Upload
parc-a-xerox-company
View
558
Download
0
Embed Size (px)
DESCRIPTION
A Content-Centric Approach for Requesting and Disseminating Monitoring Information in Wireless Mesh Networks Julien Boite, Vania Conan, Gérard Nguengang, Mathieu Bouet (Thales Communications & Security, France), Alain Ploix, Dominique Gaïti (University of Technology of Troyes, France)
Citation preview
A Content-Centric Approach
for Requesting and Disseminating Monitoring Information !!in WMN
Julien Boite – CCNxCon 2012 – 12/09/2012
2 / 2 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Outline
Context
! Wireless Mesh Networks ! Adaptive Flow-based Gateway Selection Objective
Measurement Approaches
! End-to-end Measurements Scalability Issue ! Proposed Hop-by-Hop Measurements Approach
A Use Case For Content-Centric Networking
! Transposition to the CCN world
Conclusion and Future Work
3 / 3 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Context
Wireless Mesh Networks
! Nowadays deployed as access networks " Rural dead zones, urban zones, campus-wide
" Commercial deployments use WiFi / WiMax [1][2]
! Provide users with a wide range of communication services " Heterogeneous quality constraints
! Gateways and backhaul links play a key role " Most of traffic goes to the Internet [3]
" Multiple GWs spread over the network
! Different backhaul technologies " ADSL, Ethernet, wireless, HF, satellite
" Heterogeneous and dynamic performances !"#$%&'%
!"#!"$
!"%!"& !"#$%$&&
4 / 4 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Context
Adaptive Flow-based Gateway Selection Objective
! How to ensure that QoS-constrained flows always benefit from the best quality available?
" Fluctuating performances inside the Wireless Mesh Network
" Time-varying capabilities of gateways and backhaul networks
" Diversity of applications and flows’ constraints
! Limitations of ad hoc routing protocols [4] " Specific metrics have been proposed (ETX, ETT, etc.) [5]
# Allow taking into account links quality, load, performances inside the WMN
" But still a unique metric to choose a default gateway on each mesh router
! Limitations of load balancing schemes [6-9]
" Focus on avoiding congestion or resource usage optimization
" Does not consider the QoS requirements of flows
" Dos not consider heterogeneous and dynamic performance of backhaul links
!"#$%&'%!"#
!"$
!"%!"&
!"#$%$&&
$ Right choices for QoS ??
5 / 5 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Context
Adaptive Flow-based Gateway Selection Objective
! Objective " Dynamically map the requirements of QoS-constrained flows onto the gateway and backhaul network that best deals with it
" Configure forwarding accordingly, at the flow level of granularity
! Software entities embedded on mesh routers " Form an overlay, communicate one with another
" Interact with local node (gather info, enforce config) # Trigger E2E measurements through each GW
# Apply forwarding decisions
" Run algorithms for intelligent reactive decision making # Performance comparison with regard to flows’ constrains
" Existing framework for implementation [10] Network Element
Monitor
Decide
Execute
6 / 6 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Measurement Approaches
! How to efficiently monitor end-to-end performances to feed the decision making process?
Network Element
Monitor
Decide
Execute
7 / 7 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Measurement Approaches
End-to-end Measurements Scalability Issue
! Monitoring performances offered by gateways and their backhaul network " Passive probing $ considered, but not sufficient
" Active probing generates overhead $ scalability of end-to-end measurements?
