LOCATION MANAGEMENT IN DISTRIBUTED,

SERVICE-ORIENTED COMMAND AND CONTROL SYSTEMS

Thomas Nitsche

Research Institute for Communication, Information Processing and Ergonomics (FGAN-FKIE)

Keywords: Location management, non-homogenous distrib

ution, neighbourhood, areas of interest, network centric

operations, command and control.

Abstract: In this paper we propose an efficient location management scheme for large amounts of mobile users and

other objects in distributed, service-oriented systems. To efficiently observe geographic areas of interest

(AOI) in command and control information systems (C2IS), i.e. to compute the AOI within a C2IS, we

introduce the concept of region services. These services contain all objects of a fixed geographic region. To

handle in-homogenous distributions of objects we propose a combination of regular and hierarchical

regions. A user-specific C2IS instance can now directly and efficiently establish subscription-relations to

the relevant objects around its AOI in order to obtain information about the position, status and behaviour of

these objects. If objects including the current user itself now dynamically change their position we merely

have to update the information relations to those few objects that enter or leave a region within the AOI,

instead of having to consider all objects within the global information grid. Region services thus do not only

improve the efficiency for generating a static common operational picture but can also handle any dynamic

changes of object positions.

1 INTRODUCTION

Location management, i.e. the determination and

tracking of object positions, plays an important role

in all systems where such location information is

used. In ubiquitous systems it delivers valuable

information about the geographic context of a user

and his surrounding.

In this paper we handle the question how such

location-awa

reness can be realized in large

distributed systems. Objects in such systems are

mobile users, sensors as well as spatial objects

representing specific geographic locations like

bridges or street corners. These (mobile) objects can

be represented as service instances or agents. To

adapt themselves to the current situation they must

be able to observe the relevant context, i.e. the other

objects in their direct neighbourhood, but also

potentially far more remote objects that might have

an effect on them.

In network-centric operations (NCO) (Alberts &

Hay

es, 2003; Kruse, Adkins & Holloman, 2005;

Wilson, 2004) as well as in crises management for

disaster scenarios (Arnold, et al., 2004; Denning,

2006a; Jungert, Hallberg & Hundstad, 2006)

command and control information systems (C2IS)

(Alberts & Hayes, 2006) are used for managing the

operations. In both cases the user has to make proper

decisions and react quickly, i.e. in real-time,

according to the current situation.

Since the number of objects and users is too

large

to be directly observable – it may reach

millions in network-centric operations – the C2IS

has to provide mechanisms to filter the available

information to the mission-specific parts that are

relevant to the corresponding users.

Areas of interest (AOI) are geographic areas like

he surrounding of a user’s position and his area of

responsibility where the user wants to get informed

about other objects and users that either are within

that specific area or that may have an effect on that

area. The objects themselves are in general

distributed in-homogenously over the overall

operational area (cf., e.g., (Mitschke & Peter, 2001)

for distribution of users within a disaster scenario)

and the objects are not static but may change their

positions with different speed: While pedestrians are

moving slowly, vehicles or even helicopters or air-

planes may change their position at very high speed.

Since each of the objects may correspond to a user

of the C2IS, we therefore have to manage the

location of potentially millions of mobile users in

parallel. Since the applications require a reaction in

144

Nitsche T. (2007).

LOCATION MANAGEMENT IN DISTRIBUTED, SERVICE-ORIENTED COMMAND AND CONTROL SYSTEMS.

In Proceedings of the Second International Conference on Software and Data Technologies - PL/DPS/KE/WsMUSE, pages 144-151

DOI: 10.5220/0001347001440151

 SciTePress

real-time, the corresponding location management

and AOI computation has to be done in an efficient

manner as well.

The remainder of this paper is organized as

follows. In Section 2 we describe the global

information grid and how geographically based

areas of interest can be defined in command and

control information systems. The location

management in C2IS is discussed in Section 3. An

efficient algorithm based on the concept of region

services working for non-homogenous distributions

is described in Section 4. Finally, Section 5

concludes.

