WAVE-WP (WORLD PROCESSING) TECHNOLOGY

Peter Sapaty

Masanori Sugisaka

Institute of Mathematical Machines and Systems

National Academy of Sciences,

Glushkova Ave 42, 03187 Kiev, Ukraine

Department of Electrical and Electronic Engineering

Oita University, 700 Oaza Dannoharu, 870-1192 Japan

Keywords: parallel and distributed processing, distributed control, space navigation, spati

al pattern-matching, WAVE-

WP language, system integrity, system management, distributed simulation, cooperative robotics, open

systems.

Abstract. A new computational and control model of parallel and distributed nature has been developed. It comprises

self-evolving, space-conquering automaton, high-level system navigation and coordination language,

describing system problems in a spatial pattern-matching mode, and related distributed control mechanisms

for management of physical, virtual, and combined worlds. The model allows us to obtain complex spatial

solutions in a compact, integral, and seamless way. It can be effectively used for the creation, integration,

simulation, processing, management and control of a variety of dynamic and open systems – from physical

to biological, and from artificial to natural.

1 INTRODUCTION

The use of computers is steadily shifting from

computations to coordination and management of

complex distributed and dynamic systems, in both

civil and military areas. To speed up the progress in

this direction, we need qualitatively different

knowledge processing and control models as

traditional ones, like Turing machine or cellular

automata, underlying conventional computers and

programming in them, were originally oriented on

computations. We also need models that can

describe much broader activities than traditional

information processing, to operate in real worlds,

with physical movement and physical matter and

objects relocation and manipulation.

The execution of such models may need to

olve any existing manned or unmanned hardware

and software systems in the world, human beings

including. A possible general-purpose model of this

type, called WAVE-WP (or World Processing), as a

further development of the distributed network

processing WAVE paradigm (Sapaty, 1999), will be

summarized in this paper.

2 PROBLEMS OF DISTRIBUTED

PROCESSING AND CONTROL

TECHNOLOGIES

Single machine solutions are often showing highest

possible integrity as a system – with all resources

directly available, and control being global, direct,

and absolute. On the contrary, the existing

distributed computing and control models and

technologies follow the analytical approach,

representing and studying systems as consisting

from pieces, or agents, exchanging messages or

sharing common (distributed) objects. This results in

known difficulties of obtaining the needed global

behavior from local activities, especially if there

many of them and their number and interactions

change over time.

Sapaty P. and Sugisaka M. (2004).

WAVE-WP (WORLD PROCESSING) TECHNOLOGY.

In Proceedings of the First International Conference on Informatics in Control, Automation and Robotics, pages 92-102

DOI: 10.5220/0001152400920102

 SciTePress

Overhead in making large distributed systems

operate as a single controllable entity and follow

global goals, by using such approaches, may be

enormous, as symbolically shown in Fig. 1, see also

(Sapaty, 2002), and the necessity of runtime

recovery after failures may aggravate the situation

further.

Integral

single-

machine

solution

Breaking

into pieces

to be

distributed

communication &

synchronization

control

Distributed

solution

Figure 1: Overhead of distributed system solutions.

For solving complex problems in real, especially

unpredictable and hostile environments, we may

need distributed systems of much higher integrity,

being based on quite different (than traditional

pieces-to-whole) ideologies and methodologies.

The WAVE-WP model tries to attack the

problem by starting from the opposite side – from

the whole, allowing us to express it directly, while

abstracting from possible system’s parts and their

interactions, delegating the latter to an efficient

automatic implementation.

3 THE WAVE-WP CONTROL

AUTOMATON

The automaton effectively inherits the integrity of

traditional sequential programming over localized

memory, but for working now with the real

distributed world, while allowing its parallel

navigation in an active pattern flow and matching

mode – as a single spatial process.

The automaton may start from any point of the

distributed world or system to be controlled,

dynamically covering its parts or the whole, and

mounting of a variety of parallel and distributed

knowledge and control infrastructures. Implanting

distributed “soul” into the system organization, the

automaton increases the system's integrity,

capability of pursuing local and global goals,

assessing distributed situations, making autonomous

decisions, and recovering from indiscriminate

damages.

In quite other applications, it may need to

destroy the very system it propagates through and/or

operates in (or certain, for example, malicious

incursions or infrastructures in it). Many spatially

cooperating or competing parallel WAVE-WP

automata may evolve on the same system's body

serving, say, as deliberative, reactive, and/or

reflective spatial processes.

