Multirate Opportunistic Routing Using IETC Metric

1
Multi-rate Opportunistic Routing Using IETC Metric

• January 8, 2009
• Chien-chun Hung

2
Basic idea

• Develop opportunistic routing in wireless
  multi-hop network considering multi-rate
  characteristic and integrated end-to-end
  throughput.

3
Agenda

• why considering opportunistic routing ?
• why considering multi-rate ?
• Multi-rate Geographical Opportunistic Routing.
• why considering end-to-end performance ?
• why considering integrated performance ?
• Our approach
• ITC metric
• Future work

4
WHY CONSIDERING OPPORTUNISTIC ROUTING?
5
Why considering opportunistic routing?

• Opportunistic routing exploits the wireless
  broadcast nature and takes the advantages of
  spatial diversity, selecting multiple neighboring
  nodes to help relay data packets during one
  transmission.

6
Why considering opportunistic routing?
7
Why considering opportunistic routing?

• As more than one neighbor involved in forwarding
  process, the possibility of at least one
  forwarder receives the packets correctly
  increases, reducing the number of retransmission
  and further improving transmission throughput.
• ? therefore

8
Why considering opportunistic routing?

• Transmit data packets involving multiple
  forwarders instead of single next hop.
  conclusion A
• Conclusion A multiple forwarders instead of
  single path.

9
WHY CONSIDERING MULTI-RATE ?
10
Why considering multi-rate?

• Some cases that base rate must be applied.
• Some cases that higher rate must be applied.
• Some cases that both higher and base rate must be
  applied.

11
Base rateusing base rate can reach into the
transmission range far away
12
Higher ratehigher rate can achieve high
throughput within short range
13
However
14
Both higher/base rates are needed
S
11 Mbps P 0.75
F
1 Mbps P 0.5
? Therefore
D
15
Why considering multi-rate?

• Considering non-base rate instead of base rate
  only exploiting the benefit of multi-rate
  instead of single-rate. conclusion B
• Conclusion A multiple forwarders instead of
  single path.
• Conclusion B multi-rate instead of single-rate.

16
Multiple forwarders vs. single next
hop Multi-rate vs. single rate
17
MULTI-RATE GEOGRAPHICAL OPPORTUNISTIC ROUTING
18
MGOR

• Multi-rate Geographical Opportunistic Routing
• Compared with conventional GOR, except using
  multi-rate advantage, MGOR takes both
  geographical information and link reliability
  into consideration during candidate selection and
  forwarding prioritization.

• ON THROUGHPUT EFFCIENCY OF GEOGRAPHIC
  OPPORTUNISTIC ROUITNG IN MULTIHOP NETWORKS
  Mobile Networks and Application 2007
• MULTI-RATE GEOGRAPHIC OPPORTUNISTIC ROUTING IN
  WIRELESS AD HOC NETWORKS MILCOM 2007
• ON END-TO-END THROUGHPUT OF OPPORTUNISTIC
  ROUTING IN MULTIRATE AND MULTIHOP WIRELESS
  NETWORKS INFOCOM 2007

19
Impact of transmission rate
20
Impact of forwarding strategy

• F1 gt F2 gt F3 gt F4 1.86 Gbmps
• F3 gt F2 gt F4 gt F1 1.99 Gbmps
• Lower-priority forwarders have to wait for a long
  period until higher-priority forwarders confirm
  their reception and transmission. However, since
  farther nodes tend to have limited reliability,
  the long period is wasted. This situation is
  getting worse when the coordination delay
  increases.

21
Impact of candidate selection

• Candidate set ltF1,F4gt 1.46 Gbmps
• Candidate set ltF3,F2gt 1.67 Gbmps
• Candidate set ltF1,F2,F3,F4gt 1.86 Gbmps
• Candidate set ltF3,F2,F4,F1gt 1.99 Gbmps
• Candidate set ltF3,F2,F4gt largest throughput

• Coordination delay ? , of forwarders ?
• Coordination delay ? , of forwarders ?

