+ All Categories
Home > Documents > 5 saso2012-presentation

5 saso2012-presentation

Date post: 14-Jul-2015
Category:
Upload: ale-cignetti
View: 113 times
Download: 0 times
Share this document with a friend
Popular Tags:
19
Motivation Scenario Algorithm Evaluation A Cognitive-Inspired Model for Self-Organizing Networks ASENSIS 2012 Daniel Borkmann 0 Andrea Guazzini 12 Emanuele Massaro 3 Stefan Rudolph 4 0 Communication Systems Group, ETH Zurich, Switzerland 1 Institute for Informatics and Telematics, National Research Council, Pisa, Italy 2 Department of Psychology, University of Florence, Italy 3 Department of Informatics and Systems, University of Florence, Italy 4 Organic Computing Group, University of Augsburg, Germany 10th September, 2012 Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 1 / 19
Transcript
Page 1: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

A Cognitive-Inspired Model for Self-OrganizingNetworks

ASENSIS 2012

Daniel Borkmann0 Andrea Guazzini12 Emanuele Massaro3

Stefan Rudolph4

0Communication Systems Group, ETH Zurich, Switzerland1Institute for Informatics and Telematics, National Research Council, Pisa, Italy

2Department of Psychology, University of Florence, Italy3Department of Informatics and Systems, University of Florence, Italy

4Organic Computing Group, University of Augsburg, Germany

10th September, 2012

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 1 / 19

Page 2: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Large Scale Network

Source: http://de.wikipedia.org/w/index.php?title=Internet&oldid=107566536

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 2 / 19

Page 3: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Motivation

Large Scale Networks emergeInternetPervasive ComputingOften used: Overlay networks

Problems of overlay networksStructured: Hard without global informationUnstructured: No optimization of network structure

IdeaSelf-optimization of an overlay networkThrough a cognitive-inspired model

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 3 / 19

Page 4: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Scenario

Connected network of n nodesStatic, nodes don’t disappear or appearEach holdes one item (e.g. a service or data)Each wants to retrieve items with respect to its energyEach has a limited number of links from 1 . . .mEach node can change its links

Optimization problems: change links in order toRetrieve all items with the minimum number of hopsMaximize the number of items with a fixed amount of hops

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 4 / 19

Page 5: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Cognitive-Inspired Hub DetectionDiffusion and Competitive Interaction

At startA is the adjacency matrixEvery node i has a state vector Si (short term memory)S(k)

i is the probability that node i belongs to community kEvery node belongs to its own community

Update of the state vectors

S(t + 12 ) = mSik (t)+(1−m)∑j AijSjk (t)

S(t +1) =Sα

ik (t+12 )

∑j Sαij (t+

12 )

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 5 / 19

Page 6: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Cognitive-Inspired Hub DetectionDiffusion and Competitive Interaction

EntropyEi =−∑(Sj · log(Sj))Plateaus show sub-clustersWhen curvature changes sign, save information in temporarymemory box

0.00

1.00

2.00

3.00

4.00

5.00

6.00

0 5 10 15 20 25 30 35 40

Entr

op

y

Time

Shannon entropy of information

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 6 / 19

Page 7: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Cognitive-Inspired Hub DetectionCognitive Dissonance

Cognitive concept found by social psychologistsReduces conflicting cognitionsCreates consistent belief system

Here: Dij :=∑k |Sik−Sjk |

2

Interesting for adaption of α :

If∣∣∣E t−1

i +Dt−1i

Ki

∣∣∣− ∣∣∣E ti +Dt

iKi

∣∣∣< ε for more than τ∗ times

Set αi = 1.5|η(0,σ)|+1

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 7 / 19

Page 8: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Cognitive-Inspired Hub DetectionLong Term Memory

Store potential hubs in the Long Term MemoryFind B1 time positions by sorting with respect to first derivativeSort the remaining vectors with respect to the entropyFind the potential hubs in the state vectors

Use Long Term Buffer of size B2

The last B2 sets of size B1 are stored (bounded rationality)This creates a (B1,B2) matrix

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 8 / 19

Page 9: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Rewiring

With help of this Long Term Memory, we can can create a “hublist" for each node

Rewiring steps:1. Find the weakest X% of the nodes2. Choose Y% of the nodes at random3. Each of these nodes closes a connection to a non-hub4. Each of these nodes opens a new connection to a potential hub

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 9 / 19

Page 10: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Network Example

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 10 / 19

Page 11: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Network Example

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 11 / 19

Page 12: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Network Example

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 12 / 19

Page 13: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Network Example

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 13 / 19

Page 14: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Network Example

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 14 / 19

Page 15: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Numerical SimulationScenarios

1. Maximization of the reachable items of the nodesThe energy (hops) is limitedWeakest nodes: Minimum number of items

2. Minimization of used energyAll item will be reached in every stepWeakest nodes: Maximum number of energy

Randomized AlgorithmFor comparisonDoes not use hub list

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 15 / 19

Page 16: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Numerical SimulationParameters

Number of nodes n

Mean connectivity

Mean extra connectivity

Number of unique items I

Number of items to retrieve Imax

Hub detection: m,α

Rewiring

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 16 / 19

Page 17: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

EvaluationResults for maximization of retrieved items

33.00

33.50

34.00

34.50

35.00

35.50

36.00

36.50

37.00

37.50

38.00

0 200 400 600 800 1000

Mean fi

tness

(I c

urr

)

Round

Topology Optimization

Rewiring, cognitive approach Rewiring, randomized approach

Setting: Mean over 50 runs, n = 200, mean_conn= 4, extra_conn= 4, I = 50, Imax = 45, rw_weak= 0.09, rw_rand= 0.03

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 17 / 19

Page 18: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

EvaluationResults for the minimization of energy

148.00

150.00

152.00

154.00

156.00

158.00

160.00

162.00

164.00

166.00

0 100 200 300 400 500 600 700 800 900 1000

Mean e

nerg

y

Round

Topology Optimization

Rewiring, cognitive approach Rewiring, randomzied approach

Setting: Mean over 50 runs, n = 200, mean_conn= 4, extra_conn= 4, I = 50, Imax = 45, rw_weak= 0.09, rw_rand= 0.03

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 18 / 19

Page 19: 5 saso2012-presentation

Motivation Scenario Algorithm Evaluation

Conclusion

ContributionsDevelopment of a cognitive model for community detectionApplication of information for self-optimization of a networkComparison with a randomized algorithm

Future Work(i) Evaluate the algorithm on a wide range of large scale network

topologies(ii) Localize the decision making of a node when to rewire or not(iii) Introduce more dynamics into items and nodes

Borkmann, Guazzini, Massaro, Rudolph Self-Organizing Networks 19 / 19


Recommended