World representation. One of the main

features of WAVE-WP is the representation of

distributed worlds it operates in, as described in

(Sapaty, 2000).

Physical world (or PW) is continuous and

infinite in WAVE-WP. Existing at any its point, and

possibly performing a job, is considered as residing

in a node (with physical coordinates). Such a node,

reflecting only occupancy at the point, has no

personal identity or content. It vanishes with the

termination of all occupancies in it.

Virtual world (or VW) is discrete and interlinked

in WAVE-WP, and is represented, similar to WAVE

(Sapaty, 1999), by a distributed Knowledge Network

(KN). Its persistent nodes may contain established

concepts or facts, and (also persistent) links

(oriented and non-oriented, connecting the nodes)

may reflect different relations between the nodes.

The same model can also operate with the united

(or PVW) world, in which any element may have

features of the both worlds. A simplified example of

such a world is shown in Fig. 2.

node1

link1

node6

node3

Node4

node2

link8

link9

link3

link7

link5

link4

x12.5 y4

x13.5 y9

x10.8 y3

x7.5 y4.5

x3.5 y9

x2.8 y4.5

x10.5 y9.7

x5 y6.8

link4

Persistent virtual nodes

Physical-

Virtual

nodes

Temporary physical nodes

link10

Temporary links

Persistent

links

Figure 2: United distributed physical-virtual world.

A variety of effective access mechanisms to

nodes, links and their groups, say, by physical

coordinates, electronic addresses, by names, via

traversing links, etc. (classified as tunnel and surface

navigation) abound in the model, using both

selective and broadcasting access modes.

WAVE-WP (WORLD PROCESSING) TECHNOLOGY

Waves. Solutions of any problems in this

formalized world in WAVE-WP are represented as

its coordinated parallel navigation (or exploration,

invasion, grasping, coverage, flooding, conquest,

etc.) by some higher-level forces, or waves (Sapaty,

1999, 2000, 2002). These bring local operations,

control and transitional data directly into the needed

points of the world, to perform jobs there. The

obtained results, together with the same or other

operations may, in their turn, invade the other world

parts, and so on. In a most abstract form this “agent-

less” spatial process may look like shown in Fig. 3,

with parallel asynchronous navigation, spatial

intersection, branching and looping.

Wave 1

Wave 2

Distributed

physical &

virtual world

Spatial

looping

of waves

Sets of space

locations

reached &

logical

synchronization

Intersection

of moves in

space

Mutually

dependent

branching

of moves

Figure 3: Parallel conquest of distributed worlds by waves.

During the world navigation, which may be loose

and free or strictly (both hierarchically and

horizontally) controlled, waves can modify the very

world they evolve in and move through, as well as

create it from scratch (including any distributed

structures and topologies). Waves may also settle

persistent cooperative processes in its different

points, subsequently influencing, governing its

further development and evolution in the way

required.

4 HIGHER-LEVEL WAVE-WP

LANGUAGE

The system language expressing full details of this

new control automaton has been developed. Having

recursive space-navigating and space-penetrating

nature, it can operate with both information and

physical matter. The language can also be used as a

traditional one, so no integration with (and/or

interfaces to) other programming models and

systems may be needed for solving complex

distributed knowledge processing and control

problems.

Very compact syntax of the language, as shown

in Fig.4 (see also Sapaty, Sugisaka, 2002), makes it

particularly suitable for direct interpretation in

distributed environments, being supported by

effective program code mobility in computer

networks. In this description, braces set up zero or

more repetitions of a construct with a delimiter at its

right; square brackets identify an optional construct;

semicolon allows for sequential, while comma for

parallel invocation of program parts; and

parentheses are used for structuring of WAVE-WP

programs (or waves).

Successive program parts, or advances, develop

from all nodes of the set of nodes reached (SNR) by

the previous advance, whereas parallel or

independent parts, moves, constituting the advances,

develop from the same nodes, while splitting

processes and adding their own SNRs to the

resultant SNR of the advance.

Figure 4: Syntax of WAVE-WP language.

Elementary acts represent data processing, hops

in both physical and virtual spaces, and local control.

Rules establish non-local constraints and contexts

over space-evolving waves like, for example, the

ability to create networks, also allowing WAVE-WP

to be used as a conventional language. Variables,

called spatial (as being scattered in space), can be of

the three types: nodal, associated with virtual or

physical nodes and shared by different waves;

frontal, propagating with waves as their sole

property; and environmental, accessing elements of

internal and external environments navigated by

waves.

