSIM-DDS is developed.

DDS middleware is added to the communication

layer of the original AddSIM and some modification

of Graphical User Interface (GUI) and Kernel layer

for accommodate the DDS. Figure 3 shows the

distributed simulation concept using AddSIM-DDS.

Figure 3: Distributed simulation concept using AddSIM-

DDS.

In this point, we have to synchronize the spatial

database because all spatial AddSIM players, almost

physical combat weapon systems, shall share same

spatial data for engagement. AddSIM-DDS is

designed to synchronize the all spatial DB’s instantly

using DDS and not to journal DB’s. In addition,

distributed AddSIM-DDS nodes are synchronized to

correctly execute the entire simulation without

temporal causality errors.

3 METHOD

This section describes modeled scenario, overall

structure of the model, and the way for simulation

construction on distributed nodes. We utilize

AddSIM-DDS to develop the system models

participating in air-defense operations and construct

federation.

3.1 Modeled Scenario

Our model includes enemy and friend entities with

focus on a ground-to-air engagement situation.

Especially, for the purpose of performance prediction

and evaluation of our defensive systems, we focused

on air defense operations area. With the presented

scenario, we can analyze relationships between

performance variables of defensive systems and

operational effectiveness. For example, we analyzed

how the delay of C2 network affects the final miss

distance, the closest distance between aircraft and

missile.

The concept of the scenario is described in Figure

4. When a fleet of enemy aircrafts are coming into our

defense area, our multi-function radar detects and

AddSIM

Player

AddSIM

Player

AddSIM

Player

AddSIM

Player

AddSIM

Player

AddSIM

Player

AddSIM Kernel

(Master)

AddSIM Kernel

(Remote)

Spatial DB

Journaling

Distributed Object

Middleware (TAO)

Data Centric Middleware (DDS)

Spatial DB

Journaling

DB A

AddSIM Kernel

DDS Service

Module

AddSIM

Player

AddSIM

Player

AddSIM

Player

Spatial DB

Journaling

DB B

AddSIM Kernel

DDS Service

Module

AddSIM

Player

AddSIM

Player

AddSIM

Player

Spatial DB

Journaling

DB X

AddSIM Kernel

DDS Service

Module

AddSIM

Player

AddSIM

Player

AddSIM

Player

Local Interface

Damage Assessment Interface

Remote Interface(Interaction)

∙∙∙

SIMULTECH 2016 - 6th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

272

tracks them. After analyzing and data fusing, radar

sends fire commands to a launcher with the target

information. Then, fire commands are transmitted to

a launcher across C2 network with some delays. After

receiving the information, the launcher calculates

launch direction and the midcourse way-point of

inertial navigation guidance. After all the necessary

information for launch is injected to a missile, the

launcher sends a launch signal to the missile. The

fired missile flies to a pre-determined way-point

using inertial navigation guidance, and switches to

homing guidance flight with a seeker. Enemy

aircrafts warned on an incoming missile attack can do

an evasive flight according to their operational

concepts.

Figure 4: A brief scenario of air-defense engagement.

3.2 Simulation Construction

We designed and implemented simulation models to

represent the entities described in Section 3.1. Our

distributed air-defense engagement model (ADEM)

includes continuous-time models representing

dynamics of continuous systems such as multi-

function radar (MFR), missiles (MSL), launchers

(LCR), and air threats (ATS). It also includes discrete

event models which describe the logic of engagement

control systems (ECS), embedded control systems in

missiles, and overall controls of simulation scenarios.

We used AddSIM to develop the engagement

model. System-level entities in the scenario are

modeled as AddSIM players. AddSIM players may

have many functional and physical sub-components

hierarchically, which represents behavior or physical

structure of the system. Hierarchical structure of our

simulation model is illustrated in Figure 5. Detailed

behavior and mathematical formulations are based on

the previous research (Oh and Kim, 2012). In this

study, we focused on the construction of a DDS-based

distributed simulation using AddSIM-DDS.

Figure 5: Hierarchical structure of ADEM.

Figure 6: Structure of a distributed simulation.

The entire federation of ADEM consists of four

AddSIM-DDS federates corresponding to AddSIM

players. Figure 6 shows the distributed structure of

ADEM simulation. All information exchanged in the

AddSIM-DDS federation is divided into two

categories. One is an instantaneous message for

interrupt or notification, the other is persistent object

data throughout the simulation. This corresponds to

an interaction and object class in HLA, and an

interface and spatial data in AddSIM. As described

Section 2.2, every distributed AddSIM node has its

own spatial database and contents of the database are

synchronized by distributing updated data through

DDS middleware. This spatial database plays an

environmental model in the domain of an agent-based

simulation.

