neuron inspired collaborative transmission in wsns

Information Systems Architecture Science Research Division

Neuron inspired collaborative transmission in WSNsMobiquitous 2011, 06.12.–09.12., Copenhagen, Denmark

Stephan Sigg, Predrag Jakimovski, Florian Becker, Hedda R. Schmidtke,Alexander Neumann, Yusheng Ji, Michael BeiglThe 8th International ICST Conference on Mobile and Ubiquitous Systems: Computing,

Networking and Services, MobiQuitous 2012, 06.12.–09.12.2011, Copenhagen, Denmark

Motivation Neuron systems in WSNs Collaborative transmission for neuronal structures Results Conclusion and future work

Smart Spaces

Stephan Sigg | Neuron inspired collaborative transmission | 2 Information Systems Architecture Science Research Division


Smart Spaces

Smart Spaces are ...

Sensing environment

Traditionally visual and audio system

Sensing and acting as a living being



Smart Spaces



Smart Spaces

Smart Space ?



Smart Spaces

How can we command the enormouspotential of ubiquitously available sensing

and computing power?



Motivation



Outline

Motivation

Neuron systems in WSNs

Collaborative transmission for neuronal structures

Results

Conclusion and future work




xyw

z

xz

ww

y

1

kn

m

0

11 11w11

(1) (2)

0km

(2)

(1)

Neural network with one hidden layer:

yk(−→x ,−→w ) = σ

M∑j=1

w(2)jk h

(D∑i=1

w(1)ij xi + w

(1)0j

)+ w

(2)0k

Stephan Sigg | Neuron inspired collaborative transmission | 12 Information Systems

Architecture Science Research Division



No straightforward mapping to wireless networks

NN are inherently clocked

Noise and interference due to wirelesscommunication



Outline

Motivation



Results





ζsum =ι∑

i=1

<(e j(fi t+γi )

)



Outline

Motivation



Results




Results



Results

Dendrites 180, 240, 300, 500Synapses 2, 3, 4, 5Transmit frequency 2.4 GHzTransmission power 1 mWAntenna gain 0 dB (Rx/Tx)Random noise power −103 dBm

Pathloss Ptx

(λ

4πd

)2GtxGrx



Results

180 240 300 5000

0.2

0.4

0.6

0.8

1

Count of dendrite nodes

Med

ian

BE

R

Deterioration of BER with increasing count of dendrite nodes



Results

2 3 4 50.9

1

1.1

1.2

1.3

1.4

1.5

1.6

Count of synapses nodes

BE

R d

ete

rio

ratio

n n

orm

alis

ed

on

th

e o

ptim

um

va

lue

Deterioration of BER with increasing count of synapse nodes

180 dendrite nodes

240 dendrite nodes300 dendrite nodes




Conclusion

NN-overlay over a wireless sensor network

Communication among nodes via distributedbeamforming

Computation of arbitrary functions

Smart space implementation possible




Future work

Instrumentation with hardware nodes

Computation of relevant functions

Validation in very large networks

Reduce synchronisation time and accuracy


Questions?

Stephan [email protected]


neuron inspired collaborative transmission in wsns

Documents