This recursive navigational structure of the

language allows us to express highly parallel and

fully decentralized, albeit strongly controlled and

coordinated, operations in distributed worlds in a

most compact way – in the form of integral space

processing and transformation formulae. These

formulae resemble data processing expressions of

traditional programming languages, but can now

operate in and process the whole distributed world.

wave Æ { advance ; }

advance Æ { move , }

move Æ constant | variable | { move act } |

[ rule] ( wave )

constant Æ information | physical-matter

variable Æ nodal | frontal | environmental

act Æ flow-act | fusion-act

rule Æ forward-rule | echo-rule

ICINCO 2004 - INTELLIGENT CONTROL SYSTEMS AND OPTIMIZATION

5 IMPLEMENTATION BASICS

On the implementation layer, the automaton widely

uses high-level mobile cooperative program code

self-spreading and replicating in networks, and can

be easily implemented on any existing software or

hardware platform. As the automaton can describe

direct movement and processing in physical world,

its implementation may need to involve a multiple

mobile hardware – with or without human

participation. A network of (hardware or software)

communicating WAVE-WP language interpreters

(or WI), which can be mobile if installed in manned

or unmanned vehicles, should be embedded into the

distributed world to be controlled (Sapaty, 1999,

2002), placing WIs in world’s most sensitive points,

as shown in Fig. 5.

During the spatial execution of system scenarios

in WAVE-WP, individual interpreters can make

local information and physical matter processing as

well as physical movement in space. They can also

partition, modify and replicate program code,

sending it to other interpreters (along with local

transitional data), dynamically forming track-based

distributed interpretation infrastructures.

The automaton can also exploit other systems as

computational and control resources, with or without

preliminary consent, i.e. in a (remotely controlled)

virus-like mode. For example, existing network

attacks (especially DDoS) may be considered as a

possible malicious (simplified, degenerated, and

distorted) implementation of the automaton.

Spatial scenario 1

Spatial scenario 2

Distributed

physical &

virtual World

Creating distributed

infrastructures

WAVE-WP

interpreters

Sensitive

points

Figure 5: A network of interpreters with self-spreading

spatial scenarios in WAVE-WP.

6 THE DISTRIBUTED WAVE-WP

INTERPRETER

The WAVE-WP interpreter consists of a number of

specialized processors working asynchronously and

in parallel, handling and sharing specific data

structures like waves queue, incoming and outgoing

queues, local part of the distributed knowledge

network, track forest, etc., and being responsible for

different interpretation operations (parser, data

processor, control processor, communication

processor, sensors and motion, etc.), see Fig. 6 (also

Borst, 2002; Sapaty, 1993, 1999).

The interpreter can be easily implemented on any

existing platform, in software or directly in silicon.

The existing public domain WAVE system (in C

under Unix/Solaris/linux) operating via the Internet

had been used in different countries, especially for

distributed network management (Gonzalez-

Valenzuela, Vuong, 2002) and simulation of

battlefields (Sapaty, 1999, 2002).

Frontal

Variables

Parser

Incoming

Queue

Outgoing

Queue

Track

Forest

Wave

Queue

Suspended

Waves

Nodal

Variables

Knowledge

Network

Control

Processor

Environmental

Variables

Wave

Identities

Communication

Processor

Operation

Processors

Figure 6: General organization of WAVE-WP interpreter

The interpreter may also have physical body, say,

as a mobile or humanoid robot, if engaged in

operations in physical world, or can be body-

mounted on humans. The whole network of such

“doers” can be mobile, changing structure, as robots

or humans can be moving at runtime (Sapaty,

Sugisaka, 2002a). Elementary expressions and

operations of WAVE-WP language may trigger a

combination of physical movements of doers in

space, transference of information and physical

matter between the doers both electronically, at a

distance, and in a direct, tactile contact, as well as

cooperation and group behavior of doers.