" Model for computing the overhead (nb packets) in grid topologies # Dng = nb hops from mesh router n to GW g
# Mp = nb probe packets ; N = nb nodes ; G = nb GWs
!"#$%&'%!"#
!"$
!"%!"&
!"#$%$&&
'
(
0
5000
10000
15000
20000
25000
0 10 20 30 40 50 60 70 80 90 100
Ove
rhe
ad
(n
b p
ack
ets
)
Nb nodes
Overhead inside the WMN
Overhead outside the WMN
Total overhead
8 / 8 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Measurement Approaches
Proposed Hop-by-Hop Measurements Approach
! To reduce the overhead generated to monitor performances offered by gateways and their backhaul network
" Hop-by-hop probing mechanism + Control protocol and memory on nodes
Probe Control message Memory
Wireless link Backhaul link
R1
9 / 9 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Measurement Approaches
Proposed Hop-by-Hop Measurements Approach
! To reduce the overhead generated to monitor performances offered by gateways and their backhaul network
" Probing mechanism + Control protocol and memory
Probe Control message Memory
Wireless link Backhaul link
R1
10 / 10 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Measurement Approaches
Proposed Hop-by-Hop Measurements Approach
! To reduce the overhead generated to monitor performances offered by gateways and their backhaul network
" Probing mechanism + Control protocol and memory
Probe Control message Memory
Wireless link Backhaul link
R1
11 / 11 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Measurement Approaches
Proposed Hop-by-Hop Measurements Approach
! To reduce the overhead generated to monitor performances offered by gateways and their backhaul network
" Probing mechanism + Control protocol and memory
Probe Control message Memory
Wireless link Backhaul link
R1
12 / 12 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Measurement Approaches
Proposed Hop-by-Hop Measurements Approach
! To reduce the overhead generated to monitor performances offered by gateways and their backhaul network
" Probing mechanism + Control protocol and memory
Probe Control message Memory
Wireless link Backhaul link
R1
R2
13 / 13 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Measurement Approaches
Proposed Hop-by-Hop Measurements Approach
! To reduce the overhead generated to monitor performances offered by gateways and their backhaul network
" Probing mechanism + Control protocol and memory
Probe Control message Memory
Wireless link Backhaul link
R2 R1
14 / 14 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Measurement Approaches
Proposed Hop-by-Hop Measurements Approach
! To reduce the overhead generated to monitor performances offered by gateways and their backhaul network
" Probing mechanism + Control protocol and memory
Probe Control message Memory
Wireless link Backhaul link
R2 R1
15 / 15 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Measurement Approaches
Proposed Hop-by-Hop Measurements Approach
! To reduce the overhead generated to monitor performances offered by gateways and their backhaul network
" Probing mechanism + Control protocol and memory
Probe Control message Memory
Wireless link Backhaul link
R2 R1
16 / 16 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Measurement Approaches
Proposed Hop-by-Hop Measurements Approach
! Results compared to E2E measurements " Potential for reducing overhead inside and outside the WMN
" Model for computing the overhead +
!"#$%&'%!"#
!"$
!"%!"&
!"#$%$&&
'
(
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
0 10 20 30 40 50 60 70 80 90 100
Ove
rhe
ad
(n
b p
ack
ets
)
Nb nodes
E2E strategy inside WMN
E2E strategy outside WMN
E2E strategy total overhead
Hop-by-Hop strategy inside WMN
Hop-by-Hop strategy outside WMN
Hop-by-Hop strategy total overhead
Ms = nb signaling packets
Pj,n = probability that node at distance j from requesting node n provides the result without generating more probes
17 / 17 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
A Use Case for Content-Centric Networking
Transposition to the CCN world
! Monitoring results $ content " Naming (for hierarchical prefix-based routing):
! Gateways $ content repository " Content produced if not available
! Requests / Responses $ Interest packets / Data packets " Interest packets sent to the IP
next hops towards gateways
! Memory $ content store " Dealing with measurements validity: FreshnessSeconds1
! What about measurements and results aggregation ?
/domain.net/measurement/GW_ID/requestor/target
1http://www.ccnx.org/releases/latest/doc/technical/Staleness.html
18 / 18 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
A Use Case for Content-Centric Networking
Transposition to the CCN world
! Face mapped to an application " Forward unserved Interest packets to this face
$ Entry in the PIT for content name CN1 with incoming Face
" Trigger a measurement up to the IP next hop towards a gateway
" Send a new Interest packet to the next hop towards the gateway (specific Face)
$ New entry in the PIT with the application’s Face, content name must different than CN1
" This PIT entry is consumed when receiving the Data packet, forwarded to the application
" Aggregate measurement results
" Generate a new Data packet with the originally requested content name CN1 $ The first entry added in the PIT is consumed and the Data packets is forwarded back towards the requestor
! Manipulating FIB entries is necessary to achieve this routing behaviour
19 / 19 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Conclusion and Future Work
Conclusion
! Hop-by-Hop Measurement Approach " Great potential for reducing overhead both inside and outside a WMN
" The more mesh routers request for identical measurements and the more frequent these requests, the lowest the overhead
" Probing mechanisms + control protocol for requesting and dissimenating measurment results
! Could be implemented using CCN mechanisms " Interception of messages by an application to perform measurements and aggregate results
Future Work
! CCN implementation ! Study the overhead in more realistic scenarios
" Average distance to obtain a result ?
