Addressing the Challenges of Tactical Information Management in Net-Centric Systems With DDS

Tactical information management sys-tems increasingly run in net-centric

environments characterized by thousandsof platforms, sensors, decision nodes, andcomputers connected together toexchange information, support sense-making, enable collaborative decisionmaking, and effect changes in the physicalenvironment. For example, the GlobalInformation Grid (GIG) is an ambitiousnet-centric environment being designed toensure that different services and coalitionpartners, as well as individuals participat-ing to specific missions, can collaborateeffectively and deliver appropriate fire-power, information, or other essentialassets to warfighters in a timely, depend-able, and secure manner [1]. Achievingthis vision requires the following capabili-ties from the distributed middleware soft-ware:• Shared operational picture. A key

requirement for mission-critical net-centric systems is the ability to sharean operational picture with planners,warfighters, and operators in real-time.

• Ensure the right data gets to theright place at the right time by satis-fying end-to-end QoS requirements,such as latency, jitter, throughput,dependability, and scalability.

• Interoperability and portability inheterogeneous environments. Sincenet-centric systems are faced withunprecedented challenges in terms ofplatform and network heterogeneity,they are necessary.

• Support for dynamic coalitions. Inmany net-centric tactical informationmanagement systems, dynamicallyformed coalition of nodes will need toshare a common operational pictureand exchange data seamlessly.

Prior middleware technologies (such as

the Common Object Request BrokerArchitecture [CORBA] Event Service andNotification Service, the Java MessageService [JMS], and various other propri-etary middleware products) have histori-cally lacked key architectural and QoScapabilities, such as dependability, surviv-ability, scalability, determinism, security,and confidentiality needed by net-centricsystems for tactical information manage-ment. To address these limitations – andto better support tactical informationmanagement in net-centric systems likethe GIG – the OMG has adopted theDDS specification, which is a standard forQoS-enabled data-centric pub/sub com-munication aimed at net-centric tacticalinformation management systems [2].DDS is used in a wide range of militaryand commercial systems including navalcombat management systems, commercialair traffic control, transportation manage-ment, automated stock trading systems,and semiconductor fabrication devices.

The remainder of this article presentsan overview of DDS that is geared tosoftware architects. We also discuss theDDS QoS policies that are the most rele-vant for net-centric tactical informationmanagement systems. Finally, we explainhow DDS has been applied in practice toaddress key challenges of developing andoperating distributed software in currentand planned net-centric tactical informa-tion management systems.

Overview of DDSDDS provides the following capabilitiesfor net-centric tactical information sys-tems:• Universal access to information

from a wide variety of sources that runover potentially heterogeneous hard-ware/software platforms and net-

works.• An orchestrated information bus

that aggregates, filters, and prioritizesthe delivery of this information towork effectively under the restrictionsof transient and enduring resourceconstraints.

• Continuous adaptation to changesin the operating environment, such asdynamic network topologies, publisherand subscriber membership changes,and intermittent connectivity.

• Standard QoS policies and mecha-nisms that enable applications andadministrators to customize the wayinformation is delivered, received, andprocessed in the appropriate form andlevel of detail to users at multiple lev-els in net-centric systems.This section describes the key capabil-

ities and entities in DDS and shows howits QoS policies can be used to specify andenforce performance-related requirementsof net-centric tactical information man-agement systems. Figure 1 shows the vari-ous profiles and layers in the DDS stan-dard. The lower layer defines a Data-Centric Publish Subscribe (DCPS) plat-form, whose goal is to provide efficient,scalable, predictable, and resource-awaredata distribution. The higher layer is theData Local Reconstruction Layer (DLRL),which is an optional interface that pro-vides an object-oriented facáde atop theDCPS. The DLRL can be used to maptopics onto object fields and defines navi-gable associations between objects.