2 AREAS OF INTERST

IN COMMAND

AND CONTROL SYSTEMS

2.1 The Global Information Grid

The global information grid (GIG) (Blais, Goerger,

Richmond, Gates & Willis, 2005; U.S. Government

Accountability Office, 2004) forms the technical

basis for realizing the power of the network-centric

operations concept. As is the computational or data

grid formed as a connection of compute or data

centres with the users (Foster & Kesselmann, 1998),

so is the information grid: It is based on the global

connection of all users and systems, ranging from

sensors over the command and control systems to

actors. In principle, every system or user may be

able to communicate with everybody else in the

network and can have access to remote services and

data. In the information grid, all available

information can thus be shared among the different

users, provided that the required security criteria are

satisfied.

Sharing all available information is, however,

not sufficient. From the information point of view,

using all available data leads to an explosion of the

information space, which cannot be handled

appropriately by humans and computers. To handle

the problem of information overload (Denning,

2006b) we have to restrict the data in an operational

picture to only that information that is relevant for

the corresponding user and his current mission

(Hayes-Roth, 2006; Nitsche, 2006).

A command and control information system

(C2IS) therefore has to provide corresponding

filtering mechanisms. Areas of interest (AOI) are

one such concept.

2.2 Areas of Interest

An area of interest (AOI) is the area of concern to

the user. Here we concentrate on geographically

based areas. The area of interest to a user of a C2IS

will generally include the (geographical) area of

operation of the corresponding user. Moreover, the

AOI will normally also include some surrounding

area to monitor the behaviour of neighbouring users

(be they friendly or hostile), who could influence the

successful completion of the user’s mission.

In a C2IS the AOI of a user is thus defined as an

area like the surrounding of his position or the

surrounding of his area of operation which he wants

to observe. That means that he wants to get informed

about other objects that are within that specific area.

This is feasible since most of the data presented in a

common operational picture (COP) (Mittu &

Segaria, 2000) by a C2IS have some spatial

reference. This includes (physical) positions of

mobile users (Hightower, Borriello, 2001)

representing own and foreign forces, spatial data like

information about streets or bridges, and others.

The shape of an AOI can be seen as a

combination of simple circular or rectangular

surroundings of the current user’s position and other

spatial data (cf., e.g., (Lisper & Holerin, 2002)). To

simplify the presentation we abstract from the exact

shape of the AOI for the remainder of this paper. We

only require that we are able to check if a certain

position is inside a given shape. Moreover, we also

abstract from the exact kind of objects stored in the

C2IS. We only need some attribute values about

their position and status, but do not distinguish

otherwise between them here.

2.3 Incorporating Effect Ranges

The simple form of AOIs defined above can be

extended based on the potential impact of objects on

us on their effect range, but also on their potential

impact in the future based on planned activities

(Schade & Hieb, 2006).

For incorporating the effect range of objects we

can use object properties like their speed, their

direction of movement or the range of the object or

their weapons to determine a distance from within

they can be a potential threat or supporter (cf., eg.,

(McEneaney & Singh, 2007; Scott, Wan, Rico,

Furusho & Cummings, 2007)).

In Figure 1, for instance, we filtered all objects

whose effect range (which is simplified visualized as

a green circle) does not intersect with the AOI. Thus

only those two objects located directly within the

LOCATION MANAGEMENT IN DISTRIBUTED, SERVICE-ORIENTED COMMAND AND CONTROL SYSTEMS

145

area of interest itself and the object in the lower right

part of the figure remain relevant in this example.

Figure 1: Area of interest (simplified visualized as red

circle) with effect range (simplified visualized as green

circles) of objects (blue dots) taken into account.

3 LOCATION MANAGEMENT

IN C2IS

Such user-parameterised areas of interest as

described before thus define a set of objects that the

user is interested in, i.e. that are relevant for his

operational picture. For its computation we hence

have to determine which objects are contained with

the AOI because they are within a specific proximity

of that user or might have an effect on him.

