on the construction of data aggregation tree with minimum energy cost in wireless sensor networks:...

45
On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National Tsing Hua University Tung-Wei Kuo and Ming-Jer Department of Computer Science Hsinchu 30013,

Upload: gladys-wright

Post on 27-Dec-2015

216 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

On the Construction of Data Aggregation Tree

with Minimum Energy Cost in Wireless Sensor Networks:

NP-Completeness and Approximation Algorithms

National Tsing Hua University

Tung-Wei Kuo and Ming-Jer Tsai

Department of Computer Science

Hsinchu 30013, Taiwan, ROC

Page 2: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

2

Motivation (1/2)

• In a wireless sensor network (WSN), a sink collects reports from each sensor periodically. • For example:– In a building– Collecting data like 1. temperature, 2. concentration of CO, 3. power consumed by some equipment.

Page 3: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

3

Motivation (2/2)

• Sensors are equipped with an AC power plug or sustained power supply.• The Octopus X WSN[1] :

[1] Octopus wireless sensor network, http://163.13.128.59/.Our goal is to minimize the total energy cost.

Page 4: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

4

• Data aggregation is a way to reduce the number of transmitted packets. –The energy cost is decreased.– It is performed according to the aggregation ratio, q [2].

[2] C. Liu and G. Cao, “Distributed monitoring and aggregation in wireless sensor networks,” in IEEE INFOCOM, 2010.

The aggregation ratio, q, is the size of report that can be aggregated into 1 packet.

Data aggregation (1/2)

Page 5: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

5

Data aggregation (2/2)

2143

30℃

31℃

q = 3An example

n(transmitted packets) = 5

sink

29℃

31℃28℃32℃

31℃ 29℃

31℃30℃29℃32℃31℃ 28℃

Page 6: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

6

We can simulate this using our model by setting q to large enough (e.g. 4)

Data aggregation model:a special case when q = ∞

• Simulate n(transmitted packets) of MAX query

2143

30℃

31℃

q = 4sink

29℃

31℃28℃32℃

31℃ 29℃

31℃ 32℃

Each node sends exactly one packet31℃ 29℃

31℃30℃29℃32℃31℃28℃Max temperature query

Page 7: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

7

Problem definition• A static routing tree is considered here.• To estimate the energy cost, we consider– Tx, the energy to transmit a packet, and– Rx, the energy to receive a packet.

Given the aggregation ratio q, Tx, and Rx:We want to find an optimal tree to minimize the energy cost.

Page 8: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

8

31℃30℃

31℃

Why does routing structure matter?

2143

30℃

31℃

q = 3sink

29℃

31℃28℃32℃

Tx = 2Rx = 1energy cost = (2+1)⨉5This is a shortest path tree.Let’s see the optimal tree.29℃

29℃31℃

29℃31℃30℃

29℃

29℃32℃31℃ 28℃32℃31℃28℃

Shortest path tree may NOT be an optimal tree.energy cost = (2+1)⨉4

Page 9: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

9

NP-completeness

• This problem is NP-complete.• Idea of the proof:– Does there exist a tree such that every node sends only one packet?

• We will design an approximation algorithm.

Page 10: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

10

Our approximation algorithm

Our Algorithm: Shortest path tree.• It is a 2-approximation algorithm.• Other benefits:1. Distributed implementation.2. Only one input: the network topology.

Page 11: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

11

Page 12: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

12

A new problem –when relay nodes exist

• Relay nodes do not generate reports.• A feasible routing tree only needs to span all non-relay nodes in this problem.

43

sink21

43

sink2A feasible routing tree A relay node.

Page 13: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

13

• Steiner tree and shortest path tree:• Bad news: bad approximation ratios• Good news: perform well on some caseq is small q is largeShortest path treeSteiner tree

We want to combine this 2 advantages

Inspiration(1/2)

Page 14: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

14[3] F. S. Salman, J. Cheriyan, R. Ravi, and S. Subramanian, “Approximating the single-sink link-installation problem in network design,” SIAM J. on Optimization, vol. 11, pp. 595–610, 2000.

• We want a subgraph such that1. The path for each non-relay node is short.2. The number of spanned edges is small.• Salman et al. compute a subgraph that has the above properties [3].

But, the subgraph might not be a tree.

Inspiration(2/2)

Page 15: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

15

Our algorithm:A shortest path tree on Salman’s subgraph• It is a 7-approximation algorithm.• Only one input: the network topology.

Our approximation algorithm

Page 16: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

16

• Using the subgraph, Salman et al. design a 7-approximation algorithm for the Capacitated Network Design (CND) problem.• The CND problem is similar to ours except that … • Difference: the solution may NOT be a tree.

A better approximation algorithm (1/3)

Page 17: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

17

Our algorithm:A shortest path tree on the CND problem’s approximation solutionA better approximation algorithm

(2/3)

For any λ-approximation algorithm of the CND problem, there is a corresponding 2λ-approximation algorithm for our problem.