A separate specification, called theReal-Time Publish/Subscribe (RTPS)DDS interoperability wire protocol,defines the standard network protocolused to exchange data between publishersand subscribers that use different imple-mentations of DDS [3]. The remainder of

Addressing the Challenges of Tactical InformationManagement in Net-Centric Systems With DDS

Recent trends in net-centric systems motivate the development of tactical information management capabilities that ensure theright information is delivered to the right place at the right time to satisfy quality of service (QoS) requirements in heteroge-neous environments. This article presents an architectural overview of the Object Management Group’s (OMG) DataDistribution Service (DDS), which is a standards-based QoS-enabled data-centric middleware platform that enables applica-tions to communicate by publishing information they have and subscribing to information they need in a timely manner. DDSis an important distributed software technology for mission-critical Department of Defense (DoD) net-centric systems becauseit supports the following: (1) location independence, via anonymous publish/subscribe (pub/sub) protocols that enable commu-nication between collocated or remote publishers and subscribers, (2) scalability, by supporting large numbers of topics, datareaders, and data writers and platform portability, and (3) interoperability, via standard interfaces and transport protocols.

Dr. Douglas C. Schmidt, Dr. Angelo Corsaro, and Hans van’t HagPrismTech Corporation

24 CROSSTALK The Journal of Defense Software Engineering March 2008

Software Engineering Technology

March 2008 www.stsc.hill.af.mil 25

this section describes the conceptualmodel of DDS and explains the QoS poli-cies that are most relevant for net-centrictactical information management systems.

DDS Conceptual Model Domains and PartitionsDDS applications send and receive datawithin a domain. Domains can be dividedinto partitions that allow the separationand protection of different data flows.Although DDS entities can belong to dif-ferent domains, only participants withinthe same domain can communicate, whichhelps isolate and optimize communicationwithin communities that share commoninterests. For example, each communica-tion layer within the GIG could be associ-ated with a DDS domain and further sub-divided into partitions. This approach iso-lates domain participants across layers,which enables effective use of resourcesand helps enforce security and confiden-tiality policies.

Global Data SpaceDDS provides a strongly typed global dataspace within each domain in which appli-cations produce and consume the dynam-ically changing portions of a shared infor-mation model, as shown in Figure 2. DDS’information model capabilities are similarto those of relational databases, exceptthat DDS’ global data space is completelydistributed, QoS-aware, and allows anony-mous and asynchronous sharing of acommon information model. The DDSinformation model is the only knowledgepublishers and subscribers need to com-municate, i.e., they need not be aware ofeach other nor be concerned with low-level network programming details, suchas Internet protocol addresses, port num-bers, remote object references, or servicenames. By allowing data to flow where andwhen needed, DDS’s global data spaceenables the sharing of tactical informationand situational awareness informationneeded to implement net-centric tacticalinformation management systems.

TopicA DDS topic is an association between adata type, a set of QoS, and a uniquename, as shown in Figure 3 (see page 26).A topic is also the unit of informationcontained in DDS’ global data space andis used by applications to define theirinformation model and associate QoSpolicies with it. DDS applications in net-centric systems define their informationmodel by identifying topics that are rele-vant for solving their requirements andorganizing them into either relational or

object-oriented models. DDS thus allowsthe expression of the system informationmodel as either a 1) topic relational model,which can be thought of as an extensionof the familiar entity relationship diagramsused in data bases, decorated with QoS, or2) an object-oriented model, which canalso be synthesized as an object-orientedview of the relational model.

The DCPS layer provides support forrelational modeling, while the DLRLextends the DCPS with an object-orientedfacade, so that applications can eithercompletely ignore the DCPS relationalmodels or build an object model atop theDLRL. Data associated with DDS topicsare expressed using types defined by thestandard OMG Interface DefinitionLanguage (IDL), which simplifies theinter-working between DDS and CORBA.Relationships between topics can be cap-tured via keys that can be used to distin-guish between different instances of thesame topic.

In net-centric tactical information sys-tems, an information model will be associ-ated with every layer in which DDS-baseddata exchange occurs. This informationmodel, which can comply with DoD orNorth American Trade Organization stan-dards, is the lingua franca used by the dif-ferent applications in coalitions toexchange information and seamlesslyinteroperate. Likewise, the QoS policiesdecorating the information model deter-mine how the data is disseminated, per-sisted, and received in the global dataspace.