3.1 Service-Oriented C2IS

In this paper we assume a service-oriented

architecture of the C2IS (Käthner & Spielmann,

2004). The overall functionality is provided in the

form of services with user-specific instances, each of

them can be distributed onto different computers in

the grid which allows the C2IS to operate in a

decentralized manner. The shared data of the global

information space is thus not necessarily contained

in a central database but can also be distributed

within the grid among the different service

instances, where each of them only hold a local

portion of the overall information.

Relevant for the location management is a so-

called COP-service. The user-specific instance of

that service stores the data about the user itself and

its current status like its position. This information

can be delivered to other COP-services using a

publish-subscribe-approach. A user-specific COP-

service can subscribe to those objects relevant for

this user in order to become informed about their

position, status and future changes.

3.2 Simple Approaches and its

Limitations

However, before a COP-service can establish

subscription-relations to the relevant objects, we first

have to determine these relevant objects, i.e. we

have to find out which objects are actually within

our AOI.

In case that we were satisfied with an AOI

containing only objects in our direct neighbourhood

we might get the location information of these

objects just for free. A special routing protocol for

the radio communication in mobile ad-hoc networks

(MANETs) additionally transmits GPS data

(Bachran, Bongartz & Tiderko, 2005). Unfortunately

this is a proprietary protocol, so it only works if all

radio communication devices use this protocol

which we cannot assume to be valid, especially in

multinational operations or disaster scenarios

incorporating different organisations with different

equipment. Moreover, this approach is restricted to

the tactical level and the direct surrounding of

ourselves. Objects that are located far more remote

but are still relevant for our AOI, e.g. because of

their large effect range, cannot be handled here. The

same restriction holds for objects that are connected

directly via fibre cable instead of radio

communication.

Note that an IP-based location management

scheme (cf., e.g., (Shin, Park, Jung & Kim, 2006))

also does not work here: Due to the different

organisations involved that may use their own sub-

networks as well as command and control

hierarchies the geographic neighbourhood is not

necessarily correlated with that in a network, even if

we used tactical radio on the lower tactical levels.

Similar holds for RFID-based tracking systems like

(Satoh, 2006).

Thus in general we have to compute the AOI

explicitly.

Let N be the total number of objects available

within the grid. If no further information is

available, the AOI computation for a single user

requires O(N) time, since we have to check the

position of all N objects.

Unfortunately, we are not the only user in the

system. According to the NCO approach in principle

all users (except for hostile units) can do the same,

i.e. they may have an associated COP-service that

computes its own AOI. This implies that all objects

may define their own local areas of interest for

which they have to check the positions of all other

objects in turn. Implemented naively, this would

lead to an algorithm of quadratic time-complexity

O(N

), while synchronized all-to-all algorithms can

do this in O(N log N) time (Barnes & Hut, 1986).

Effect

ICSOFT 2007 - International Conference on Software and Data Technologies

146

This is, however, still not satisfactory. The

reason is that the above complexity only holds for a

single AOI computation with static object positions.

In practice, however, we have a situation where the

objects are mobile, i.e., they change their positions

at any time. This implies that some objects leave our

AOI, while others may enter it. Subscription-

relations to objects within out AOI only provide us

with position updates of those objects that we are

already monitoring. So we can determine if an object

leaves the AOI, but we will not know if another

objects moves towards us. Even worse, we may (and

in general will) move ourselves towards other

objects that are not in our AOI yet and hence are not

monitored by us. As a result, dynamic changes of

object positions will soon make the AOI outdated.

As a consequence we had to re-compute the AOI

in regular intervals in order to update it. The

dynamic behaviour of the objects thus forces us to

execute the above described algorithm over and over

again. Since it is a global algorithm this not only

takes the computation time but – in case of

distributed services – also the time for the all-to-all

communication scheme wasting a lot of bandwidth.