ADEM: Distributed Air-

defense Engagement Model

Air Threat

Radar

Control

Dynamics

Multi-function

Radar

Detect

Network

Launcher

Launch

Missile

Navigation

Seeker

Guidance

Autopilot

Dynamics

: System Model (Player)

: Subsystem Function

AddSIM Kernel AddSIM Kernel

Data Centric Middleware (DDS)

Spatial DB

Journaling

AddSIM Kernel

DDS Service

Module

Spatial DB

Journaling

AddSIM Kernel

DDS Service

Module

DDS Service

Module

ATS player

↓ DetectInfo

MFR player

Spatial DB

Journaling

DDS Service

Module

Spatial DB

Journaling

DetectInfo ↓

↑ LaunchCmdInfo

↓ LaunchCmdInfo

MSL playerLCR player

A DDS-based Distributed Simulation for Anti-air Missile Systems

273

Table 1: Topics for data exchange in ADEM federation.

Topic Attribute Data type Description QoS

DetectInfo

Detect_time STime* Detected time (s)

TOPIC_QOS

_DEFAULT

KEEP_ALL

_HISTORY_QOS

BY_SOURCE

_TIMESTAMP

_DESTINATIONO

RDER_QOS

VOLATILE_DUR

ABILITY_QOS

RELIABLE_RELI

ABILITY_QOS

Lockon Bool Lock on flag

Target_ID Integer Target identifier (enumeration)

Target_type Integer Target type (enumeration)

Target_pos[3] Double[3] Target position (East, North, Up) (m)

Target_vel Double Target velocity (m/s)

Target_azim Double Target azimuth (rad)

LaunchCmd

Info

Launch_cmd Bool Launch command flag

Waypoint_pos[3] Double[3]

Waypoint for inertial navigation

(East, North, Up) (m)

SpatialInfo

Simulation_time STime Simulation timestamp

Object_ID String Object identifier

Object_parent_ID String Parent identifier

Object_Type Integer Object type

IFF Integer

Identification friend or foe

(enumeration)

RCS Double Radar cross-section (m

)

Damage_state Integer Damage state

Object_state Integer Object state

Object_pos[3] Double[3]

Object position

(Latitude, longitude, and altitude)

Object_orient[3] Double[3]

Object attitude

(Yaw, pitch, and roll)

Object_LineVel[3] Double[3] Linear velocity in the ENU

Object_LineAcc[3] Double[3] Linear acceleration in the ENU

Object_RotVel[3] Double[3] Rotational velocity (rad/s)

Object_RotAcc[3] Double[3] Rotational acceleration (rad/s

)

AdditionalState[10] Double[10] Additional state variables

*STime is a structure defined in AddSIM to represent the simulation time without floating-point error.

Table 2: Topic for time synchronization in AddSIM-DDS.

Topic Attribute Data type Description QoS

TimeInfo

Publisher_ID String Publisher identifier

Same as in Table 1

Executer_ID String Executer identifier

Time_stamp STime Federate time

Time_type Integer Time category (enumeration)

We defined two types of message topics –

‘DetectInfo’ and ‘LaunchCmdInfo’. As described in

Figure 6, allied force federates exchange them for

launch information exchange. While these topics

should be defined and exchanged explicitly by model

developers, topics for spatial information exchange

and time synchronization are created and exchanged

autonomously by AddSIM-DDS. They are

‘SpatialInfo’ and ‘TimeInfo’ topics. Table 1 and 2

show DDS topics and their attributes which are

defined in ADEM federation.

Table 3: Publish-subscribe relationship between federates.

Topic ATS MFR LCR MSL

DetectInfo - P S -

LaunchCmdInfo - - P S

SpatialInfo P/S P/S P/S P/S

TimeInfo P/S P/S P/S P/S

Table 3 shows publish and subscribe relations

between federates.

SIMULTECH 2016 - 6th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

274

4 EXPERIMENTS AND RESULTS

We installed four AddSIM-DDS federates on

physically distributed computers in the laboratory.

After multiple simulations, we confirmed that the

result of the distributed federated simulation are same

as the result on a single AddSIM node. Figure 7

shows the result of the distributed air-defense

simulation in which number of enemy aircrafts and

our anti-air missiles are involved. It depicts the same

trajectories as the results of sequential simulation on

a single node. Furthermore we verified all results

quantitatively by comparing the simulated log.

Therefore, we can simulate a large and complex

air-defense engagement model effectively in a

distributed environment, without worries of

insufficient memory or long simulation execution

times.

Figure 7: Simulation result of the distributed air-defense

simulation.

AddSIM provides some external interfaces for

interoperation with legacy models. A HLA/RTI

interface is one of them (Kim, Oh, and Hwang, 2013).

AddSIM also provides DDS connectivity via the

kernel (Kim et al., 2014). Although AddSIM utilizes

the two middleware services in a different

architectural way, we were very curious on the

performance difference between the two services.

For comparison, we built another set of ADEM

which are the same as the previous model except for

using HLA instead of DDS. In order to observe the

change in the total simulation time in accordance with

increasing traffic, we changed number of updates per

unit simulation time (1/timestep). The comparison

result is shown in Figure 8. X axis shows the number

of updates per unit simulation time. We gave the

seven changes from 10 to 10,000. The data size of

each update are 1,192 bytes on the average -

SpatialInfo (266 bytes) and TimeInfo (32 bytes).

Sometimes additional traffic are generated -

DetectInfo (22 bytes) and LaunchCmdInfo (14 bytes),

but the impact on the overall traffic is relatively small.