Publishers and SubscribersIn net-centric tactical information man-agement systems, publishers and sub-scribers correspond to a range of domainparticipants such as embedded devices,Unmanned Air Vehicles (UAVs), soldiers’equipment, as well as planning and simula-tion services in operations centers. DDSapplications use data writers to publish

Data Centric Publish Subscribe (DCPS)

Data Local Reconstruction Layer (DLRL)

A

B

CD

E

F

m

J

K

TopicType QoS

Name

Figure 1: Profiles and Layers in the DDS Standard

Data Centric Publish Subscribe (DCPS)

Data Local Reconstruction Layer (DLRL)

A

B

CD

E

F

m

J

K

Name

Figure 2: DDS Global Data Space in a Domain

Addressing the Challenges of Tactical Information Management in Net-Centric Systems With DDS

data values to the global data space of adomain and data readers to receive data. Apublisher is a factory that creates andmanages a group of data writers with sim-ilar behavior or QoS policies, as shown inFigure 4. A subscriber is a factory that cre-ates and manages data readers, as shownin Figure 4.

Publishers can declare their intent toproduce data on particular topics withassociated QoS, and they distribute thedata on those topics to the global dataspace. Subscribers receive topic data in theglobal data space that match their sub-scriptions (the rules that define what rep-resents a matching subscription aredescribed below). QoS policies allow pub-lishers and subscribers to define, first,their local behavior, such as the number ofhistorical data samples they require andthe maximum update-rate at which theywant to receive data, and, second, howdata should be treated once in transit withrespect to reliability, urgency, importance,and durability. Topics can also be annotat-ed with these QoS policies to drive thebehavior of the data-distribution. TheQoS policies of pre-defined topics serveas defaults for publishers and subscribersand can therefore ensure consistencybetween requested and offered QoS.

Subscriptions and MatchingA subscription is an operation that associ-ates a subscriber to its matching publish-ers, as shown in the center of Figure 4. Inaddition to the topic-based subscriptions

described, DDS also supports content-basedsubscription, in which a subset of the stan-dard Structured Query Language (SQL) isused to specify subscription filters. InDDS a matching subscription must matchthe following two types of a topic’s prop-erties: (1) its features, such as its type,name, key, and content; (2) its QoS poli-cies, which are described in the QoS Policiessection.

The matching process for QoS uses arequested/offered (RxO) model shown inTable 1, where the requested QoS must beless than or equal to the offered QoS. Forexample, subscribers requesting reliabledata delivery cannot communicate withpublishers that only distribute data usingbest effort delivery. Likewise, subscriberscannot request a topic update whose dead-line is smaller than that declared by anypublishers.

The subscription matching mechanismprovided by DDS enforces a powerfulform of design by contract [4], where QoS isused together with type information todecide whether publishers and subscriberscan communicate. This extended form ofdesign by contract helps ensure that net-centric systems will operate as intended,both from functional and QoS perspec-tives. These assurances are essential in thedevelopment, deployment, and operationof mission-critical net-centric tacticalinformation management systems.

DiscoveryAnother key feature of DDS is that allinformation needed to establish commu-nication can be discovered automatically,in a completely distributed manner.Applications dynamically declare theirintent to become publishers and/or sub-scribers of one or more topics to the DDSmiddleware, which uses this informationto establish the proper communicationpaths between discovered entities. This

capability supports dynamic scenarioscommon in net-centric tactical informa-tion management where cooperatingdomain participants join and leavethroughout system operation.

QoS PoliciesDDS is designed for mission-critical net-centric systems where the right answerdelivered too late becomes the wronganswer. To meet timing requirements it isessential that the middleware controls andoptimizes the use of resources, such as net-work bandwidth, memory, and CPU time.Table 1 shows the rich set of QoS policiesthat DDS provides to control and limittopic (T), data reader (DR), data writer(DW), publisher (P), and subscriber (S)resources and topic QoS properties, suchas persistence, reliability, and timeliness [2].Below we discuss the DDS QoS policiesthat are the most relevant for net-centrictactical information management systems.