But even if we ignored the above described

computational and communication efficiency

problems of the algorithm and simply re-computed

the AOI in very short time intervals we still would

have the problem that we do not know if and when

the AOI becomes outdated due to un-monitored

objects entering our AOI without notice.

4 REGION SERVICES

The reason why the above approach is so inefficient

is that we repeatedly check the positions of all

available objects within the grid, while there are in

practice only a few objects relevant for our area of

interest, their number being in general much less

than N, i.e., the total number of objects. In the naïve

implementation we therefore filter out objects

according to their position from the full object set

rather than just checking if there exist any objects

with a specific position (in our surrounding) and

combining these small sub-sets directly.

4.1 Concept

Approaches in multi-cast communication schemes

(Carzaniga, Rosenblum & Wolf, 2001; Carzaniga,

Rutherford & Wolf, 2004; Sebé & Domingo-Ferrer,

2007) lead us to the idea of a region service. Such a

service defines a certain geographical region of the

world and contains a list of all objects that are

located within this region.

Based on a C2IS software architecture consisting

of COP-services, we hence can extend this by a set

of region services, each of them being responsible

for a certain region. In its simplest form we can

divide the earth (or at least our full operational area)

into regions of the same size (cf. Figure 2-(a)).

(a) Regular Regions (b) Hierarchical Regions

Figure 2: Division of an area into (a) regular regions with

the same size each, and (b) hierarchically defined regions

based on a quad-tree division with different sizes but

containing approximately the same number of objects.

In cellular radio networks such regions appear

naturally due to the limited range of radio

communication. A mobile user thus has connections

to only a few cells that determine the user’s location

area. Location management here means paging and

location update (Chew, Yeo & Kuan, 2007).

However, in network-centric operations the users

are connected via different communication channels

which include radio networks, satellite connections

and fibre lines. In the global information grid, where

all objects are connected to, is the geographic

proximity in general not directly visible. Even if two

users are in direct neighbourhood of each other, they

may use different kinds of radio communication

devices due their affiliation to different

organisations or nations and hence do communicate

not directly with each other via radio but via their

corresponding home organisations.

Moreover, the location of a user is actually not a

single position but potentially a larger area of its

effect range. We can thus do not directly use the

concepts used in (cellular) radio networks for

general command and control information systems.

We hence use directly the spatial position

information of an object (like its GPS position).

Figure 2-(a) shows the division of an area into

regular regions, each of them being of the same size.

In such a regular division the borders of each region

can be computed very easily and the test if a certain

position falls within a specific region can be done

very efficiently. (This computation of the affiliated

region to a certain position corresponds to the

paging process in cellular radio networks.)

Each of the spatial objects can thus be added to

one of the region services in constant time. To detect

the objects within the area of interest (AOI) for a

user, we now only have to request the objects from

LOCATION MANAGEMENT IN DISTRIBUTED, SERVICE-ORIENTED COMMAND AND CONTROL SYSTEMS

147

those region services that overlap with the user’s

AOI. This takes O(N/R) time on average for each

region service, with R being their number. If the

number R of regions is sufficiently large we can thus

achieve that each region service is – on average –

responsible for only a restricted number of objects.

Unfortunately, the objects in a military domain

are in general not regularly distributed around the

world but concentrated on the battlefields. This

implies that the number of objects located within

each regular region may vary significantly: While

some regions may be (almost) empty, other regions

may contain a large amount of objects. This not only

affects the efficiency of the algorithm described

before but may even prevent the required real-time

behaviour of the command and control information

system.

A similar problem arises in massively

multiplayer online real-time games. Here a zoning

approach is used to distribute the server load onto

parallel servers, each of them being responsible for a

different partition, i.e. zone, of the game world.

However, while this zoning concept scales with the

number of players and the size of the users, it does

not scale with the density of players within a certain

area. If the object density within one region is too

large, the server does not respond in real-time

anymore. (Müller & Gorlatch, 2006) proposes the

replication of heavily populated zones onto different

game servers on the grid. Here each server contains

the whole game state, but only processes the so-

called active entities as part of the global state and

afterwards broadcasts its local computation result.

