Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global...
-
Upload
arthur-todd -
Category
Documents
-
view
215 -
download
0
Transcript of Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global...
![Page 1: Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global Telecommunications Conference 1.](https://reader035.fdocuments.net/reader035/viewer/2022062517/56649f145503460f94c28631/html5/thumbnails/1.jpg)
1
Collaborative Sampling in Wireless Sensor
Networks
Minglei Huang
Yu Hen Hu2010 IEEE Global Telecommunications Conference
![Page 2: Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global Telecommunications Conference 1.](https://reader035.fdocuments.net/reader035/viewer/2022062517/56649f145503460f94c28631/html5/thumbnails/2.jpg)
2
Introduction
Sensor consumes a lot of energy when it communicates with others.
With prior knowledge of correlation between two sensor nodes, the amount of communication can be greatly reduced.1) Select a representative node
2) Divide the field into Voronoi cells and approximate the underlying function
![Page 3: Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global Telecommunications Conference 1.](https://reader035.fdocuments.net/reader035/viewer/2022062517/56649f145503460f94c28631/html5/thumbnails/3.jpg)
3
Problem Formulation(1/3)
Assumptions:1) Broadcast links are symmetric.
2) The broadcast range and energy cost of all sensors are the same.
3) The broadcast time is negligible compared to the backoff time.
4) Fusion Center(FC) has no power constraints.
5) No handshaking before each broadcast.
6) All broadcasts are heard and decoded correctly by the FC.
![Page 4: Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global Telecommunications Conference 1.](https://reader035.fdocuments.net/reader035/viewer/2022062517/56649f145503460f94c28631/html5/thumbnails/4.jpg)
4
Problem Formulation(2/3)
N sensor nodes locate at (i=1,…,N) having instantaneous reading at time k
Using all the reported readings() to estimate the next reading
Error function
g(.) : An estimator that approximate the underlying function
![Page 5: Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global Telecommunications Conference 1.](https://reader035.fdocuments.net/reader035/viewer/2022062517/56649f145503460f94c28631/html5/thumbnails/5.jpg)
5
Problem Formulation(3/3)
Define as a set of sensor nodes that has broadcasted by the report
The optimal node to report at
![Page 6: Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global Telecommunications Conference 1.](https://reader035.fdocuments.net/reader035/viewer/2022062517/56649f145503460f94c28631/html5/thumbnails/6.jpg)
6
Using Distributed Delay(1/2)
Delay due to data aggregation
When to clock out data as it is processed by nodes have significant performance impact in terms of data accuracy and freshness
A back off delay that is inversely proportional to the prediction error at each node
[12] I. Solis and K. Obraczka, “The impact of timing in data aggregation for sensor networks,” in Proceedings of the IEEE ICC, Paris, France, Jun. 2004.
![Page 7: Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global Telecommunications Conference 1.](https://reader035.fdocuments.net/reader035/viewer/2022062517/56649f145503460f94c28631/html5/thumbnails/7.jpg)
7
Using Distributed Delay(2/2)
Packets arrive at FC can be modeled as Poisson Process
=1-
c=
![Page 8: Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global Telecommunications Conference 1.](https://reader035.fdocuments.net/reader035/viewer/2022062517/56649f145503460f94c28631/html5/thumbnails/8.jpg)
8
Algorithm
![Page 9: Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global Telecommunications Conference 1.](https://reader035.fdocuments.net/reader035/viewer/2022062517/56649f145503460f94c28631/html5/thumbnails/9.jpg)
9
Complexity
Goal : Select k N samplesAssume the broadcast range is rN, 0Compare 3 sampling method
1) Oracle – the best possible performance
2) Greedy Oracle
3) Collaborative Sampling
![Page 10: Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global Telecommunications Conference 1.](https://reader035.fdocuments.net/reader035/viewer/2022062517/56649f145503460f94c28631/html5/thumbnails/10.jpg)
10
Experim
ent R
esu
lts
![Page 11: Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global Telecommunications Conference 1.](https://reader035.fdocuments.net/reader035/viewer/2022062517/56649f145503460f94c28631/html5/thumbnails/11.jpg)
Experim
ent R
esu
lts
11
We could save communication and computations by using the stopping criteria to let the algorithm finish sooner.
![Page 12: Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global Telecommunications Conference 1.](https://reader035.fdocuments.net/reader035/viewer/2022062517/56649f145503460f94c28631/html5/thumbnails/12.jpg)
12
Conclusion
The proposed method actively samples the data in network by scheduling the broadcasts in a distributed fashion.
Apply the idea of performing quick approximation of a underlying function in WSN.
The error bound can be tightened and performance of real life data has to be tested.