Data AvailabilityDDS provides the following QoS policiesthat control the availability of data todomain participants:• The Durability QoS policy controls

the lifetime of the data written to theglobal data space in a domain.Supported durability levels include thefollowing: (1) volatile, which specifiesthat once data is published it is notmaintained by DDS for delivery to latejoining applications; (2) transient local,which specifies that publishers storedata locally so that late joining sub-scribers get the last published item if apublisher is still alive; (3) transient,which ensures that the global dataspace maintains the information out-side the local scope of any publishersfor use by late joining subscribers; and(4) persistent, which ensures that theglobal data space stores the informa-tion persistently so it is available to latejoiners even after the shutdown andrestart of the system. Durability isachieved by relying on a durability ser-vice whose properties are configuredby means of the DURABILITY_SER-VICE QoS.

• The LIFESPAN QoS policy controlsthe interval of time during which adata sample is valid. The default valueis infinite, with alternative values beingthe time-span for which the data canbe considered valid.

• The HISTORY QoS policy controlsthe number of data samples (i.e., sub-sequent writes of the same topic) thatmust be stored for readers or writers.Possible values are the last sample, the



TopicType QoS

Name

Figure 3: DDS Topic

Publisher

DataWriter

Topic

Type

QoS

Name

writes

QoS

QoS

DataWriter

Topic

Typewrites

Subscriber

DataReaderreads

QoS

DataReaderreads

...

QoS

Name

QoS

QoS QoS

QoS matching

QoS matching

......

Figure 4: DDS Publisher/Writer Subscriber/Reader and Subscription/QoS Matching



last n samples, or all samples.These QoS policies provide the DDS

global data space with the ability to coop-erate in highly dynamic environmentscharacterized by continuous joining andleaving of publisher/subscribers. Thiscapability makes it possible for net-centrictactical information management systemsto share a common operational pictureregardless of the dynamism that charac-terizes portions of the systems, such ascoalitions of soldiers collaborating inurban environments or coordinated UAVsin support of tactical operations.

Data DeliveryDDS provides the following QoS policiesthat control how data is delivered andwhich publishers are allowed to write aspecific topic:• The PRESENTATION QoS policy

gives control on how changes to theinformation model are presented tosubscribers. This QoS gives control onthe ordering as well as the coherencyof data updates. The scope at which itis applied is defined by the accessscope, which can be one ofINSTANCE, TOPIC, or GROUPlevel.

• The RELIABILITY QoS policy con-trols the level of reliability associatedwith data diffusion. Possible choicesare RELIABLE and BEST_EFFORTdistribution.

• The PARTITION QoS policy givescontrol over the association betweenDDS partitions (represented by astring name) and a specific instance ofa publisher/subscriber.

• The DESTINATION_ORDER QoSpolicy controls the order of changesmade by publishers to some instanceof a given topic. DDS allows theordering of different changes accord-ing to source or destination time-stamps.

• The OWNERSHIP QoS policy con-trols which writer owns the write-access to a topic when there are multi-ple writers and ownership is EXCLU-SIVE. Only the writer with the highestOWNERSHIP_STRENGTH canpublish the data. If the OWNERSHIPQoS policy value is shared, multiplewriters can concurrently update atopic. OWNERSHIP thus helps tomanage replicated publishers of thesame data.These QoS policies control the relia-

bility and availability of the data, thusallowing the delivery of the right data tothe right place at the right time. More elab-orate ways of selecting the right data are

offered by the DDS content-awarenessprofile that allows applications to selectinformation of interest based upon theircontent.

Data TimelinessDDS provides the following QoS policiesthat control the timeliness properties ofdistributed data:• The DEADLINE QoS policy allows

applications to define the maximuminter-arrival time for data. DDS can beconfigured to automatically notifyapplications when deadlines aremissed.

• The LATENCY_BUDGET QoS pol-icy provides a means for applicationsto inform DDS of the urgency associ-ated with transmitted data. The latencybudget specifies the time period with-in which DDS must distribute theinformation. This time period startsfrom the moment the data is writtenby a publisher until it is available in thesubscriber’s data-cache ready for useby reader(s).

• The TRANSPORT_PRIORITY QoSpolicy allows applications to controlthe importance associated with a topic

or with a topic instance, thus allowinga DDS implementation to prioritizemore important data relative to lessimportant data.These QoS policies make it possible to

ensure that tactical information needed toreconstruct the shared operational pictureis delivered in a timely manner.