22
MGOR

• EOT To characterize the behavior of GOR in terms
  of bit-meter advancement per second (bmps) during
  candidate selection and relay prioritization
  within one-hop neighbors.
• Packet advancement
• Link reliability

23
Challenges for MGOR

• MGOR takes both distance advancement and link
  reliability into consideration, making the best
  decision among one-hop neighbors.
• However, the routing information out of one-hop
  neighbors plays a more critical role during
  candidate selection and prioritization.
• End-to-end performance vs. one-hop performance

• ETX A high throughput path metric for multi-hop
  wireless routing, MobiCom03
• ExOR Opportunistic multi-hop routing for
  wireless networks, SIGCOM05

24
WHY CONSIDERING END-TO-END PERFORMANCE?
25
Why considering e2e performance?

• Consider only the advancement within one hop,
  failed to take the link reliability after one hop
  into account.
• F1 and F2 achieve the same packet advancement.
• MGOR is likely to choose only F1 as forwarder
  however, due to the poor link quality between F1
  and D, as well as F2 and D, F1 and F2 should both
  be chosen.

26
Why considering e2e performance?

• Hole Problem
• MGOR is likely to choose a forwarder which has
  no available forwarders after one hop
  transmission. This is due to the limitation that
  MGOR failed to consider the whole condition from
  the forwarder to the destination.

27
Why considering e2e performance?
400 m, p 0.5
320 m, p 0.6
450 m
28
Why considering e2e performance?
P 0.6 ETX 1.67
P 0 ETX 8
P 0.7 ETX 1.43
P 0.8 ETX 1.25
P 0.5 ETX 2
29
Why considering e2e performance?

• Consider the whole end-to-end performance during
  candidate selection and relay prioritization
  instead of performance within one hop.
  conclusion C
• Conclusion A multiple forwarders instead of
  single path
• Conclusion B multi-rate instead of single-rate
• Conclusion C e2e performance instead of one-hop
  performance

30
WHY CONSIDERING INTEGRATED END-TO-END PERFORMANCE
?
31
Why considering integrated end-to-end performance?
32
Why considering integrated end-to-end performance?
P 0.8 ETX 1.25
F11
F1
P 1 ETX 1
P 0.75 ETX 1.33
S
D
P 0.7 ETX 1.43
P 1 ETX 1
P 0.95 ETX 1.05
F21
F2
33
Why considering integrated end-to-end performance?

• The drawbacks of ETX approach
• Failed to consider the integrated performance
  after one-hop neighbors.
• Failed to consider the link quality within
  one-hop neighbors.

34
Why considering integrated end-to-end performance?

• Since opportunistic routing takes the advantage
  of multiple neighbors to increase reliability and
  further improves transmission throughput, we
  think that integrated e2e performance, revealing
  the potential ability provided by multiple
  forwarders, should be considered during candidate
  selection and prioritization.

35
Why considering integrated end-to-end performance?

• Considering integrated end-to-end performance
  instead of single path performance
• conclusion D
• Conclusion A multiple forwarders instead of
  single path
• Conclusion B multi-rate instead of single-rate
• Conclusion C e2e performance instead of one-hop
  performance
• Conclusion D integrated e2e performance instead
  of single path performance

36
OUR APPROACH
37
Our approach

• We aim at developing a metric representing the
  Integrated End-to-end Transmission Cost (IETC)
  from any node to destination.
• Each transmitter can therefore selects forwarders
  and prioritize them based on IETC metric.

38
Algorism

• Each node periodically broadcasts probing packets
  within one-hop neighbors using different rates,
  measuring the packet receiving ratio (PRR) as the
  link reliability.
• A backward recursive computation for calculating
  one nodes IETC value for a specific destination.
  The IETC value of destination is 0.
• After finishing calculation at one node, the
  minimal IETC value will be broadcasted,
  representing the IETC value to the destination of
  that node.

S
D
39
IETC metric
Rate Ri
IETC1
IETCX
IETC2
IETCY
IETCS
IETCD 0
IETC3
IETCZ
We assume that prioritization is set as A gt B gt C
where
represents the probability of at least one
forwarder receives the packets correctively.
40
Simulation

• (under construction.)

41
Conclusion

• An opportunistic routing protocol exploiting
  multi-rate characteristic.
• Propose an integrated end-to-end performance
  metric as a basis for candidate selection and
  prioritization.

42
Future work

• Simulation work
• Exploiting spatial reuse to improve transmission
  throughput.
• Developing a scalable routing protocol supporting
  multiple connections.