a dynamic topology control algorithm for wireless sensor … · ad hoc now 2015 june 29th, 2015...

A Dynamic Topology ControlAlgorithm for Wireless Sensor

NetworksGerry Siegemund, Volker Turau and Christoph Weyer

Ad Hoc Now 2015June 29th, 2015

TUHHTUHHInstitute of TelematicsInstitute of TelematicsHamburg University of TechnologyHamburg University of Technology

IntroductionIntroduction

Topology Control

J Topology control algorithms (TCAs) for wireless sensornetworks� dynamically form a graph representing communication

network to be used by applications� while maintaining local and global graph properties such as

I high link qualitiesI low node degree,I k-connectedness,I small diameter

J TCAs aim at reduced interference, less energyconsumption, and increased network capacity

* Essence: TCAs perform a deliberate choice on each node’sneighbors (physical topology is thinned out)

G. Siegemund, V. Turau, and Ch. Weyer A Dynamic Topology Control Algorithm for Wireless Sensor NetworksG. Siegemund, V. Turau, and Ch. Weyer A Dynamic Topology Control Algorithm for Wireless Sensor Networks 11


Topology Control

0

12

3

4

5

6

7

8

9

10

11

12

13

14

15

17

16

19

18

21

20

23

22

25

24

27

26

29

28

31

30

34

35

32

33

38

39

36

37

42

43

40

41

46

47

44

45

51

50

49

48

55

54

53

52

59

58

57

56

63

62

61

60

68

69

70

71

64

65

66

67

76

77

78

79

72

73

74

75

85

84

87

86

81

80

83

82

93

92

95

94

89

88

91

90

102

103

100

101

98

99

96

97

110

111

108

109

106

107

104

105

119

118

117

116

115

114

113

112

127

126

125

124

123

122

121

120

137

136

139

138

141

140

143

142

129

128

131

130

133

132

135

134

152

153

154

155

156

157

158

159

144

145

146

147

148

149

150

151

171

170

169

168

175

174

173

172

163

162

161

160

167

166

165

164

186

187

184

185

190

191

188

189

178

179

176

177

182

183180

181

205

204

207

206

201

200

203

202

197

196

199

198

193

192

195

194

220

221

222

223

216

217

218

219

212

213

214

215

208

209

210

211

224

Original Topology

�

��

�

�

�

�

�

�

�

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

Controlled Topology



Topology Control Algorithms

J TCAs based on transmission power control� Raising transmission power

ù increases number of direct communication partners÷ reduces number of concurrent transmissions

J Many TCAs disregard÷ limited resources in WSNs÷ disregard fluctuations over time, i.e., they keep a computed

topology forever

* Agility and stability of TCA have important influence of QoSfor applications



Existing Work

New RETC[5] EELTC[17] STC[14] CONREAP[8] LEEP[4] XTC[18]

Distributed YES YES NO YES NO YES YES

Dynamic YES YES NO NO NO YES NO

Stable YES NO YES YES YES NO YES

Agile YES YES NO NO NO YES NO

Memory de-NO YES - YES - NO YES

pends on Δ


RequirementsRequirements

Notation

J Model: Undirected graph G(t) = (V , E(t))� E(t) set of edges at time t� (u, v) ∈ E(t) iff u and v can communicate directly at time t

J TCA dynamically computes subgraph Gc(t) = (V , Ec(t)) ofG(t)� Ec(t) ⊆ E(t)

* TCA aims to smoothly update Ec(t) such that a list ofcriteria are maximized while avoiding oszillation



Criteria for Gc(t)

1. Link quality� Q(e) ≥ Qmin for each e ∈ Ec(t) and any time t

2. Symmetry� Bidirectional communication

3. Connectivity4. Bounded degree

� Each node in Gc(t) should have at most CN neighborsregardless of the size of V or E(t)

5. k-Spanner� Distance between two nodes in Gc(t) should be at most

k -times the corresponding distance in G(t)* More practical criterion: Minimize average length of paths

between any pair of nodes



Criteria for Gc(t)

6. Stability� If e ∈ Ec(t0) and Q(e) ≥ Qmin at time t0 then e should

remain in Ec(t) for at least t ∈ [t0, t0 + TN ]� e should remain in Ec(t) for t ≥ t0 as long as Q(e) ≥ Qmin

and other criteria are fulfilled7. Agility

� If Ec(t) does not satisfy criteria 1 to 5 and there existse ∈ E(t)∖Ec(t) that can improve one criterion then include einto Ec(t)

� If there exist e ∈ Ec(t0) with Q(e) < Qtol then e should bediscarded from Ec during [t0, t0 + TN ].