ResourcesDDS defines the following QoS policiesto control the network and computingresources that are essential to meet datadissemination requirements:• The TIME_BASED_FILTER QoS

policy allows applications to specifythe minimum inter-arrival timebetween data samples, therebyexpressing their capability to consumeinformation at a maximum rate.Samples that are produced at a fasterpace are not delivered. This policyhelps a DDS implementation optimizenetwork bandwidth, memory, and pro-cessing power for subscribers that areconnected over limited bandwidth net-works or which have limited comput-ing capabilities.

• The RESOURCE_LIMITS QoS poli-

QoS Policy Applicability RxO Modifiable

DURABILITY

DURABILITYSERVICE

LIFESPAN

HISTORY

PRESENTATION

RELIABILITY

PARTITION

DESTINATIONORDER

OWNERSHIP

OWNERSHIPSTRENGTH

DEADLINE

LATENCYBUDGET

TRANSPORTPRIORITY

TIME BASEDFILTER

RESOURCELIMITS

USER_DATA

TOPIC_DATA

GROUP_DATA

T, DR, DW Y N

DataAvailability

T, DW N N

T, DW - Y

T, DR, DW N N

P, S Y N

Data Delivery

T, DR, DW Y N

P, S N Y

T, DR, DW Y N

T, DR, DW Y N

DW - Y

T, DR, DW Y Y

DataTimeliness

T, DR, DW Y Y

T, DW - Y

DR - Y

ResourcesT, DR, DW N N

DP, DR, DW N Y

ConfigurationT N Y

P, S N Y

Table 1: Key QoS Policies for Net-Centric Systems

cy allows applications to control theamount of message buffering per-formed by a DDS implementation.DDS’s QoS policies support the vari-

ous elements and operating scenarios thatconstitute net-centric tactical informationmanagement. By controlling these QoSpolicies it is possible to scale DDS fromlow-end embedded systems connectedwith narrow and noisy radio links, to high-end servers connected to high-speedfiber-optic networks.

ConfigurationThe QoS policies described above providecontrol over the most important aspectsof data delivery, availability, timeliness,and resource usage. In addition, DDS alsosupports the definition and distribution ofuser specified bootstrapping informationvia the following QoS policies:• The USER DATA QoS policy allows

applications to associate a sequence ofoctets to domain participant data read-ers and data writers. This data is thendistributed by means of the DCPSparticipant built-in topic. This QoSpolicy is commonly used to distributesecurity credentials.

• The TOPIC_DATA QoS policyallows applications to associate asequence of octets with a topic. Thisbootstrapping information is distrib-uted by means of the DCPS Topicbuilt-in topic. A common use of thisQoS policy is to extend topics withadditional information, or meta-infor-mation, such as eXtensible MarkupLanguage schemas.

• The GROUP_DATA QoS policyallows applications to associate asequence of octets with publishers andsubscribers. This bootstrapping infor-mation is distributed by means of theDCPS subscription,and DCPS publi-cation built-in topics, respectively. Atypical use of this information is toallow additional application controlover subscriptions matching.

DDS Success StoriesAlthough DDS is a relatively new standard(adopted by the OMG in 2004), it hasbeen adopted quickly due to its ability toaddress key requirements of data distribu-tion in net-centric systems, as well as thematurity and quality of available imple-mentations, which are based on decades ofexperience developing data-centric mid-dleware for mission-critical systems.Moreover, DDS has been mandated bythe U.S. Navy’s Open ArchitectureComputing Environment as the standardpublish/subscribe technology to use in

next-generation combat management sys-tems, and Defense Information SystemsAgency as the standard technology forpublish/subscribe to be used in all new orupgraded systems [5, 6]. Several majordefense programs, such as the U.S. Navy’sDDG-1000 land attack destroyer, U.S.Army’s Future Combat Systems (FCS),and the Thales Tactical Information AndCommand System Operating System(TACTICOS), also adopted DDS evenbefore it was mandated, underscoringDDS’ ability to address the data distribu-tion challenges of next generation net-centric defense systems.