The following program first hops into some node

a of the distributed KN (located, say, in Doer.1),

picks up a certain matter from the environment there

into a frontal variable F, also information (integer 5)

into F1, then propagates via virtual link p to another

node b (which may happen to reside in Doer.2),

bringing the matter and information into the latter

doer. The rest of program (or wave) will be

performed from the destination doer.

direct#a; F=“matter”; F1=5; p#b; wave

WAVE-WP (WORLD PROCESSING) TECHNOLOGY

Coming to node b with F1 and wave could be

done purely wirelessly, but the transference of F

needs direct physical contact, which can be

performed by either Doer.1 or Doer.2 (or both)

moving for a meeting, or by engagement of another

doer, say Doer.3, for the matter’s transference, as

shown in Fig. 7.

Control, wave, frontal variables

Doer.1

Doer.2

Doer.3

Direct

contact

Direct

contact

Request & coordinates

Doer

movement

Confirmation

Waves

Physical matter

Information

Node

Link

Order of

operations

Figure 7: Transference of physical matter between doers

7 DIRECT ACCESS TO

PHYSICAL WORLD

The model can describe complex operations in PW

on a semantic level, abstracting from

implementation, with parallel control directly

evolving in physical space, covering the latter.

Multiple operations can be performed cooperatively

by groups of robots or humans, with transfers of

information and physical matter between the

interpreters embedded into physical bodies,

communicating both electronically and mechanically.

A coordinated delivery of physical matter to remote

points and its synchronized processing by a group of

robots had been demonstrated (Sapaty, Sugisaka,

2002a). A description of coordinated robotic column

movement and its synchronized operation had been

shown (Sapaty, Sugisaka, 2001).

As an elementary example of direct operations in

PW, the following program, applied in a certain

workplace, brings from remote locations sand,

cement and water, mixing them and making concrete,

and then delivers the obtained result into another

remote location (assigning it to nodal variable

Nconcrete there), using spatial programs w1-w4

for accessing the mentioned locations physically

(see Fig. 8):

(W4; Nconcrete) =

(w1; “3 tons of sand”) +

(w2; “2 tons of cement”) +

(w3; “4 tons of water”)

sand

cement

water

work place

concrete

place 1

place 3

place 2

Figure 8: Direct operations on physical matter.

8 DISTRIBUTED KNOWLEDGE

REPRESENTATION AND

PROCESSING

Dynamically creating arbitrary knowledge networks

in distributed spaces, which can be modified at

runtime, WAVE-WP can implement any knowledge

processing and control systems in parallel and fully

distributed way, similar to WAVE (Sapaty, 1999). A

program package in WAVE was demonstrated for

basic problems of the graph and network theory,

where each graph node can reside on a separate

computer. The package included finding spanning

trees, shortest paths, articulation points, maximum

cliques, diameter and radius, etc., also self-

recovering network topologies after indiscriminate

damages of their nodes and links. Parallel simulation

of other control models in WAVE-WP, like Petri

nets, was demonstrated too.

As an example, let us consider a KN of Fig. 9,

which may be arbitrarily distributed between

computers of Internet or robots, say, Robot.1 and

Robot.2.

Masanori

Peter

Bob

Doug

John

beer

whiskey

colleagues

know

respect

hate

Robot.1

Robot.2

colleagues

2, 4

Figure 9: Distributed knowledge representation.

ICINCO 2004 - INTELLIGENT CONTROL SYSTEMS AND OPTIMIZATION

In WAVE-WP, complex knowledge processing

tasks can be formulated and solved in parallel,

regardless of the knowledge distribution. Say, to

answer the question: “Who of John’s colleagues is

respected by both Masanori and Peter”, the

following simple program will be sufficient:

direct#John; colleagues#any;

and(

andparallel(-respect#(Masanori, Peter)),

USER = CONTENT)

It provides the final answer Doug, with the

decision-finding process shown in bold and stages

numbered in Fig. 9 (parallel operations emerging on

stage 3).

9 OPERATING IN PHYSICAL

WORLD UNDER THE

GUIDANCE OF VIRTUAL

WORLD

Operating in the unity of physical and virtual worlds,

the WAVE-WP model can effectively investigate

physical worlds and create their reflection in the

form of distributed virtual worlds. The latter can

guide further movement and search operations in the

distributed PW, modifying and updating itself, and

so on. Such physical-virtual world unity in one

model, where both worlds can be open and dynamic,

allows us to create integral intelligent manned or

unmanned systems operating effectively in

unpredictable environments.

This has been demonstrated on an example of

optimized parallel territory search by a group of

robots sharing distributed knowledge representation

of the area of interest, which can be modified and

updated (say, reduced or extended) at runtime

(Sapaty, Sugisaka, 2003). Parallel operations in PW

can be optimized in advance by solving the

problems in VW first, in a parallel simulation mode,

mapping subsequently or simultaneously the

obtained solutions into the PW.