" Impact of timers duration ?
20 / 20 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Questions ?
Thank you!
Contact: [email protected]
Paper
Q2SWinet 2012 – “Scaling End-to-end Measurements in Heterogeneous Wireless Mesh Networks”
Acknowledgments
Vania Conan Thales Communications & Security Gérard Nguengang Thales Communications & Security Mathieu Bouet Thales Communications & Security Alain Ploix Université de Technologie de Troyes Dominique Gaïti Université de Technologie de Troyes
21 / 21 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
References
[1] Meraki Inc. (2011). [Online]. Available: http://meraki.com [2] Ozone. (2010). [Online]. Available: http://www.ozone.net [3] J. Jun and M.L. Sichitiu, “The nominal capacity of Wireless Mesh Networks”, Wireless Communications, IEEE, vol. 10, no. 5, pp. 8-14, 2003 [4] M. Abolhasan, T. Wysocki and E. Dutkiewicz, “A review of routing protocols for mobile ad hoc networks”, Ad Hoc Networks, vol. 2, no. 1, pp.1-22, 2004 [5] M. Campista, P. Esposito, I. Moraes, L. Costa, O. Duarte, D. Passos, C. de Albuquerque, D. Saade and M. Rubinstein, “Routing metrics and protocols for wireless mesh networks”, Network, IEEE, vol. 22, no. 1, pp. 6–12, 2008 [6] D. Nandiraju, L. Santhanam, N. Nandiraju and D.P. Agrawal, “Achieving load balancing in wireless mesh networks through multiple gateways”, in Proc. of 2006 IEEE MASS, 2006, pp. 807-812 [7] Y. Bejerano, S.-J. Han and A. Kumar, “Efficient load-balancing routing for wireless mesh networks”, Computer Networks, vol. 51, no. 10, pp. 2450–2466, 2007 [8] J. Baliosian, J. Visca, E. Grampín, L. Vidal and M. Giachino, “A rule-based distributed system for self-optimization of constrained devices”, in Proc. of 2009 IFIP/IEEE Int. Symposium on Integrated Network Management (IM), 2009, pp. 41–48 [9] E. Ancillotti, R. Bruno and M. Conti, “Load-balanced routing and gateway selection in wireless mesh networks: Design, implementation and experimentation”, in Proc. of 2010 IEEE Int. Symposium on WoWMoM, 2010, pp. 1–7 [10] J. Boite, G. Nguengang, M. Israël, V. Conan, “CONEMAF: A modular multi agent framework for autonomic network management”, in Proc. of the Int. Conference on Agents and Artificial Intelligence (ICAART), 2010, vol. 2, pp. 224–231 [11] Networking Named Content, 2009, V. Jacobson, D.K. Smetters, J.D. Thornton, M.F. Plass, N.H. Briggs and R.L. Braynard, in Proceedings of the 5th international conference on Emerging networking experiments and technologies, pp. 1-12 [12] Project CCNx™. http://www.ccnx.org, Sep. 2009
22 / 22 /
DS
C/D
T/C
EA
/TA
I – 9
/28/
12
Measurement Approaches
Proposed Hop-by-Hop Measurements Approach
! Monitoring performances offered by gateways and their backhaul network " Hop-by-hop strategy properties
# The more mesh routers request for identical measurements, the lowest overhead
# If all nodes request for the same measurement (metric/dest) before timers expire, induced overhead is the same for all topologies
# If not all nodes request for the same measurement, induced overhead is topology-dependent
0
20
40
60
80
100
120
140
160
0 2 4 6 8 10 12 14 16 18 20
Ove
rhe
ad
(n
b p
ack
ets
)
Nb nodes performing a measurement (k)
n=20
n=18
n=1Linear topology for each n in [1;20]
Dense topology for each n in [1;20]
GW GW...
...
123N
1
2 3 N
Linear topology
Star topology
GW GW...
...
123N
1
2 3 N
Linear topology
Star topology