For example, the TACTICOS combatmanagement system developed by ThalesNaval Netherlands is based on an imple-mentation of DDS that allows them toachieve very good scalability, from small

ships to aircraft carrier grade, as well ashigh performance, availability, and deter-minism even under temporary overloadconditions [7, 8]. TACTICOS is currentlyin use in 15 navies worldwide serving 20ships-classes ranging from small patrolboats up to large frigates. The utilizationof DDS is instrumental in its successsince it provides both the scalability tosupport thousands of applications run-ning on more than 150 distributed com-puters on a frigate size system. Anotherkey feature of DDS is its battle-damageresistance, meaning that software can bedynamically re-allocated to the remainingcomputer pool in case of an error on aspecific computer. The DDS PersistenceProfile support is instrumental in thisdynamic reallocation since it allows appli-cations to store their internal state into theDDS middleware, which manages thisstate in a distributed and fault-tolerant wayso that restarted applications can continuewhat they were doing before they crashed.

The DDS implementation used onTACTICOS supports a data-centricapproach where at the start of the systemdesign, the information model can be cap-tured, annotated with proper QoS poli-cies, and then shared between multipleparties. This federated architecture is com-mon in existing and planned coalition-baseddevelopments where multiple partiesjointly implement the overall combat sys-tem. DDS provides the fault-tolerantinformation backbone onto which allthese applications are deployed and is thusresponsible for providing each applicationwith the right information at the righttime.

Along with the rapid adoption ofDDS in the defense domain, its use is alsosteadily growing in other domains, such astransportation, telecommunications, andfinance. For example, in the context ofAir Traffic Control and Management,DDS has been selected as thepublish/subscribe middleware for distrib-uting flight data plans in CoFlight [9],which is the next generation EuropeanFlight Data Processor. In general, DDS isan appropriate middleware technology forapplication domains that require rich sup-port for QoS policies and high-perfor-mance and dependability standards-based,commercial-off-the-shelf implementa-tions.

Concluding RemarksDDS is a standards-based QoS-enableddata-centric publish/subscribe middle-ware that provides a feature rich data-cen-tric real-time platform to support theneeds of current and planned net-centrictactical information management systems.Its powerful set of QoS policies, togetherwith its scalable architecture, makes it aneffective and mature choice for solving thedata distribution and information manage-ment problems net-centric systems [10].Next, we summarize how DDS addressesthe key challenges outlined in the intro-duction in a standard and interoperablemanner:• Shared operational picture. DDS

provides effective support for thesetypes of applications via its QoS poli-cies for defining the scope, content,and QoS of the data model that under-lies the operational picture.

• The right data at the right time atthe right place via DDS QoS policiesthat enable a fine-grained control overinformation delivery, such as the abili-ty to control many aspects of data dis-semination to ensure timely deliveryand optimal resource usage.

• Heterogeneous environment. By



DDS provides thefault-tolerant information

backbone onto whichall these applications are

deployed and is thusresponsible for providingeach application withthe right informationat the right time.



About the Authors

Douglas C. Schmidt,Ph.D., is a professor ofcomputer science atVanderbilt Universityand is the chief technicalofficer of PrismTech.

His expertise focuses on distributedcomputing middleware, object-orientedpatterns and frameworks, and distrib-uted real-time and embedded systems.He has authored nine books and morethan 350 papers in top technical journals,conferences, and books that cover high-performance communication softwaresystems, real-time distributed comput-ing, and object-oriented patterns forconcurrent and distributed systems.

PrismTech Corporation6 Lincoln Knoll LNSTE 100Burlington, MA 01803Phone: (781) 270-1177 Fax: (781) 238-1700E-mail: doug.schmidt

@prismtech.com

Angelo Corsaro, Ph.D.,is the OpenSplice DDSproduct marketing man-ager at PrismTech andco-chairs the OMG DDSSpecial Interest Group

and the Real-Time Embedded andSpecialized Services task force. He iswell-known in the distributed real-timeand embedded systems middlewarecommunity and has a wealth of experi-ence in hard real-time embedded sys-tems, large-scale, and very large-scaledistributed systems such as defense,aerospace, homeland security and trans-portation systems.