For example, the space to be searched can be

described by a set of connected polygons and

represented in WAVE-WP by a KN reflecting the

polygons connectivity graph and individual polygon

data, as shown in Fig. 10. This spatially shared KN

can be arbitrarily distributed between robots and

constantly updated by them in different points.

p5p4p3

Robot.1

Robot.3

Robot.2

Figure 10: Space model distribution between robots

For example, by visiting polygons only once and

moving to free neighboring polygons, while

checking the states of surrounding neighbors

sequentially, the program, consisting of three

competing branches starting in polygons 1, 5, and 7

by different robots, will be as follows (using specific

external procedure

move_clean):

direct#(1,5,7); M=1; (p#)?move_clean;

repeat(or(g#; grasp(M==nil; M=1));

(p#)?move_clean)

10 INTELLIGENT NETWORK

MANAGEMENT

Integrating traditional network management tools

and systems, and dynamically extracting higher-

level knowledge from raw data via them, WAVE-

WP, same as WAVE, establishes a higher,

intelligent layer, allowing us to analyze varying

network topologies, regulate network load, and

redirect traffic in case of line failures or congestions

(Sapaty, Zorn, 1991). Representing a universal

spatial control model based on coordinated code

mobility and dynamic tracking, WAVE-WP can also

be used for creating essentially new, universal and

intelligent, network protocols. These, along with

traditional data delivery, will be able to make local

and global automatic decisions on the network

management, self-recovery after failures, and

runtime topology restructuring and optimization.

As the simplest example for distributed network

management in WAVE-WP, the following

maximum parallel program creates shortest path tree

(SPT) from some node a, covering the whole

network in a spatial navigation mode, by self-

replicating mobile code (see also Sapaty, 1999):

direct#a;

repeat(#;Fd+=L;Nd==,Nd>Fd;Nd=Fd;Np=B)

WAVE-WP (WORLD PROCESSING) TECHNOLOGY

Another program, starting from node e, collects

the shortest path from a to e in a reverse order via

the SPT found and recorded in a computer network

by the previous program:

direct#e;repeat(Fp&=C;#Np);U=Fp

The work of these two programs is shown in Fig. 11.

e,d,b,a

result

Shortest path tree creation program

Shortest path collection program

Link weights

Figure 11: Distributed dynamic shortest path solution.

11 ADVANCED CRISIS

REACTION FORCES

Smaller, dynamic armies, with dramatically

increased mobility and lethality, represent nowadays

the main direction, also challenge, in the

development of advanced military forces oriented on

crisis situations. These forces should be capable of

conducting non-traditional combat, peacekeeping,

and recovery operations, withstanding asymmetric

threats, and operating in unpredictable environments.

They may also effectively use multiple unmanned

units, such as mobile robots.

The WAVE-WP technology can quickly

assemble a highly operational battle force from

dissimilar (and possibly casual) units, setting

intelligent command and control (CC)

infrastructures over them (Sapaty, 2000a). These

infrastructures can follow global mission scenarios

and make autonomous decisions in manned,

automated, or fully unmanned mode. In case of

damages, the technology can either restore the

previous CC infrastructure (with, possibly, reduced

set of units), or make complete runtime

reassembling of the entire force, with a new CC

infrastructure, taking into account remaining

operational units and their physical locations. Both

local and global restructuring and reassembling can

be carried out at runtime, without the loss of overall

operational capability.

As an example, the following parallel program

creates and regularly updates an optimal hierarchical

CC infrastructure based on a physical neighborhood,

with top of the hierarchy always associated with the

most central unit of a distributed force. Individual

units (manned or unmanned vehicles, computerized

humans) can be on a constant move, but the optimal

spatial CC hierarchy is maintained under any

circumstances (including indiscriminate destruction

of units).

repeat(

Faver=average(direct#all; WHERE);

Nstart=min(direct#all;

(Faver, WHERE)?distance_ADDRESS):2;

direct#Nstart;

quit(

Frange=r40; N=1;

repeat(

direct#Frange; grasp(N==nil; N=1);

[any#any; LINK=nil];

[create(-infra#BACK)]));

USER=ADDRESS; TIME+=360)

A possible hierarchical infrastructure, created

and maintained by this program, is shown in Fig. 12.

infra

Most central unit

Mobile

crisis

reaction

units

Figure 12: Creating of a neighborhood hierarchy

12 MASSIVE COOPERATIVE

ROBOTICS

Autonomous robotic armies for civil and military

applications, comprising thousands of cooperating

mobile units, may well become today’s reality in

WAVE-WP (Sapaty, Sugisaka, 2002a). High-level

mission scenarios can be injected into such

organizations from any robotic unit, dynamically

covering and grasping the whole distributed system.

