Cluster Based Routing Protocol original slides, School of Computing, National University of Singapore, 1999.

1. CBRP: A Cluster-based Routing
Protocol for Mobile Ad hoc Networks
Presented by: Jiang Mingliang
Supervised by: Dr Y.C. Tay, Dr Philip Long

2. Presentation Outline
Project Overview and Objectives
Related Works
CBRP: Motivations
CBRP: the Details
Performance Evaluation
Conclusion and Future Work

3. Project Overview
Mobile Ad hoc Networks (MANET), its
applications and challenges
IETF working group MANET

4. Project Overview
MANET characteristics ( & the difficulties
for routing protocols)
Dynamic Topology
Limited Link Bandwidth
Limited Power Supply for Mobile Node
Need to scale to large networks

5. Project Objective
Design a routing protocol for MANET that
is:
efficient
scalable
distributed and simple to implement
Evaluate CBRP through simulation
compare with different design alternatives
compare against other MANET protocols

6. Related Works
Existing MANET protocols:
MANET
routing
protocols
discover routes
on-demand (re-active)
Maintain updated
routes (pro-active)
Source routing
Table driven
Variation of
distant vector?
Variations of
link state routing?
DSR
AODV, ABR,
TORA
DSDV
OLSR

7. Related Works
Problems with pro-active routing protocols
high overhead in
periodic/triggered routing table updates
low convergence rate
waste in maintaining routes that are not going
to be used!!
Simulating results have shown RIP, OSPF,
DSDV fails to converge in highly dynamic
MANET.

8. Related Works
Re-active Routing Protocols
prohibitive flooding traffic in route discovery
route acquisition delay
every route breakage causes a new route
discovery
Works in trying to reduce flooding traffic
LAR (GPS for every mobile node?)
DSR (aggressive caching)

9. CBRP: Motivations
Design Objective:
a distributed, efficient, scalable protocol
Major design decisions:
use clustering approach to minimize on-
demand route discovery traffic
use “local repair” to reduce route acquisition
delay and new route discovery traffic
suggest a solution to use uni-directional links

10. CBRP: Protocol Overview

11. Cluster Formation
Mechanism:
Variations of “min-id” cluster formation algorithm.
Nodes periodically exchange HELLO pkts to
maintain a neighbor table
neighbor status (C_HEAD, C_MEMBER, C_UNDECIDED)
link status (uni-directional link, bi-directional link)
maintain a 2-hop-topology link state table
Objective:
Form small, stable clusters with only local information
Node ID Node Status
Neighbor ID Neighbor status Link status
… … …
Adjacent cluster ID
…
HELLO
message
format:

12. Cluster Formation (an example)
Variation of Min-ID
Minimal change
Define Undecided State
Aggressive Undecided -> Clusterhead
e.g. 2’s neighbor table
3
8
4
1
5
2
6
7
9
10
11
Nbr ID Nbr status Link status
7 member Bi-directional
6 C_head Bi-directional
4 member Bi-directional
1 C_head Bi-directional

13. Adjacent Cluster Discovery
3
8
4
1
5
2
6
7
9
10
11
Objective:
For clusterheads 3 hops away to discover each other
Mechanism:
Cluster Adjacency Table exchanged
in HELLO message
e.g. 4’s Cluster Adjacency Table
Adj cluster ID Gateway
8 9
6 2

14. Route Discovery
Source S “floods” all clusterheads with Route Request Packets
(RREQ) to discover destination D
[3]
[3,1,8,11]
1
2
4
5 6
7
8
9
10
3
11
3 (S)
11 (D)
[3,1]
[3,1,6]
[3,1,8]

15. Route Reply
Route reply packet (RREP) is sent back to source along
reversed “loose source route” of clusterheads.
Each clusterhead along the way incrementally compute a
hop-by-hop strict source route.
1
2
4
5 6
7
8
9
10
3
11
3 (S)
11 (D)
the reversed
loose source route of
RREP: [11,8,1,3]
[11][11,9]
[11,9,4]
[11,9,4,3]
the computed
strict source route of
3->11 is: [11,9,4,3]
[11,9,4]

16. Route Reply
Route reply packet (RREP) is sent back to source along
reversed “loose source route” of clusterheads.
Each clusterhead along the way incrementally compute a
hop-by-hop strict source route.
1
2
4
5 6
7
8
9
10
3
11
3 (S)
11 (D)
the reversed
loose source route of
RREP: [11,8,1,3]
the computed
strict source route of
3->11 is: [11,9,4,3]

17. Route Error Detection
1
2
4
5 6
7
8
9
10
3
11
3 (S)
11 (D)
Use source routing for actual packet forwarding
A forwarding node sends a Route Error Message (ERR) to
packet source if the next hop in source route is unreachable
Source route header of data
packet: [3,4,9,11]
Route error (ERR)
down link: {9->11}