While in games such a bad responsiveness is at

most annoying, in real-life like in network-centric

operations the inability to locate the different objects

in time may actually result in real casualties.

The region services in our approach are basically

storing data (the local objects in that region) and are

not computing-services. (The actual computation is

done in the COP-services. Note that the mapping of

service-instances to actual servers is outside of the

scope of this paper.) The replication thus does not

make sense here. However, the idea of parallelizing

the load onto different (sub-)servers can be adopted

if we divide a region (with its object set) into sub-

regions (with its corresponding sub-sets of

associated objects).

An adaptive approach to the definition of regions

therefore has to take the actual distribution of

objects into account. Organizing region areas

hierarchically as a quad- or oct-tree (in case of two-

or three-dimensional coordinates, see Figure 2-(b))

allows region services to be defined in such a way

that they all contain (almost) the same number K of

objects. This leads to a O(K) constant time algorithm

for retrieving the objects within an AOI, provided

the size of the AOI is small compared to the whole

area, i.e. the AOI covers only a fixed number of

regions. However, the check which region a certain

position belongs to requires O(log (N/K)) time in

this case, so dynamic object movements are more

expensive here than for regular regions.

The concept of region services not only improves

the efficiency for generating a static operational

picture but can also handle dynamic changes of

object positions. In general all user objects are

mobile. Thus the objects within and near the AOI of

a user may leave or enter the AOI dynamically.

However, not only the surrounding objects but also

the user itself may move. Therefore a static

computation of the AOI does not work, but we have

to compute and update the AOI dynamically: If an

object is located in the area of one of the regions that

overlaps with our AOI, we can subscribe to that

object for position updates. If that object finally

leaves the area of that region and is thus out of our

AOI, we can cancel the subscription of that object.

If, on the other hand, a currently un-monitored

object changes its position in such a way that it is

entering the area of one of the regions that overlap

with our AOI, we will be informed by the region

service about this object and can immediately

subscribe to it. We therefore only have to check for a

limited amount of objects if they are within our AOI.

Note that a region service should in average

contain not only one but multiple objects, because

this shall give the best trade-off between 1) the time

required to manage the objects within the region

service and their potential dynamic movements from

one region to another region on one hand, and 2) the

time to determine the region service(s), i.e. to check

where a certain object belongs to, on the other hand.

4.2 Location Management using

Region Services

The actual location management algorithm for

computing the areas of interest with the support of

region services is described here. See also Figures 3

and 4 for an illustration.

Effect

Figure 3: Computing the AOI with effect ranges using

region services. The resulting area is shown in Figure 1.

ICSOFT 2007 - International Conference on Software and Data Technologies

148

Figure 4: Phases of the location management process.

Phases 1-2 (see Figure 4) set the pre-conditions

by initializing the region services. Phases 3-6 fetch

the location information of the objects in the user’s

surrounding of the current user. Finally in phase 7

the AOI is computed locally. Phases 2 and 5 can be

omitted for the location management without effect

ranges.

1 Register objects at corresponding region

services of current position: First all objects

have to register themselves at their

corresponding region service, i.e. they publish

themselves.

In case of regular regions it takes constant

time for each object and can be done in parallel

for each region service, while in case of

hierarchical regions it takes O(log R) time for

each object to find its proper region, and since

adding new objects may change the region

hierarchy this has to be done sequentially.

Once all the objects are registered at their

corresponding region service, their COP-

services take care of their movements: If an

object leaves the area of one region and enters

the area of another region, the service

automatically de-registers the object at the old

region service and registers itself at the new

one. These dynamic updates may happen in

parallel to the AOI computation, provided we

ensure that an object is registered at the new

region service before it de-registers itself in the

old region.

2 Register objects at region services of current

effect range: Then all objects additionally have

to register at the corresponding region service

of all those regions that they have an effect on.

