Towards proactive computing systems

Leo Motus and Merik Meriste

Tallinn University of Technologyand University of Tartu

21st January 2008 1Laboratory for Proactive Technologies

Alternative research approaches

• Divide and rule approach (extensive research)

e.g. in computer science: focusing on components -- methods, tools, and theory developed for Turing computation; the mainstream research from1960 up to 1980 and further

• Understanding the essence (intensive research)

e.g. in computer science: focusing on interaction of components – discussions on the essence of super-Turing computation, respective methods, tools, and applications; resumed in mid 1970-es and gaining popularity due to large number of ubiquitous computing applications

Compare also with the message by P. Anderson (1972) ”More is different”

21st January 2008 Laboratory for Proactive Technologies 2

Explicit and implicit interactions

21st January 2008 Laboratory for Proactive Technologies 3

Environment

Computer 1

Computer 2

Computer 3

Computer 4

Laboratory for Proactive Technologies 4

A generic building block of the world

[J.Liu, K.C. Tsui Toward nature-inspired computing. 59-64, CACM, 2006]21st January 2008

Laboratory for Proactive Technologies 5

action

situational awareness

Situation aware non-algorithmic

Non-algorithmic

Algorithmic

Algorithmic Partially controlled

Proactive Topological time

Topological and metric time

+ other notions of SA

Conventional models of

computationInteractivecomputing

Situation aware interactive computing

Models of computation

situational awareness

interaction

21st January 2008

Laboratory for Proactive Technologies 6

Computing in a proactive system

Sequential Interaction Machines

Persistent Turing machine

Multi-stream Interaction Machines

Situation-aware, Multi-stream Interaction Machines

Algorithmically parallel execution

Time-constraint algorithm theory

Algorithmic approach (Turing machines, Petri nets, etc)

e.g. temporal logic, timed automata, timed Petri nets

21st January 2008

Laboratory for Proactive Technologies 7

Sequential interaction machine

is based on the concept of Persistent Turing Machine

(PTM):

• PTM is a history-aware Turing machine

• “Turing machines, transition systems, and interaction” Information and Computation 194 (2004) 101-128, by Goldin D, et alii.

output (o1) output (o2) output (o3) output (o4)

initial state

input (i0) input (o1,i1) input (o1,o2,i2) input (o2,o3,i3)

…

21st January 2008

Laboratory for Proactive Technologies 8

Multi-stream interaction machine

output (o1) output (o2) output (o3) output (o4)

initial state

input (i0) input (o1,i1) input (o1,o2,i2) input (o2,o3,i3)

…

output (1) output (2) output (3) output (o4)

initial state

input (i) input (1,2) input (1,2,3) input (2,3,4)

…

21st January 2008

21st January 2008 Laboratory for Proactive Technologies 9

Multi-streams and Q-model (1)

pi: T (pi ) x dom pi → val pi

pj : T (pj ) x dom pj → val pj

ij: val pi x T (pi ) x T (pj ) → proj val pi

dom pj

K (ij ,t) T (pi ) , t T (pj )

T (pi ) = {ti0, ti

1, ti2, … ti

n, ti(n+1),

… }

The definition of K (ij ,t) depends on

whether T (pi ) T

(pj ) , or

T (pi ) → T (pj ), or

T (pi ) T (pj )

21st January 2008 Laboratory for Proactive Technologies 10

Multi-streams in Q-model (2)

pi

pj

ij

{in1i, in2

i, …, inni, … } →

→ {out1i, out2

i, out3i, …, outn

i, out(n+1)i, …}

{in1j, in2

j, …, inkj, in(k+1)

j … } →

{out1j, out2

j, …, outkj, out(k+1)

i, …}

K (ij ,t)

Laboratory for Proactive Technologies 11

…...

p-PTM1

…

p-PTMj

p-PTMn

In-s1

In-sj

In-sn

Out-s1

Out-sk

Out-sm

i-PTM1

i-PTM2

i-PTM3

Multi-stream interaction machine

21st January 2008

Laboratory for Proactive Technologies

A multi-stream application

Squad

Far away HQFar away HQInteractive map

Interactive mapSensor networks

Portable unmanned vehicles

ReconnaissanceLocal HQ

Far away HQ

Interactive map

Interactive map

Interactive map

Interactive map

21st January 2008 12

Some results obtained in CDC

• Applications of and development tools for multi-agents– Generic agent generator and engine (KRATT)– Interactive agent-based digital maps– Pilot applications of smart dust

• Ad hoc networks – research and experiments with heterogeneous and mobile nodes, dynamically changing topology and interaction pattern

• Research into models of computation– Taxonomy of computation models– Describing models by situation-sensitive stream

functions

21st January 2008 Laboratory for Proactive Technologies 13

A selection of publications (1)

Models of computation

– M. Meriste, L. Motus (2002) „On models for time-sensitive interactive computing”, Intern. Conference on Computational Science, LNCS 2329, Springer, 156-165

– L. Motus, M. Meriste, W. Dosch (2005) „Time-awareness and Proactivity in Models of Interactive Computation”, Electronic Notes in Theoretical Computer Science 141, 69-95,

– Dosch, W.; Meriste, M.; Motus, L. (2007). “Enriching Interactive Components with Again Commands”. Proceedings of the 2007 IEEE International Conference on Electro/Information Technology. IEEE Computer Society, 200 - 205

21st January 2008 Laboratory for Proactive Technologies 14

A selection of publications (2)

Multi-agent systems– L. Motus, M. Meriste, T. Kelder, J. Helekivi, V. Kimlaychuk (2003) „A

test-bed for time-sensitive agents – some involved problems”, 9th IEEE Intern. Conf. on Emerging Technologies and Factory Automation, Lisbon,, vol. 2, 645-651

– M. Meriste, L. Motus, T. Kelder, J. Helekivi (2005) „C# templates for time-aware agents” , 7th International Conference on Enterprise Information Systems, Miami, vol.4, 247--250

– M. Meriste, L. Motus, T. Kelder, J. Helekivi, A. Marandi, J. Preden (2005) „Location awareness of information agents”, in Advances in Databases and Information Systems, Springer, 199 – 208.

21st January 2008 Laboratory for Proactive Technologies 15

A selection of publications (3)

Ad hoc networks– L. Motus, M. Meriste, J. Preden (2006) „Network Enabled

Capabilities – Grassroots Perspectives”, NATO RTO/IST Symposium on Dynamic Communication Management, 16/1 – 16/13

– J. Preden (2006) „Communication area based positioning”, in: Mobile Ad-hoc and Sensor Systems (MASS), The 3rd IEEE International Conference on Mobile Ad-hoc and Sensor Systems, 336-347

– J. Preden, M. Sarkans, Otto T. (2007) “Diagnostics of Machining and Assembly Systems by Networked Motes”, Machining Engineering, vol.7 no 1-2, 68-77

21st January 2008 Laboratory for Proactive Technologies 16

Future engagements

• A joint research project “Proactivity in the artificial world” 2008 – 2013 , involving researchers from TUT and UT and comprising– Situation-aware, multi-stream computation and

multi-agent systems – Self-organising systems, and other self-X systems – Ad hoc networks– Application prototypes (smart buildings, monitoring

and diagnostic systems, etc )– Behavioural robotics (together with institute of

mechatronics and institute of machinery)• Network enabled capability project (2008 – 2010), and

other collaboration with NATO RTO• Emergent behaviour in complex systems – decision

pending

21st January 2008 Laboratory for Proactive Technologies 17