18. Local Route Repair in CBRP
Objective
Increase Packet Delivery Ratio
Save Route Rediscovery flooding traffic
Reduce overall route acquisition delay
Mechanism
Spatial Locality

19. Local Route Repair
1
2
4
5 6
7
8
9
10
3
11
3 (S)
11 (D)
A forwarding node repairs a broken route using its 2-hop-topology
information and modifies source route header accordingly.
Destination node sends a gratuitous route reply to inform source
of the modified route
Source route header of data
packet: [3,4,9,11]
Route error (ERR)
down link: {9->11}

20. Local Route Repair
1
2
4
5 6
7
8
9
10
3
11
3 (S)
11 (D)
A forwarding node repairs a broken route using its 2-hop-topology
information and modifies source route header accordingly.
Destination node sends a gratuitous route reply to inform source
of the modified route
Source route header of data
packet: [3,4,9,11]
Modified source route
[3,4,9,8,11]

21. Local Route Repair
1
2
4
5 6
7
8
9
10
3
11
3 (S)
11 (D)
A forwarding node repairs a broken route using its 2-hop-topology
information and modifies source route header accordingly.
Destination node sends a gratuitous route reply to inform source
of the modified route
Source route header of data
packet: [3,4,9,11]
Gratuitous route reply
[3,4,9,8,11]

22. Utilize Unidirectional links
Cause of unidirectional links
Hidden Terminal
Difference in transmitter power or receiver
sensitivity.
Pitfalls with unilinks
Discovery of (dead) unilinks
Problems with 802.11 RTS/CTS/Snd/Ack,
ARP

23. Utilize Unidirectional links
Selective use of Unilinks in CBRP
1 2
7
8 3
5
4
10
6
9

24. Supercluster
Taking advantage of hidden stability from
the changing topology
Better support for natural mobility patterns
Merge stable clusters into supercluster
to be further studied

25. Performance Evaluation
Goals
show the robustness of CBRP’s packet delivery with
reduced overhead.
evaluate how CBRP scales to larger networks
compare different design alternatives (with/without local
repair)
compare CBRP with other MANET routing protocols
Tools
ns (network simulator) with wireless extension.
features
models Lucent WaveLAN DSSS radio with signal
attenuation, collision and capture.
implements IEEE 802.11 link layer

26. Simulation Environment
Mobility Model (random way-point)
Nodes move within a fixed rectangular area m x n
Each node chooses a random destination and move
toward it at a speed uniformly distributed between 0 and
max_speed
When reaching its destination, a node pauses for
pause_time before start moving again.
Traffic Model
A node creates a session with a randomly selected
destination node.
Packets of fixed size 128 byte are sent with constant
sending rate of 4 pkts/sec

27. Simulation Parameters
Simulator parameters
CBRP implementation parameters
channel bandwidth 2Mbps transmission range 250m
max_speed 20m/s simulated time 600s
Route Request Retransmit Interval
(exponential backoff)
500ms
Timeout for packets without a route 30s
Network interface buffer size 50
Send buffer size at the packet originator 50

28. 1. Packet delivery ratio
with respect to network mobility
Network mobility is directly affected by pause_time.
pause_time has value {0, 30s, 60s, 120s, 300s, 600s} with 0
representing constant mobility and 600s signifying a stationary network.
Packet Delivery Ratio for 50-node network
(30 CBR sources, 128-byte packets)
0.7
0.75
0.8
0.85
0.9
0.95
1
0 150 300 450 600
pause time
packetdeliveryratio
CBRP
CBRP-w/o repair
DSR
DSDV

29. 2. Packet delivery ratio with respect to
network size
Simulated network of nodes {25, 50, 75, 100, 150} with constant
mobility, 60% of nodes have active CBR sessions.
Packet Delivery Ratio
with increasing number of nodes
0.5
0.6
0.7
0.8
0.9
1
25 50 75 100 125 150
number of nodes
packetdeliveryratio
CBRP
CBRP-w/o repair
DSR

30. 2. Routing Overhead with respect to
network size
Routing overhead(normalized) = #routing pkts sent/ #data pkts delivered.
Routing Overhead
with increasing number of nodes
0
2
4
6
8
25 50 75 100 125 150
number of nodes
routingoverhead
CBRP
CBRP-w/o repair
DSR

31. Milestones
Aug 98, CBRP as Internet Draft
Aug 98, in Chicago Presentation to the
IETF
Oct 98, presentation to MMlab, EE, NUS
Nov 98, Presentation to IETF in Orlando
Mar 99, paper submitted to Globecom99

32. Limitations of CBRP
Source Routing, overhead bytes per
packet
Clusters small, 2 levels of hierarchy,
scalable to an extend

33. Conclusion
CBRP is a robust/scalable routing
protocol superior to the existing proposals
Further study on Superclustering
QoS, Multicast support in CBRP