Massive coordinated operations in physical

spaces can automatically emerge during parallel

interpretation of the WAVE-WP scenarios, while

preserving the overall system integrity and ability of

pursuing global goals. The scenarios can be

represented in such a way that they should (and will

be able to) survive by any means, while individual

ICINCO 2004 - INTELLIGENT CONTROL SYSTEMS AND OPTIMIZATION

robots may fail, with system intelligence being the

feature of the whole campaign, rather than of

individual robots.

The following program sets up a number of

cyclic routes (with waypoints given) to be patrolled,

and then makes parallel branches compete and seize

available robotic bodies themselves, to follow these

routes. If a robot finds some irregularities on its own

route (using infrared sensors), it reports this to the

user of the whole system, and also reprograms other

robots in order to follow its own route for some

period of time. After the expiration of this time,

these other robots return to patrol their own routes.

The work of this program (for only two robots

supposedly available) is depicted in Fig. 13.

Fpoints=(x4y3,x7y3,x9y7,x5y10,x2y6),

Fpoints=(x11y4,x11y8,x13y10,x15y8,x14y3),

.............................;

or(direct#all; grasp(Nmark==; Nmark=1));

Np=Fpoints;

repeat(

WHERE=Np:1; Np&=Np:1; Np:1=nil; Fp=Np;

free(?infraredCheck; USER=alarm_WHERE;

direct#any; Np=Fp; TIME+=1800;

Np=Fpoints))

x13 y10

x7 y3

x5 y10

x2 y6

x4 y3

x9 y7

x11 y4

x11 y8

x14 y3

x15 y8

Robot.1

Robot.2

Alarm !

Robot.2

waypoints

Route of

Robot.1

Route of

Robot.2

Figure 13: Cooperative region patrol by two mobile robots

13 DISTRIBUTED ROAD AND AIR

TRAFFIC MANAGEMENT

Distributed computer networks working in WAVE-

WP and covering the space to be controlled can be

efficiently used for both road and air traffic

management. The WAVE model provides

simultaneous tracking of multiple objects in PW by

mobile intelligence spreading in the VW, via

computer networks (Sapaty, Corbin, Seidensticker,

1996; Sapaty, 1999; Sapaty, Klimenko, Sugisaka,

2004).

Assigning personal active mobile code to each

object under control brings high flexibility to the

distributed control system, with parallel and

cooperative tracking of multiple objects and making

non-local decisions, along with runtime optimization

and routing. This may especially be important in

crisis situations, where a priori flight schedules

become useless, and the distributed management

infrastructure is damaged. The surface roads can be

destroyed too, and traffic routing may need to be

fully dynamic, in order to reach destinations by

individual vehicles in suitable time. Related

distributed management scenarios with mobile

intelligence had been demonstrated live in WAVE

via the Internet in different projects.

As an elementary example, the following simple

mobile program (using an external procedure

Fobject?seen) seizes control of a certain

moving object, and then follows its PW movement,

propagating itself via the computer network. It

launches subordinate search agents in neighboring

computers (associated, say with neighboring radar

stations) unless the disappeared object is found

elsewhere and followed again. The work of this

program is shown in Fig. 14.

Fobject=TRW562;

repeat(

repeat(Fobject?seen; TIME+=10);

any#any; Fobject?seen)

Many such objects can be simultaneously seized

and controlled by mobile intelligences in WAVE-

WP, and any payload can be added to the program

above, including detailed studying of the behavior of

controlled objects, with their possible subsequent

rerouting or destruction.

Moving aerial

object

Self-

replicating

tracking

agent

Tracking

failure

Tracking

failure

Figure 14: Distributed tracking of moving objects.

WAVE-WP (WORLD PROCESSING) TECHNOLOGY

14 AUTONOMOUS DISTRIBUTED

COGNITIVE SYSTEMS

Cognitive systems belong to the most advanced class

of intelligent systems -- the ones aware of what they

are doing. While cognitive systems include reactive

and deliberative processes, they also incorporate

mechanisms for self-reflection and adaptive self-

modification.