Page 18: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

18

• When all the report sizes are the same:–We obtain a 5.1-approximation algorithm – It is based on Hassin’s 2.55-CND approximation algorithm [4].

• In other case:–We obtain a 7.1-approximation algorithm for our problem.– It is based on Hassin’s 3.55-CND approximation algorithm [4].

A better approximation algorithm (3/3)

[4] R. Hassin, R. Ravi, and F. S. Salman, “Approximation algorithms for a capacitated network design problem,” Algorithmica, vol. 38, pp. 417–431, 2004.

Page 19: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

19

Page 20: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

20

Simulation• Simulation Settings:– 100 sensors are randomly placed in a 100*100 field– Transmission range = 20– Tx = 2, Rx = 1– Report size = 1 (uniform report size), or 1~5 (non-uniform report size)– Aggregation ratio = 2, 4, 6, …, 50 for uniform report size, and 2, 4, 6, …, 100 for non-uniform report size

• The result is obtained by averaging data of 30 different networks.

Page 21: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

21

Simulation• We will compute a lower bound (LB).• LB = the maximum of 2 other lower bounds1. The optimal value if fractional packets are allowed (min cost flow problem)

• E.g. report size = 5, aggregation ratio = 10 → transmit 0.5 packet, instead of 1 packet2. Minimum number of spanned edges (Steiner tree problem)• We use a 2-approximation algorithm to compute Steiner tree [5].[5] L. Kou, G. Markowsky, and L. Berman, “A fast algorithm for steiner trees,” Acta Informatica, vol. 15, pp. 141–145, 1981.

Page 22: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

300

500

Simulation

-without relay node

22

5 10 15 20 25 30 35 40 45 50Aggregation Ratio

Energy

Cost

200

400

600700800900

1000 Lower Bound: Uniform Report Size

Lower Bound: Uniform Report SizeLower Bound: Uniform Report Size

Page 23: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

300

500

Simulation

-without relay node

23

5 10 15 20 25 30 35 40 45 50Aggregation Ratio

Energy

Cost

200

400

600700800900

1000 Lower Bound: Uniform Report SizeLower Bound: Non-Uniform Report Size

Lower Bound: Non-Uniform Report Size Lower Bound: Non-Uniform Report Size

Page 24: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

300

500

Simulation

-without relay node

24

5 10 15 20 25 30 35 40 45 50Aggregation Ratio

Energy

Cost

200

400

600700800900

1000 Shortest Path Tree: Uniform Report SizeLower Bound: Uniform Report SizeLower Bound: Non-Uniform Report Size

Shortest Path Tree: Uniform Report Size Shortest Path Tree: Uniform Report Size

Page 25: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

300

500

Simulation

-without relay node

25

5 10 15 20 25 30 35 40 45 50Aggregation Ratio

Energy

Cost

200

400

600700800900

1000 Shortest Path Tree: Uniform Report SizeShortest Path Tree: Non-Uniform Report SizeLower Bound: Uniform Report SizeLower Bound: Non-Uniform Report Size

Shortest Path Tree: Non-Uniform Report SizeShortest Path Tree: Non-Uniform Report Size

Page 26: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

300500300

500

Simulation

-without relay node

26

5 10 15 20 25 30 35 40 45 50Aggregation Ratio

Energy

Cost

200

400

600700800900

1000 Shortest Path Tree: Uniform Report SizeShortest Path Tree: Non-Uniform Report SizeLower Bound: Uniform Report SizeLower Bound: Non-Uniform Report Size

The ratios are less than 2

Page 27: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

300

500

Simulation

-without relay node

27

5 10 15 20 25 30 35 40 45 50Aggregation Ratio

Energy

Cost

200

400

600700800900

1000 Shortest Path Tree: Uniform Report SizeShortest Path Tree: Non-Uniform Report SizeLower Bound: Uniform Report SizeLower Bound: Non-Uniform Report Size

The performances are close to the optimums when the aggregation ratio is large

Page 28: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

Simulation

-without relay node

28

5 10 15 20 25 30 35 40 45 50Aggregation Ratio

Energy

Cost

200300400500600700800900

1000Arbitrary Spanning Tree: Uniform Report Size

Shortest Path Tree: Uniform Report SizeShortest Path Tree: Non-Uniform Report SizeLower Bound: Uniform Report SizeLower Bound: Non-Uniform Report Size

Arbitrary Spanning Tree: Uniform Report SizeArbitrary Spanning Tree: Uniform Report Size

Page 29: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

Arbitrary Spanning Tree: Non-Uniform Report Size

Simulation

-without relay node

29

5 10 15 20 25 30 35 40 45 50Aggregation Ratio

Energy

Cost

200300400500600700800900

1000Arbitrary Spanning Tree: Uniform Report SizeArbitrary Spanning Tree: Non-Uniform Report Size