AlgorithmAlgorithm

Topology Control Algorithm

J Basis of TCA: Link quality estimator (LQE)J LQE periodically computes for each link e = (u, v) a quality

value Q(e)� Only links with quality above Qmin are considered useful

J Each node maintains three lists of fixed size� N: current neighborhood� S: standby list� A: assessment list

J Membership in lists depends on above defined criteriaJ Promotion: From not listed å A å S å NJ Agility and stability is supported by time-triggered transitions

among lists (timers TA and TS)


AlgorithmAlgorithm

Topology Control Algorithm

A S

notlisted N

|A|<

CA

on timeoutTA,

Q < Qmin

Q ≥ Qmin, |S| < CS

or force replacement

ontim

eout T S

orrep

lacem

entfro

mA

Q < Qtol

or replacement from S

Q≥

Qm

in ,|N|<

CN

orforce

replacement

Transition DiagramG. Siegemund, V. Turau, and Ch. Weyer A Dynamic Topology Control Algorithm for Wireless Sensor NetworksG. Siegemund, V. Turau, and Ch. Weyer A Dynamic Topology Control Algorithm for Wireless Sensor Networks 99

AlgorithmAlgorithm

Promoting nodes from A to S

upon expiration of timer TA for p ∈ A doA.remove(p);if p.Q ≥ Qmin then

if |S| < CS thenS.add(p);

elseq := minQ S;if q.Q < p.Q then

S.remove(q);S.add(p);


AlgorithmAlgorithm

Promoting nodes from S to N

with Period TN dofor all p ∈ S in descending order do

if p.Q ≥ Qmin thenif |N| < CN then

N.add(p);S.remove(p);

elseq := replacementCandidate(N, p);if q ̸= null then

N.remove(q);N.add(p);S.remove(p);


AlgorithmAlgorithm

Replacement Candidates

J A node p ∈ S with p.Q ≤ Qmin replaces a node q ∈ N if thisimproves the local topology defined by N

J Goal: Optimize criteria 1, 2, 3, and 5J Challenge: Criteria of global natureJ Criterion 5: Minimize average path lengths

� Each node v considers its 2-hop neighborhood Cv� Define Len(X ) =

∑︀w∈X dist(v , w)

� Find node q such that Len(Cv ∖{q} ∪ {p}) is minimal� If Len(Cv ∖{q} ∪ {p}) < Len(Cv ) then replace q by p� Challenge: How to compute Len(Cv ∖{q} ∪ {p})?

I Distributed algorithm based on weights


AlgorithmAlgorithm

Minimizing Paths Length

v

u4

u1u2

u3

p

p replaces u4


AlgorithmAlgorithm

Connectedness

v

u1u2

p

u3u4

connected component “3”

connected component “1”

From v ’s point of view u1 resp. u2 are as good as p


AlgorithmAlgorithm

Achieving Connectedness

J Preference: A node prefers new neighbors from a differentconnected component

J Distributed algorithm labels nodes with the smallestidentifier of node within same connected component of Gc

J Challenge:� Node with smallest identifier leaves component because

edge is removed or failure of node� Solution requires knowledge of upper bound of network

diameter


EvaluationEvaluation

Evaluation

J Comparison with two TCAs: XTC, LEEPJ Implementation with OMNet++ and the MiXiM frameworkJ Regular and random node placementsJ Network size and density was variedJ Used LQE: HoPSJ Node addition and removal was considered



Evaluation: Dense Setup

0 100 200 300 400 500 600 700 800 900 1 0000.00

0.20

0.40

0.60

0.80

1.00

time [rounds]

conn

ecte

dnes

s

upper quartilmedianlower quartil

0 100 200 300 400 500 600 700 800 900 1 0000

20

40

60

80

100

link

chan

ges

link changes



Evaluation: Comparision with XTC

communication range = 2

16 25 36 49 64 81 1000

5

10

communication range = 3

16 25 36 49 64 81 1000

102030

number of nodes in networknumber of nodes in network

chos

en n

umbe

rof

nei

ghbo

rs XTC maximumXTC averageCN minimum

Number of required neighbors to achieve connected network



Evaluation: Comparision with LEEPExecution of a spanning tree algorithm on top of LEEP andproposed TCA

0

100

200

erro

rsin

tree

0 50 100 150 200 250 300 350 400 450 5000

100

200

300

time [rounds]

loop

coun

terro

rs

LEEP

0

100

200

erro

rsin

tree

0

100

200

300

time [rounds]lo

opco

unte

rrors

0 50 100 150 200 250 300 350 400 450 500

Proposed TCA


SummarySummary

Summary

J Novel dynamic TCAJ Goal: Optimizing global properties of topology while

keeping number of neighbors boundedJ Small footprintJ Simulations show that TCA provides a good compromise

between agility and stabilityJ Proposed TCA is used as basis for a self-stabilizing publish

subscribe system


A Dynamic Topology ControlAlgorithm for Wireless Sensor

NetworksGerry Siegemund, Volker Turau and Christoph Weyer

Ad Hoc Now 2015June 29th, 2015

Volker TurauHamburg University of Technology

Phone +49 40 42878-3530

e-Mail [email protected]

http://www.ti5.tu-harburg.de/staff/turau

TUHHTUHHInstitute of TelematicsInstitute of TelematicsHamburg University of TechnologyHamburg University of Technology

mailto:[email protected]

http://www.ti5.tu-harburg.de/staff/turau

a dynamic topology control algorithm for wireless sensor … · ad hoc now 2015 june 29th, 2015...

Documents