The WAVE-WP paradigm allows for the

description of interacting deliberative, reactive, and

reflective processes on a semantic level, representing

the whole mission rather than individual robots

(Sapaty, Kawamura, Sugisaka, Finkelstein, 2004).

This provides new degrees of freedom for

autonomous robotic teams, where collective

behavior of robots emerges as a derivative of

parallel and distributed interpretation of WAVE-WP

language, in the united physical and virtual world.

The mission-to-hardware mapping process may

be fully distributed, not requiring central resources,

and each robot may happen to be involved at any

level of the distributed command and control process

in any moment of time, as shown in Fig. 15. Failed

robots can be automatically substituted at runtime

without loss of the overall mission integrity.

Local deliberation

Distributed

reaction

Global

distributed

reflection

Distributed

deliberation

Local

reaction

Sensors

Distributed

Knowledge

Network

Robot.1

Robot.2

Robot.3

Robot.4

Figure 15: Fully distributed cognitive system.

15 DISTRIBUTED INTERACTIVE

SIMULATION

Having full control over distributed worlds,

including their runtime creation and modification,

WAVE-WP allows for highly efficient, scalable,

distributed simulation of complex dynamic systems,

like battlefields, in open computer networks, using

potentially unlimited number of computers working

together (Sapaty, Corbin, Borst, 1995).

Due to full distribution of the simulated space

and entities operating in it, there is no need to

broadcast changes in terrain or positions of moving

entities to other computers, as usual. Each simulated

entity operates in its own (current) part of the

simulated world, within the range of its sensors,

communicating locally with other such entities,

exactly as in the real world. The entities can move

freely through the simulated space (and between

computers, if needed).

Using volatile virus-like spatial algorithms in

WAVE-WP, a fully dynamic terrain can be

effectively modeled in a distributed space. Its parts

like clouds, floods, smog, mountains, landslides, and

craters can grow, move and spread seamlessly

between computers (Darling, Sapaty, Underhill,

1996; Sapaty, 1999), as shown in Fig. 16a,b.

igure 16: Seamless distributed dynamic terrain modeling.

ICINCO 2004 - INTELLIGENT CONTROL SYSTEMS AND OPTIMIZATION

100

16 INTELLIGENT GLOBAL

DEFENSE AND SECURITY

INFRASTRUCTURES

WAVE-WP can also be effectively used in a much

broader scale, especially for the creation of

intelligent national and international infrastructures

of different natures, widely using automated and

fully automatic control and advanced robotics. Such

global systems may effectively solve the problems

of distributed air defense, where multiple hostile

objects penetrating country’s air space can be

simultaneously discovered, chased, analyzed, and

destroyed using a computerized networked radar

system as a collective artificial brain operating in

WAVE-WP, as shown in Fig. 17 (see also Sapaty,

Klimenko, Sugisaka, 2004).

Permanent

channels

Dynamic and

casual channels

Sea, undersea, aerial

or space units

Spatial solution 1

Spatial solution 2

Figure 17: Distributed infrastructures and spatial solutions

Within the unified command and control

infrastructures, provided by the technology, different

types of unmanned air vehicles may be used, for

example, as possible mobile sensor, relay, or even

air traffic management stations, supplementary to

the ground ones, especially when the latter get

damaged or operate partially.

In other non-local applications, WAVE-WP,

using worldwide computer networks, may

effectively discover and trace criminals and their

distributed organizations, penetrate into malicious

infrastructures, studying and eliminating them, with

possible additional involvement of special hardware

and troops (Sapaty, 2002).

17 CONCLUSIONS

WAVE-WP allows for a more rational and universal

integration, management, simulation, and recovery

of large complex systems than many other

approaches – by establishing a higher level of their

vision and coordination, symbolically called “over-

operability” (Sapaty, 2002) versus (and in

supplement to) the traditional “interoperability”.

Distributed system management and

coordination scenarios in WAVE-WP are often

orders of magnitude simpler and more compact than

usual, due to high level and spatial nature of the

model and language. They often help us to see the

systems and solutions in them as a whole, avoiding

tedious partitioning into parts (agents) and setting

their communication and synchronization.

These and other routines are effectively shifted

to the efficient automatic implementation by

dynamic networks of WAVE-WP interpreters.

