1. Nikos Dimokas1 Dimitrios Katsaros2 (presentation)
Detecting Energy-Efficient ‘Central’ Nodes for Cooperative Caching in Wireless Sensor Networks
Nikos Dimokas1
Dimitrios Katsaros2 (presentation)
1Aristotle University of Thessaloniki, Greece
2University of Thessaly, Greece
IEEE AINA, Barcelona (Spain), 25-28/March/2013

2. In-network processing in a camera sensor network
Query
In-network processing in a camera sensor network
Cooperation to get cached data
Response to the sink
How many foxes inside the box on 06:30?

3. Presentation outline Earlier work and background knowledge
Energy paths
Energy Betweenness centrality
Cooperative caching with EBCCoCa
Performance evaluation
Conclusions

4. Cooperative caching: Earlier work
Earlier work & background
Cooperative caching: Earlier work

5. The concept of ‘node betweenness’
Earlier work & background
The concept of ‘node betweenness’
Betweenness Centrality (BC):
The SPBC index of a node is equal to the ratio of the number of shortest paths between pairs of nodes passing by this node to the total number of shortest paths between this pair of nodes 1 9 7 2 6 5 3 4 8
paths between j and k passing through v
betweenness centrality of node v
all paths between j and k

6. The concept of ‘node betweenness’
Earlier work & background
The concept of ‘node betweenness’
Betweenness Centrality (BC):
The SPBC index of a node is equal to the ratio of the number of shortest paths between pairs of nodes passing by this node to the total number of shortest paths between this pair of nodes
1 (0)
9 (0)
7 (13)
2 (0)
6 (7)
5 (0)
3 (10)
4 (8)
8 (0)
paths between j and k passing through v
betweenness centrality of node v
all paths between j and k

7. Tunable balance between Shortest & Energy-efficient paths
Energy paths
Tunable balance between Shortest & Energy-efficient paths
shortest energy path
the multi-hop path that connects two nodes with the minimum energy dissipation among the intermediate nodes
sepuw
the fraction of the summation of edges’ weights (that correspond to the path) to the number of edges comprising the path
Initially, each node represents its remaining energy Er as a percentage of the initial energy Ei. For node u: Eu=Er/Ei
To each edge e = (u,v), we assign a weight wuv = (Eu+Ev)/2

8. Energy path values for edges: Example
Energy paths
Energy path values for edges: Example
Initially, each node has 2Joules energy
Paths from B→E
1.4
1.1
0.65
0.625
0.475
0.55
0.45
0.725
0.675
0.5 D C B H E F G A
B→C→D→E
( )/3= 0.583 1
0.8
B→A→E
( )/2= 0.5
B→H→G→F→E
( )/4=
sepBE=

9. Energy Betweenness Centrality (EBC)
Energy consideration
Latency consideration
when α = 0.0  

10. The datum discovery protocol (1/2)
Cooperative caching with EBCCoCa
The datum discovery protocol (1/2)
A sensor issues a request for an item
Searches its local cache and if it is found (local cache hit) then the K most recent access timestamps are updated
Otherwise (local cache miss), the request is broadcasted and received by the Community Caching Nodes – CCNs (selected based on EBC)
CCNs check the 2-hop neighbors of the requesting node whether they cache the datum (proximity hit)
If none of them responds (proximity cache miss), then the request is directed to the ultimate data source (e.g., a fictitious Data Center)

11. The datum discovery protocol (2/2)
Cooperative caching with EBCCoCa
The datum discovery protocol (2/2)
When a CCN receives a request, searches its cache
If it deduces that the request can be satisfied by a neighboring node (remote cache hit), forwards the request to the neighboring node
If the request can not be satisfied by this CCN, then it forwards it recursively
If none of the nodes (CCNs and ordinary nodes) can help, then requested datum is served by the Data Center (global hit)

12. Cost-based cache replacement
Cooperative caching with EBCCoCa
Cost-based cache replacement
Each sensor first purges the data that it has cached on behalf of some other node
Calculates the following function for each cached datum i
Informs the CCNs about the candidate victim
CCNs determine the new hosting node based on residual energy