Figure 8: Simulation finish time vs. number of updates in

AddSIM-DDS and -HLA.

We employed OpenDDS Ver 3.8 as the DDS

middleware and MÄK RTI Ver 4.1 as the HLA/RTI

middleware. All experiments were performed on four

distributed computers in the laboratory. Each

computer has Intel® Core™ i7 3.40GHz processor

and 8GB RAM. They are running on Microsoft

Windows 7 Ultimate and communicating via a 1GB

Ethernet network.

Although there are some constraints and

assumptions, we could check performance

degradation on the simulation finish time with

AddSIM-HLA. More importantly, the performance

degrades significantly after a certain point of traffic.

Therefore, AddSIM-DDS is more effective to

simulate a large and complex air-defense engagement

model in a distributed environment.

These results are limited to the case of AddSIM-

DDS and –HLA, and cannot be generalized to DDS

and HLA. As mentioned, this experiment includes

constraints and assumptions such as each middleware

in a different architectural way, ignoring various

parameter setting of middleware, and not performing

the statistical analyses.

5 CONCLUSIONS

We introduced a DDS-based distributed engagement

simulation approach for the development of air-

defense guided weapons systems. We constructed the

distributed simulation using AddSIM-DDS federates.

DDS gives a powerful communication infrastructure

with its real time performance, high rate messaging,

and various QoS capabilities. And it gives a great

synergy when combined with component-based high

resolution simulation environment-AddSIM.

-5000

-4000

-3000

-2000

-1000

1000

2000

500

1000

1500

North(m)

East(m)

Height(m)

Aircrafts

Missiles

200

400

600

800

1000

1200

1400

1600

1800

2000

10 100 1000 10000

Simulation finish time (sec)

Number of updates per unit simulation time

(1/timestep)

AddSIM-HLA

AddSIM-DDS

A DDS-based Distributed Simulation for Anti-air Missile Systems

275

The presented approach can be applied effectively

to a large and complex engagement simulation in

which a number of high-resolution engineering

models participate. Our experiment results show the

validity and effectiveness of the DDS-based

distributed simulation. However, the results of our

experiments are limited to the case of AddSIM-DDS

and –HLA. And we need to expend more effort on the

statistical analyses and application to other complex

engagement simulation models continuously.

REFERENCES

Andrew Foster. Prismtech. 2014. Using DDS for scalable,

high performance, real-time data sharing in next

generation Modeling & Simulation systems (DDS-

Modeling-Simulation-WP-050914).

Choi, S. Y., & Wijesekera, D. 2000. The DADSim air

defense simulation environment. In High Assurance

Systems Engineering, 2000, Fifth IEEE International

Symposim on. HASE 2000 (pp. 75-82). IEEE.

Fujimoto, R. M. 2000. Parallel and distributed simulation

systems (Vol. 300). New York: Wiley.

Hong, J. H., Seo, K. M., Seok, M. G., & Kim, T. G. 2011.

Interoperation between engagement-and engineering-

level models for effectiveness analyses.The Journal of

Defense Modeling and Simulation: Applications,

Methodology, Technology, 8(3), 143-155.

Joshi, R., & Castellote, G. P. 2006. A comparison and

mapping of data distribution service and high-level

architecture. Technology, The Netherlands. His

research interests include parallel and distributed

computing, component based architectures, and

embedded systems.

Kim, D. H., Oh, H. S., & Hwang, S. W. 2013. Integrating

legacy simulation models into component-based

weapon system simulation environment.

In Proceedings of the 2013 Summer Computer

Simulation Conference (p. 26). Society for Modeling &

Simulation International.

Kim, D., Paek, O., Lee, T., Park, S., & Bae, H. 2014. A

DDS-based distributed simulation approach for

engineering-level models. In Simulation Conference

(WSC), 2014 Winter (pp. 2919-2930). IEEE.

Lopez-Rodriguez, J. M., Martin, R., & Jimenez, P. 2011.

How to Develop True Distributed Real Time

Simulations? Mixing IEEE HLA and OMG DDS

standards. Nextel Aerospace Defense & Security

(NADS), http://simware. es.

Oh, H. S., & Kim, D. H. 2012. Generic simulation models

to evaluate integrated simulation environment.

In Advanced Methods, Techniques, and Applications in

Modeling and Simulation (pp. 413-424). Springer

Japan.

Oh, H. S., Park, S., Kim, H. J., Lee, T., Lee, S., Kim, D., ...

& Park, J. H. 2014. AddSIM: A new Korean

engagement simulation environment using high

resolution models. In Simulation Conference (WSC),

2014 Winter (pp. 2942-2953). IEEE.

OMG (Object Management Group). 2007. Data

Distribution Service for Real-time Systems

Specification Version 1.2. http://www.omg.org/spec/

DDS/1.2/PDF/formal-07-01-01.pdf.

Sung, C. H., Hong, J. H., & Kim, T. G. 2009. Interoperation

of DEVS models and differential equation models using

HLA/RTI: hybrid simulation of engineering and

engagement level models. In Proceedings of the 2009

Spring Simulation Multiconference (p. 150). Society

for Computer Simulation International.

SIMULTECH 2016 - 6th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

276