Traditional software and hardware agents are being

requested, created, and have sense only when

required in certain moments of time, during the

spatial development of self-evolving (conceptually

agent-less) parallel mission scenarios.

A detailed description of the WAVE-WP model

and its extended applications can soon be available

(Sapaty, 2004).

REFERENCES

Borst, P.M., 2002. An architecture for distributed

interpretation of mobile programs. PhD Thesis, also

published as a book, ISBN 3-8316-0103-8.

Darling, J.C.C., Sapaty, P.S., Underhill, M.J., 1996.

Distributed virtual reality: a fully dynamic approach.

In Proc. 15th Workshop on Standards for the

Interoperability of Distributed Simulations. IST UCF,

Orlando, FL.

Gonzalez-Valenzuela, S., Vuong, S.T., 2002. Evaluation

of migration strategies for mobile agents in network

routing. In Proc. The 4th International Workshop

MATA'02, Barcelona, Spain.

Sapaty, P.S., Zorn W., 1991. The WAVE model for

parallel processing and its application to computer

network management. In International Networking

Conference INET'91, Copenhagen.

Sapaty, P.S., 1993. A distributed processing system.

European Patent No. 0389655, European Patent Office.

Sapaty, P.S., Corbin, M.J., Borst, P.M., 1995. Towards the

development of large-scale distributed simulations. In

Proc. 12th Workshop on Standards for the

Interoperability of Distributed Simulations. IST UCF,

Orlando, FL.

Sapaty, P.S., Corbin, M.J., Seidensticker, S., 1996. Mobile

intelligence in distributed simulations. In Proc. 14th

Workshop on Standards for the Interoperability of

Distributed Simulations. IST UCF, Orlando, FL.

Sapaty, P.S., 1999. Mobile processing in distributed and

open environments, John Wiley & Sons. ISBN:

0471195723, New York.

WAVE-WP (WORLD PROCESSING) TECHNOLOGY

101

Sapaty, P.S., 2000. High-level spatial scenarios in WAVE.

In Proc. The Fifth International Symposium on

Artificial Life and robotics, (AROB 5th’00), Beppu,

Japan.

Sapaty, P.S., 2000a. Basic distributed control model and

technology for mobile crisis reaction forces and their

united air defense. In Proc. NATO Symosium on

System Concepts for Integrated Air Defense of

Multinational Mobile Crisis Reaction Forces.

Valencia, Spain.

Sapaty, P., Sugisaka, M., 2001. Towards the distributed

brain for collectively behaving robots. In Proc.

International Conference on Control, Automation and

Systems, ICCAS. Cheju National University, Jeju

Island, Korea.

Sapaty, P.S., 2002. Over-operability in distributed

simulation and control. The MSIAC's M&S Journal

Online. Winter Issue, Volume 4, No. 2, Alexandria,

VA,

http://www.msiac.dmso.mil/journal/WI03/sap42

_1.html

Sapaty, P.S., Sugisaka, M., 2002. A Language for

programming distributed multi-robot systems. In Proc.

The Seventh International Symposium on Artificial

Life and Robotics (AROB 7th ‘ 02). Beppu, Japan.

Sapaty, P., Sugisaka, M., 2002a. Universal distributed

brain for mobile multi-robot systems. In book

Distributed Autonomous Robotic Systems. H. Asama,

T. Arai, T. Fukuda, and T. Hasegava (Eds.), Springer-

Verlag Tokyo, SPIN: 10869189.

Sapaty, P., Sugisaka, M., 2003. Optimized space search by

distributed robotic teams. In Proc. Symposium

Unmanned Systems 2003. Baltimore Convention

Center, USA.

Sapaty, P., Klimenko, V., Sugisaka, M., 2004. Dynamic

air traffic management using distributed brain concept.

Mathematical Machines and Systems. ISSN:1028-

9763, No.1.

Sapaty, P., Kawamura, K., Sugisaka, M., Finkelstein, R.,

2004. Towards fully distributed cognitive systems.

In Proc. The Ninth International Symposium on

Artificial Life and Robotics (AROB 9th ‘ 04). Beppu,

Japan.

Sapaty, P. S., 2004. Ruling Distributed Dynamic Worlds,

to be publ. by John Wiley & Sons, New York.

ICINCO 2004 - INTELLIGENT CONTROL SYSTEMS AND OPTIMIZATION

102