Note that in this case one object may be

registered at multiple region services.

3 Determine relevant region services for the

AOI: To compute the AOI of a user, we have to

determine the relevant region services for this

user. This includes all those regions that

intersect with the shape of the AOI.

3..

4 Get object sets from the relevant regions for

the AOI: Then we read all objects from the

relevant region services determined in phase 3

before. We thus get a list of object identifiers,

or links to their corresponding COP-services.

In order to get informed if that object-set

changes (due to moving objects) we have to

create a subscription-relation to the region-

service.

In addition to the objects located within

certain regions here we also get objects that

might have an effect within that region.

5 Merge object sets: Since effect objects may be

registered at multiple region services, we have

to merge the object sets from different regions.

For example, in Figure 3 the object in the

lower right part of the figure is – due to its

large effect range – registered at two of the

relevant regions (shown in light blue, c.f.

Figure 4-4.).

6 Read object positions and effect range: Read

further information, especially the location, of

these objects. This can be done by creating

subscription-relations to the corresponding

COP-services of these objects, which also

informs us about later changes of their values

and positions.

7 Determine objects within the AOI: Finally we

check which of the objects actually belong to

the AOI and which does not. Those objects

where their position or their effect range

intersects with the shape of the AOI are to be

displayed, while the others are filtered out.

Note that if one of the objects received in

phase 4 changes its position we have to check

again if it has moved into or out of the AOI.

However, since from phase 4 we get only

objects in the direct neighbourhood of the AOI,

there are generally much less objects to be

filtered out than had been in the simple

algorithm of Section 3.

4.3 Dynamic Location Updates

The presented algorithm works well for a static

setting where the objects do not move. Here we

discuss that is also appropriate for the dynamic case.

If the user is moving, the shape of his AOI may

intersect with other regions than before. In this case

we have to update the subscription relations to the

corresponding region services. This will provide us

LOCATION MANAGEMENT IN DISTRIBUTED, SERVICE-ORIENTED COMMAND AND CONTROL SYSTEMS

149

with the objects within these regions such that we

can update the AOI.

If, on the other hand, another object (that may be

monitored by the user’s AOI) is moving in such a

way that it leaves the area of its current region, it has

to register itself at the new region-service (and de-

automatically inform the user about the change

within the object set such that the AOI can be

adapted locally.

Moreover, the object registration at the new

region service will increase the number of objects

handled by it, if not another objects leaves the region

at the same time. In the hierarchical region model

based on quad- or oct-trees this would mean that we

had to split the region into sub-regions, if a certain

maximal number of objects had been reached. (Note

that for efficiency reasons we do not split

immediately but allow more than one object within

each region.) If a region is split, we create new

region-services for the sub-regions, while the old

region-service takes the role of the parent service.

However, one problem still remains with the

hierarchical regions. In order to check which region

service a specific location belongs to, we have to

retrieve the corresponding region service by

traversing the region-tree, starting from the root

region which denotes the whole area. This leads to

O(log R) messages which is not appropriate for

small bandwidth radio networks, and may put high

loads onto the root region server.

For this reason we combine regular and

hierarchical regions: The regular regions as in

Figure 2-(a) serve as base regions, since here the

object location mapping can be done in a single step.

Only if there are too many objects within that region

due to an in-homogenous distribution we further

sub-divide that region in a hierarchical manner. The

region service for the corresponding hierarchical

base region serves as the root for the sub-tree of

regions, which finally leads to a forest of quad- or

oct-trees.

The performance of the different location

management schemes is evaluated in (Nitsche,

2007).

5 CONCLUSION

A command and control information system (C2IS)

has to provide mechanisms to filter the information

available in the C2IS to the mission-specific parts

that are relevant to the corresponding military

commander or other C2IS users. Areas of interest

(AOI) are geographic areas like the surrounding of

the user’s position and his area of responsibility

where the user wants to get informed about other

military objects, e.g., own and foreign forces, that

are either within that specific area or that may have

an effect on that area.