PrismTech Corporation6 Lincoln Knoll LNSTE 100Burlington, MA 01803Phone: (781) 270-1177 Fax: (781) 238-1700E-mail: angelo.corsaro

@prismtech.com

Hans Van’t Hag is theOpenSplice DDS prod-uct manager, atPrismTech. He hasextensive experience inapplying an information

approach towards mission-critical andreal-time net-centric systems. Hag is acontributor to the OMG DDS specifica-tion and has presented numerous paperson DDS and publish/subscribe middle-ware technologies. Prior to joiningPrismTech, he worked at Thales NavalNetherlands (TNN) where he wasresponsible for the development of thedata-centric real-time middleware asapplied in TNN’s combat system in ser-vice with 15 navies worldwide.

PrismTech Corporation6 Lincoln Knoll LN STE 100Burlington, MA 01803Phone: (781) 270-1177 Fax: (781) 238-1700E-mail: hans.vanthag

@prismtech.com

providing standard QoS policies thatcontrol the bandwidth used for pro-viding data to interested parties, DDSruns in heterogeneous platforms whileproviding different elements with acommon operational picture.

• Dynamic coalitions. The highlydynamic nature of DDS, such as itssupport for dynamic discovery, pro-vides an effective platform for sup-porting ad hoc interactions.DDS continues to evolve to meet new

operational and technical challenges ofnet-centric tactical information manage-ment systems. Three types of extensionsare currently being pursued for DDS bythe OMG. The first involves adding newplatform-specific models that fully lever-age programming language features, suchas standard C++ containers. The secondextension deals with extensible topics thatenable incremental system updates byensuring that changes in the data modeldo not break interoperability. The final setof extensions focus on network data rep-resentation and the syntax used to definetopics. For example, upcoming versions ofthe DDS standard will likely allow the def-inition of topics using XML, as well as theuse of XML or Java Script ObjectNotation as the network data representa-tion. DDS security has not yet been stan-

dardized. The OMG will be addressingthis area of standardization starting in thespring of 2008.

With multiple COTS and open-sourceimplementations and a solid track recordof success in mission-critical military andcommercial projects, DDS has a brightfuture as the standards-based middlewareof choice for net-centric tactical informa-tion systems. More information on DDSand its application in practice are availablein online forums [11, 12] where expertsdiscuss advanced features of the DDSstandard and new directions for the tech-nology, while DDS beginners can learnfrom past experiences and ask questionsabout patterns and best practices forapplying DDS in their net-centric sys-tems.u

References1. “The United States Department of

Defense Quadrennial Defense ReviewReport.” Feb. 2006 <www.defenselink.mil/qdr/report/Report20060203.pdf>.

2. OMG. “Data Distribution Service forReal-Time Systems Specification.”<www.omg.org/docs/formal/04-12-02.pdf>.

3. OMG. “Real-Time Publish SubscribeProtocol – DDS Interoperability Wire

Protocol Specification.” <www.omg.org/cgi-bin/apps/doc?ptc/06-08-02.pdf>.

4. Bertrand, Meyer. Object OrientedSoftware Construction. 2nd ed.,Prentice Hall, 2001.

5. “Open Architecture ComputingEnvironment.” <www.nswc.navy.mil/wwwDL/B/OACE>.

6. Defense Systems Information Agency.“DoD Information TechnologyStandards Registry.” <https://disronline.disa.mil>.

7. THALES. “TACTICOS CombatManagement System, Exploiting theFull DDS Potential.” <www.omg.org/docs/dds/06-12-06.pdf.>.

8. “OpenSplice DDS.” <www.prismtech.com/openplice-dds>.

9. CoFlight eFDP <www.omg.org/docs/dds/07-07-04.pdf>.

10. Xiong, Ming, et al. “EvaluatingTechnologies for Tactical InformationManagement in Net-Centric Systems.”Proceedings of the Defense Transfor-mation and Net-Centric Systems Con-ference, Apr. 9-13, 2007, Orlando, FL.

11. OMG DDS SIG Portal <portals.omg.org/dds>.

12. OMG DDS Forum <www.dds-forum.org>.