Comparison of Routing Metrics for Static Multi

ExOR: Opportunistic Multi-Hop Routing
For Wireless Networks
Sanjit Biswas & Robert Morris
Contributions
 This paper contributes the first complete design
and implementation of a link/network-layer
diversity routing technique that uses standard radio
hardware.
 It demonstrates a substantial throughput
improvement over traditional routing.
Why ExOR promises high throughput? (1)
25%
25%
S
25%
I1
I2
I3
100%
100%
D
100%
100%
25%
I4
 Reception at different node is independent, no interference
 Traditional Routing: 1/ 0.25 + 1 = 5tx
4
 ExOR: 1/ (1-(1-0.25) ) + 1 = 2.5tx
Why ExOR promises high throughput? (2)
S
N1
N2
N3
N4
N5
N6
N7
N8
D
Traditional Path




Gradual falloff of probability with distance (80%, 40%, 20%..)
Lucky longer path can reduce transmission count
Shorter path ensures some forward progress
ExOR works better with local interference than global
Four Design Challenges
 The nodes must agree on which subset of
them received each packet
 A metric to measure the probable cost of
moving packet from any node to
destination
 Choosing most useful participants
 Avoid simultaneous transmission to
minimize collisions
Node State
 Packet buffer
 Forwarding timer
 Transmission tracker
 Used to adjust forwarding tracker.
 Batch map
 Highest priority node that received the
pkt
Working…
N7
F
N8
F
F
N1
N2
N5
S
F
Batch
N4
D
N3
N6
F
1st round
2nd round
3rd round
Cost Metric: similar to ETX
 ETX: Expected Transmission Count
 ExOR uses forward delivery
probability
 Knowledge of inter-node loss rate
Evaluation Setup
 38 Roofnet nodes participated using
802.11b
 One hop at a time for fair comparison
in traditional routing.
Evaluation – 1A
 Bars higher than 1000 indicates that nodes had to transmit the
packet more than once.
Evaluation – 1B
 ExOR exhibits better throughput
Evaluation - 2
Summary
 Pros
 ExOR achieves 2x to 4x throughput
improvement for more distant pairs
 ExOR implemented on Roofnet and evaluated in
detail
 Exploits radio properties, instead of hiding them
 Does not require changes in the MAC layer
 Cons
 Not scalable to large network as traditional
routing
 Overhead in packet header (batch info)
 Batches affect the TCP performance
 What if not enough packets to make the batch?