13. Evaluation setting (1/2)
Performance evaluation
Evaluation setting (1/2)
EBCCoCa evaluated w.r.t.:
impact of α
competitor: NICoCa (α = 0.0) MONET’08]
Parameter
Default value
Range
#items (N)
1000
fixed
# requests per node
200
Smin (KB) 1
Smax (KB) 10
#nodes (n)
500
bandwidth (Mbps) 2
waiting interval (tw)
10 sec
cache size (KB)
800
200 to 800
Zipfian skewness (θ)
0.8
0.0 to 1.0

14. Evaluation setting (2/2)
Performance evaluation
Evaluation setting (2/2)
Measured quantities
network longevity (first node dies)
average latency for getting the requested data
number of remote hits
Large number of remote hits  effectiveness of cooperation  small latency

15. Impact of network size (θ = 0.8) on longevity in a sparse WSN (d = 4)
Performance evaluation
Impact of network size (θ = 0.8) on longevity in a sparse WSN (d = 4)

16. Impact of network size (θ = 0.8) on longevity in a dense WSN (d = 10)
Performance evaluation
Impact of network size (θ = 0.8) on longevity in a dense WSN (d = 10)
20% up in longevity: In a denser network, more node-disjoint paths are available

17. Impact of α (θ = 0.8) on latency in a sparse WSN (d = 4)
Performance evaluation
Impact of α (θ = 0.8) on latency in a sparse WSN (d = 4) α

18. Impact of α (θ = 0.8) on latency in a dense WSN (d = 10)
Performance evaluation
Impact of α (θ = 0.8) on latency in a dense WSN (d = 10) α

19. Impact of α (θ = 0.8) on cooperation in sparse WSN (d = 4)
Performance evaluation
Impact of α (θ = 0.8) on cooperation in sparse WSN (d = 4) α

20. Impact of α (θ = 0.8) on cooperation in dense WSN (d = 10)
Performance evaluation
Impact of α (θ = 0.8) on cooperation in dense WSN (d = 10) α

21. Summary Performance issues for WSNs
Conclusions
Summary
Performance issues for WSNs
Jointly optimize in a localized manner the nodes’ energy & latency for cooperative caching: EBCCoCa
EBCCoCa:
increases the remote hits (due to the effective sensor cooperation)
identifies energy-efficient paths
tunable for energy vs. latency considerations

22. References (a sample of significant/representative ones)
P. Nuggehalli, V. Srinivasan, C-F. Chiasserini. Energy-efficient caching strategies in ad hoc wireless networks. Proc. ACM MOBIHOC, 2003.
L. Yin and G. Cao. Supporting cooperative caching in ad hoc networks. IEEE Transactions on Mobile Computing, 5(1):77-89, 2006.
N. Dimokas, D. Katsaros, Y. Manolopoulos. Cooperative caching in wireless multimedia sensor networks. ACM Mobile Networks and Applications, 13(3-4), , 2008.
Y. Du, S.K.S. Gupta, G. Varsamopoulos. Improving on-demand data access efficiency in MANETs with cooperative caching. Ad Hoc Networks, 7(3), pp. 579–598, 2008.
M.K. Denko, J. Tian, T.K. R. Nkwe, M.S. Obaidat. Cluster-based cross-layer design for cooperative caching in mobile ad hoc networks. IEEE Systems Journal, 3(4), , 2009.
E. Chan, W. Li, D. Chen. Energy saving strategies for cooperative cache replacement in mobile ad hoc networks. Pervasive and Mobile Computing, 5(1), 77-92, 2009.
N. Chauhan, L.K. Awasthi, N. Chand, R.C. Joshi, M. Misra. Energy efficient cooperative caching in mobile ad hoc networks. International Journal of Applied Engineering Research, 1(3), , 2010.
N. Dimokas, D. Katsaros, Y. Manolopoulos. High performance, low complexity cooperative caching for wireless sensor networks. ACM Wireless Networks, 17(3), , 2011.
N. Chauhan, L.K. Awasthi, N. Chand. Cluster-based efficient caching technique for wireless sensor networks. Proc. IEEE ICLCT, 2012.

23. Thank you!