an adaptive probability broadcast- based data preservation protocol in wireless sensor networks...
TRANSCRIPT
1
An Adaptive Probability Broadcast-based Data Preservation Protocol in Wireless Sensor Networks Liang, Jun-Bin;Wang, Jianxin; Zhang, X.; Chen, Jianer
2011 IEEE International Conference on Communications (ICC)
2
Outline
•Introduction•Related Works•Network Model and Problem Statement•PBDP
▫The Probability broadcast mechanism(PBM)
▫Algorithm of PBDP•Simulations•Conclusions
3
Introduction•Goal
▫Data preservation on harsh WSN without sink.
•Challenge
▫Manage the processes of data dissemination and storage effectively.
•Proposed method
▫PBDP(Probability Broadcast-based Data Preservation)
▫also can reduce the redundancies of data transmission to conserve the energy of nodes.
4
Related Works - Growth codes[4]
•Degree of a codeword “grows” with time•At each timepoint codeword of a specific
degree has the most utility for a decoder (on average)
•This “most useful” degree grows monotonically with time
• R: Number of decoded symbols sink has
R1 R3R2 R4
d=1 d=2 d=3 d=4
Time ->
http://www.powercam.cc/slide/17704
5
Related Works - Growth codes[4]
• Consider the degree of an encoded packet:▫ Decoder has decoded r original data.▫ The probability that new received encoded packet is
immediately decodable to the decoder:
Number of decoded original data: rNumber of decoded original data: r
Impo
rtan
ce o
f Im
med
iate
ly
Dec
odab
le P
acke
t
: Low Degree
: High Degree
http://www.powercam.cc/slide/284
6
Related Works – DFCNS[5]
1. each node should store an information of the path from it to the destination.
2. Cost storage space
3. Assume grid topology
7
Related Works – EDFC[6]
Step 1 : Degree generation▫ Choose degree independently from RSD.
Step 2 : Compute steady-state distribution▫ A random walk corresponds to Markov chain model.
Step 3 : Compute probabilistic forwarding table▫ By the Metropolis algorithm
Step 4 : Compute the number of random walk (b copies)
Step 5 : Block dissemination▫ Each node disseminate b copies of its source block with its
node ID.
Step 6: Encoding
1. Require global information2. cost each node large amount of energy to send and receive large amount of data packet(maintain a large buffer).3. The real node degree may not equal to the chosen degree from RSD.
8
K=1000N=2000
9
Related Work – LTCDS-I[7]
1. Local-cluster effect may happen.
http://www.powercam.cc/slide/16907
10
•Fixing the ratio between n and k as 10%, k/n=0.1
12
1. The transmissions of CF mechanism cost large amount of energy.
2. Each node’s storage reach about 10% of network size.
13
Related Works – rateless packet[*]
13Fig. 3. Example of rateless packet initialization, encoding and dispersion phase.
http://www.powercam.cc/slide/16047
14
15
Node-centric Packet-centric
Growth codes EDFC LTCDS-I DSA-I Rateless
packet
Sink o x x x x
Synchronous o x x x x
DegreeNew
degree RSD RSD RSD RSD
dissemination
- Probabilistic forwarding table
Simple random walk
Simple random walk
Simple random walk
Global information - N>>K, M N>>k N=K N=K
Buffer size - large One data -
# of copies - b (by formula) 1 1
b (by experiment)
Mixing time - >500
Network Model• V = {} randomly distributed in a field of M*M.
• The working time of the network is broken up into time intervals.
16
Tim
e i
nte
rval Sensing
Inference
Storage
Collection
1. use EP[9] technology to estimate the number n of nodes in the network.1. compute parameters according
to .2. disseminate and store data.
1. wake up to sense its vicinity and generate data.
1. into sleep state.2. a collector enter the
network to collect data.
17
Network Model
𝑣1 𝑣8
𝑣3
𝑣7𝑣6
𝑣5𝑣4
Sensor network𝑥8 5 storage units
𝑠3• node generates a data
• put in a packet packet() of c bits for transmission.
• each node has m1 storage units , .
• is the data stored at .
• : the energy of transmission
• : the energy of reception
M
M
18
Network Model
•LNSM[10] (Log-Normal Shadowing Model)
P(d) : the probability of a node receives a packet sent from another node that is located d meters away.
r : communication range
: the path loss exponent,
r =25m=2
19
Problem statement
• How can each node disseminate its data to the network for effective storage at each time interval?