Shortest Path Tree: Uniform Report SizeShortest Path Tree: Non-Uniform Report SizeLower Bound: Uniform Report SizeLower Bound: Non-Uniform Report Size

Arbitrary Spanning Tree: Non-Uniform Report Size

Page 30: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

Simulation

-without relay node

30

5 10 15 20 25 30 35 40 45 50Aggregation Ratio

Energy

Cost

200300400500600700800900

1000Arbitrary Spanning Tree: Uniform Report SizeArbitrary Spanning Tree: Non-Uniform Report SizeThe ratios are big

Shortest Path Tree: Uniform Report SizeShortest Path Tree: Non-Uniform Report SizeLower Bound: Uniform Report SizeLower Bound: Non-Uniform Report Size

Page 31: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

31

Simulation-with relay nodeuniform report size

• Two approximation algorithms here:1. A 7-approxmiation algorithm based on Salman’s approximation algorithm. (Algorithm 1)2. A 5.1-approxmiation algorithm based on Hassin’s approximation algorithm. (Algorithm 2)• We also compare to the performance of Hassin’s algorithm directly, i.e. a non-tree routing structure.

Page 32: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

32

500Simulation-with relay node

uniform report size

5 10 15 20 25 30 35 40 45Aggregation Ratio

Energy

Cost

200250300350400450

Lower Bound Lower Bound

Lower Bound

Page 33: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

33

500Simulation-with relay node

uniform report size

5 10 15 20 25 30 35 40 45Aggregation Ratio

Energy

Cost

200250300350400450

Algorithm 1

Algorithm 1Algorithm 1

Lower Bound

Page 34: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

34

500Simulation-with relay node

uniform report size

5 10 15 20 25 30 35 40 45Aggregation Ratio

Energy

Cost

200250300350400450

Algorithm 1 Lower Bound Algorithm 2

Algorithm 2Algorithm 2

Page 35: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

35

500Simulation-with relay node

uniform report size

5 10 15 20 25 30 35 40 45Aggregation Ratio

Energy

Cost

200250300350400450

Algorithm 1Algorithm 2

The ratios are less than 2

Lower Bound

Page 36: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

36

500Simulation-with relay node

uniform report size

5 10 15 20 25 30 35 40 45Aggregation Ratio

Energy

Cost

200250300350400450

Algorithm 1Algorithm 2Hassin’s Algorithm

Hassin’s AlgorithmHassin’s Algorithm

Lower Bound

Page 37: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

37

500Simulation-with relay node

uniform report size

5 10 15 20 25 30 35 40 45Aggregation Ratio

Energy

Cost

200250300350400450

Algorithm 1Algorithm 2Hassin’s Algorithm

The performances are close

Lower Bound

Page 38: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

38

500Simulation-with relay node

uniform report size

5 10 15 20 25 30 35 40 45Aggregation Ratio

Energy

Cost

200250300350400450

Algorithm 1Algorithm 2Hassin’s AlgorithmShortest Path Tree

Shortest Path TreeShortest Path Tree

Lower Bound

Page 39: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

39

500Simulation-with relay node

uniform report size

5 10 15 20 25 30 35 40 45Aggregation Ratio

Energy

Cost

200250300350400450

Algorithm 1Algorithm 2Hassin’s AlgorithmShortest Path Tree Steiner TreeSteiner TreeSteiner Tree

Lower Bound

Page 40: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

40

500Simulation-with relay node

uniform report size

5 10 15 20 25 30 35 40 45Aggregation Ratio

Energy

Cost

200250300350400450

Algorithm 1Algorithm 2Hassin’s AlgorithmShortest Path Tree Steiner TreeWhen the aggregation ratio is small, shortest path tree performs better

Lower Bound

Page 41: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

41

500Simulation-with relay node

uniform report size

5 10 15 20 25 30 35 40 45Aggregation Ratio

Energy

Cost

200250300350400450

Algorithm 1Algorithm 2Hassin’s AlgorithmShortest Path Tree Steiner TreeWhen the aggregation ratio is large, Steiner tree is betterBoth of them perform well on average case

Lower Bound

Page 42: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

42

500Simulation-with relay node

uniform report size

5 10 15 20 25 30 35 40 45Aggregation Ratio

Energy

Cost

200250300350400450

Algorithm 1Algorithm 2Hassin’s AlgorithmShortest Path Tree Steiner TreeArbitrary Spanning TreeArbitrary Spanning TreeArbitrary Spanning Tree

Lower Bound

Page 43: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

43

Simulation-with relay nodeuniform report size

The result is similar to the previous one.

Non-

Page 44: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

44

Conclusion

• We prove the problem of constructing a data aggregation tree with minimum energy cost is NP-complete and provide a 2-approximation algorithm.• For the problem with relay nodes, we prove it is NP-complete and provide a 7-approximation algorithm.• We show any λ-approximation algorithm of the CND problem can be used to obtain a 2λ-approximation algorithm of our problem.

Page 45: On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National

45

Thank You