To efficiently observe such areas, i.e. to compute

the AOI within a C2IS, we introduce the concept of

region services. These services contain all objects of

a fixed geographic region. Regions can be defined in

a regular manner or hierarchically based on quad- or

oct-trees.

To handle in-homogenous distributions of

objects we propose a combination of regular and

hierarchical regions: The regular base regions are

used to directly derive the basic region service

responsible for the corresponding area, thus avoiding

the communication overhead necessary if we had to

read the data-dependent region-distribution from a

root region service (and its sub-services). The

concept of hierarchical regions is used in case of

larger densities of objects within a region. In the

latter case we split the region into sub-regions which

are managed by separate services.

A user-specific C2IS instance can now directly

and efficiently establish subscription-relations to the

relevant objects around its AOI in order to obtain

information about the position, status and behaviour

of these objects. If objects including the current user

itself now dynamically change their position we

merely have to update the information relations to

those few objects that enter or leave a region within

the AOI, instead of having to consider all objects

within the global information grid.

Region services thus do not only improve the

efficiency for generating a static common

operational picture but can also manage the dynamic

changes of object locations.

The proposed location management scheme

based on region services can not only be used in

command and control information systems but in all

distributed, service-oriented systems with large

amounts of mobile users.

REFERENCES

Alberts, D.S., Hayes, R.E (2003). Power to the Edge:

Command and Control in the Information Age. CCRP

Publication Series, Washington, DC, USA.

Alberts, D.S., Hayes, R.E (2006). Understanding

Command and Control. CCRP Publication Series.

Anthony, R.W. (1999). Relating Large and Small in C2

and Operations. In Command and Control Research

and Technology Symposium (CCRTS), Newport, RI.

Arnold, J., Levine, B., Mammatha, R., Lee, F., Shenoy, P.,

et al. (2004). Information-Sharing in Out-of-Hospital

ICSOFT 2007 - International Conference on Software and Data Technologies

150

Disaster Response: The Future Role of Information

Technology. Prehospital and Desaster Medicine,

19(2): 201-207.

Bachran, T., Bongartz, H. H.-J. & Tiderko, A. (2005). A

Framework for Multicast and Quality based

Forwarding in MANETs. In Proc. 3rd IASTED

International Conference on Communications and

Computer Networks (CCN’05), ACTA Press.

Blais, L., Goerger, N.C., Richmond, P., Gates, B. &

Willis, J.B. (2005). Global Information Grid Services

and Generation of the Mobility Common Operational

Picture. In 2005 Fall Simulation Interoperability

Workshop (SIW), Orlando, Florida.

Barnes, J.E., Hut, P. (1986). A hierarchical O(N log N)

force calculation algorithm. Nature, 324(4), 446-449.

Carzaniga, A., Rosenblum, D.S., & Wolf, A.L. (2001).

Design and Evaluation of a Wide-Area Event

Notification Service. ACM Transactions on Computer

Systems, 19(3): 332-383.

Carzaniga, A., Rutherford, M.J. & Wolf, A.L. (2004). A

Routing Scheme for Content-Based Networking. In

Proc. IEEE INFOCOM 2004, Hong Kong, China.

Chew, Y.H., Yeo, B.S., Kuan, D.C.M. (2007). Sensitivity

study of location management area partitioning in

cellular communication systems. Computer Networks,

51(6): 1555-1573.

Denning, P.J. (2006a). Hastily formed networks.

Communications of the ACM, 49(4), 15-20.

Denning, P.J. (2006b). Infoglut. CACM, 49(7): 15-19.

Foster, I., Kesselmann, C., editors (1998). The Grid:

Blueprint for a New Computing Infrastructure.

Morgan Kaufmann.

Hayes-Roth, F. (2006). Two Theories of Process Design

for Information Superiority: Smart Pull vs. Smart

Push. In Command and Control Research and

Technology Symposium (CCRTS), San Diego, CA.