• Goal:make the collector can recover all data even if it just visits a small number of nodes.
20
The Probability broadcast mechanism[11]
Lemma 1. [12]
Assume that each node will rebroadcast a packet after its first reception with probability p and discard it with probability 1−p. In a sufficiently large and sufficiently dense random network, there is a bimodal behavior in the network:
(1) if p ≥, the packet will be received by all nodes, where is a critical probability.(2) if p < , only a small number of nodes can receive the packet
21
The Probability broadcast mechanism
• is decided by▫analyzing a communication graph based on
the network G.
All nodes that receive the packet would form a connected sub-graph .
22
The Probability broadcast mechanism
• In LNSM▫Degree
• A connected network with n nodes, the minimum communication range of nodes is▫ [13], therefore,
▫, when the communication graph contains all nodes in the network.
• Then, is considered to contain all nodes in the network with a probability close to 1[14] when
23
The Probability broadcast mechanism
, (5) A is the event that receives the packet.
(6)
24
The Probability broadcast mechanism
(7)
Since the nodes are dispersed randomly, the degree distribution P(b) can be modeled as a Poisson point process.
(8)
(9)
25
The Probability broadcast mechanism
(10)
26
Performance of PBM
100m*100mr = 25m = 50 nJ/bit = 100 nJ/bitPacket size =100 bits
27
Algorithm of PBDP
28
Simulations
100m*100mr = 25m2 storage units
29
Conclusion
• PBDP can achieve higher decoding performance and energy efficiency than existing schemes.
30
Reference• [4]Abhinav Kamra, Vishal Misra, Jon Feldman, and Dan
Rubenstein, Growth Codes: Maximizing Sensor Network Data Persistence, in Proc. of ACM SIGCOMM, 2006.
• [5]Alexandros G. Dimakis, Vinod Prabhakaran, and Kannan Ramchandran, Decentralized Erasure Codes for Distributed Networked Storage, in: IEEE Transactions on Information Theory, Volume:52, Issue:6, June 2006
• [6]Yunfeng Lin, Ben Liang, and Baochun Li,Data Persistence in Large-scale Sensor Networks with Decentralized Fountain Codes. In Proc. of the 26th IEEE INFOCOM07, Anchorage, Alaska, May 6-12, 2007
• [7]Salah A. Aly, Zhenning Kong, and Emina Soljanin, Fountain Codes Based Distributed Storage Algorithms for Wireless Sensor Networks, Proc. 2008 IEEE/ACM Information Processing of Sensor Networks (IPSN), St. Louis, Missouri, USA, April 22-24, 2008
• [8] Aly, S.A., Youssef, M., Darwish, H.S., Zidan, M., Distributed Flooding- Based Storage Algorithms for Large-Scale Wireless Sensor Networks, IEEE International Conference on Communications (ICC 2009), 2009
31
Reference• [10] L. Quin and T. Kunz, On-demand routing in MANETs: The impact of a
realistic physical layer model, in Proceedings of the International Conference on Ad-Hoc, Mobile, and Wireless Networks, Montreal, Canada, 2003
• [11] Cigdem Sengul, Matthew J. Miller, Indranil Gupta, Adaptive probabilitybased broadcast forwarding in energy-saving sensor networks, ACM Transactions on Sensor Networks, 2008
• [12] Raman, V., Gupta, I., Performance Tradeoffs Among Percolation-Based Broadcast Protocols in Wireless Sensor Networks, 29th IEEE International Conference on Distributed Computing Systems Workshops (ICDCS 2009), 22-26 June 2009
• [13] V. Mhatre, K. Rosenberg, Design Guidelines for Wireless Sensor Networks: Communication, Clustering and Aggregation, Ad Hoc Networks, 2004.
• [14] Jin Zhu, Papavassiliou, S., On the connectivity modeling and the tradeoffs between reliability and energy efficiency in large scale wireless sensor networks, IEEE Wireless Communications and Networking (WCNC 2003), 20-20 March 2003.
• [*]Dejan Vukobratovic´, Cˇ edomir Stefanovic´, Vladimir Crnojevic´, Francesco Chiti, and Romano Fantacci, “Rateless Packet Approach for Data Gathering in Wireless Sensor Networks,” IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 28, NO. 7, EPTEMBER 2010.