Hightower, J., Borriello, G. (2001). Location Systems for

Ubiquitous Computing. IEEE Computer, 57-66.

Jungert, E., Hallberg, N., Hundstad, A. (2006). A Service-

Based Command and Control Systems Architecture

for Crises Management. International Journal of

Emergency Management, 3 (2-3): 131-148.

Kruse, J., Adkins, M. & Holloman, K.A. (2005). Network

Centric Warfare in the U.S. Navy’s Fifth Fleet. In

Proc. 38rd Hawaii International Conference on

Systems Sciences (HICSS), IEEE Computer Society.

Käthner, S., Spielmann, M. (2004). A Base Component for

Network-Based Service-Oriented C4ISR Systems. In

Coalition C4ISR Architectures and Information

Exchange Capabilities, The Hague, The Netherlands.

Lisper, B., Holerin, J. (2002). Development and

Verification of Parallel Algorithms in the Data Field

Model, In Constructive Methods for Parallel

Programming, 127-141, Nova Science Publishers.

McEneaney, W.M., Singh, R. (2007). A Computationally-

Feasible Algorithm for Estimation of Opponent

Strength in Urban Combat. In 12th International

Command and Control Research and Technology

Symposium (ICCRTS), Newport, RI, USA.

Mitschke, T., Peter, H. (2001). Handbuch für Schnell-

Einsatz-Gruppen, 3rd ed. S+K Verlag, (in German).

Mittu, R., Segaria, F. (2000). Common Operational

Picture (COP) and Common Tactical Picture (CTP)

via a Consistent Networked Information Stream

(CNIS). In Command and Control Information

Systems, Monterey, CA.

Müller, J., Gorlatch, S. (2006). Scaling Online Games on

the Grid. In Proc. 4

International Game design and

Technology Workshop and Conference (GDTW 2006).

Nitsche, T. (2006). Information Access in Tactical

Command and Control Information Systems. In 11th

International Command and Control Research and

Technology Symposium (ICCRTS), Cambridge, UK.

Nitsche, T. (2007). Evaluating Region-Based Location

Management Schemes for Service-Oriented Command

and Control Information Systems. Subm. to Mil. Com-

munication and Information Systems Conference

(MCC 2007).

Satoh, I. (2006). Location-based services in ubiquitous

computing environments. International Journal on

Digital Libraries, 6(3), 280-291, Springer.

Schade, U., Hieb, M.R. (2006). Development of Formal

Grammars to Support Coalition Command and

Control: A Battle Management Language for Orders,

Requests and Reports. In 11th International Command

and Control Research and Technology Symposium

(ICCRTS), Cambridge, UK.

Scott, S.D., Wan, J., Rico, A., Furusho, C., Cummings,

M.L. (2007). Aiding Team Support in Command and

Control Operations with Large-Screen Displays. In

Proc. of HSIS 2007: ASNE Human Systems

Integration Symposium, Annapolis, MD, USA.

Sebé, F., Domingo-Ferrer, J. (2007). Scalability and

security in biased many-to-one communication.

Computer Networks, 51(1): 1-13.

Shin, S.Y., Park, S.-H., Jung, B.H., Kim, C. (2006).

Dynamic Location Management Scheme Using Agent

in a Ubiquitous IP-Based Network. In Management of

Convergence Networks and Services, 9th Asia-Pacific

Network Operations and Management Symposium

(APNOMS 2006), LNCS 4238, 491-500, Springer.

U.S. Government Accountability Office (2004). Defense

Acquisition - The Global Information Grid and

Challenges Facing Its Implementation. Report GAO-

04-858.

Wilson, C. (2004). Network Centric Warfare: Background

and Oversight Issues for Congress. CRS Report for

Congress RL32411.

LOCATION MANAGEMENT IN DISTRIBUTED, SERVICE-ORIENTED COMMAND AND CONTROL SYSTEMS

151