You are on page 1of 27

Orthogonal Rendezvous

Routing Protocol for


Wireless Mesh Networks
Bow-Nan Cheng
Murat Yuksel
Shivkumar Kalyanaraman
By removing
position
information, can
we still efficiently
route packets?
Motivation
L3: Geographic Routing using Node IDs
(eg. GPSR, TBF etc.)
L2: ID to Location Mapping
(eg. DHT, GLS etc.)
L1: Node Localization
ORRP
N/A
Issues in Position-based Schemes
S
N
W E
(0,4)
(4,6)
(5,1)
(8,5)
(12,3)
(15,5) S
D
D(X,Y)?
?
Motivation Multi-directional
Transmission Methods
Multi-directional Antennas Tessellated FSO Transceivers
Directional communications
Model needed for ORRP
45
o
22.5
o
ORRP Introduction
Up to 69%
A
B
98%
Assumptions
Neighbors are
assigned a direction
Local Sense of
Direction
Ability to
Transmit/Receive
Directionally
Directional, smart
antennas
FSO transceivers

ORRP Design Considerations
Considerations:
High probability of connectivity without position
information [Reachability]
Scalability O(N
3/2
) total state information maintained.
(O(N
1/2
) per node state information)
Even distribution of state information leading to no
single point of failure [State Complexity]
Handles voids and sparse networks
Trade-offs
Path Stretch
Probabilistic Reachability
ORRP Proactive and Reactive Elements
Node C Fwd Table
Dest Next Cost Dir
A B 2 120
o
D D 1 230
o
Node B Fwd Table
Dest Next Cost Dir
A A 1 90
o
A
B
C
D
1. ORRP Announcements (Proactive)
Generates Rendezvous node-to-destination paths
1
1
1
1
2. ORRP Route REQuest (RREQ) Packets (Reactive)
2
2
2
2
2
3 3
3. ORRP Route REPly (RREP) Packets (Reactive)
4. Data path after route generation
4
4
Deviation Correction:
Multiplier Angle Method (MAM) Concept
180
o
t=45
o
45
o
=1t
45
o
=1t
180
o
-90
o
=-2t
A
B
C
D
E
F
G
t
o |
u t o
t
t
u
t
t
u

=
+ =

|
.
|

\
|
=

|
.
|

\
|
+ =
m
negative m m
positive m m
, 2 ,
2
max
, 2 ,
2
min
Multiplier (m)
Desired Angle
Received Angle
Loop Prevention
Actual Tx Angle
Interface Separation Angle
Deviation Angle
New Multiplier (m)
Void
u = min(+4t, +6t)
o = + t 4t
m = +2
S
R
u = min(+4t, 0)
o = + t
m = +3
u = min(+4t, 0)
o = + t
m = 0
u = min(+4t, +4t)
o = + t 4t
m = +2
u = min(+4t, +4t)
o = + t 4t
m = +3
Multiplier Angle Method (MAM)
Examples
Basic Example
VOID Navigation/Sparse
Networks Example
u = min(+4t, +6t)
o = + t 4t
m = +2
ORRP Void Navigation differences from
GPSR perimeter routing
ORRP seeks only intersections between destination ORRP
packets and source ORRP packets increased flexibility
MAM is an inherent nature of ORRP and not a special case that
switches on and off like GPSR perimeter routing
ORRP does not require location-id mappings as GPSR does
Performance Evaluation of ORRP
Metric
Reachability Percentage of nodes reachable by each node
in network (Hypothesis: high reachability)
State Complexity The total state information needed to be
maintained in the network (Hypothesis: O(N
3/2
))
Path Stretch Average ORP Path vs. Shortest Path
(Hypothesis: Low path stretch)
Analysis (without MAM)
Reachability Upper Bound
State Information Maintained at Each Node
Average Path Stretch
Packetized Simulation Scenarios Evaluated
Effect of MAM on reachability
Effect of finer-grained directionality
Total state complexity and distribution of state
Reachability Numerical Analysis
P{unreachable} =
P{intersections not in
rectangle}
4 Possible Intersection Points
1
2
3
98.3% 99.75%
57%
67.7%
Probability of
Unreach highest
at perimeters and
corners
NS2 Simulations
with MAM show
around 99%
reachability
ORRP Perimeter Issue
Perimeter/Corner Nodes Corner nodes have
higher probability of orthogonal line intersections
outside of topology bounding region
Path Stretch Analysis
Average Stretch for
various topologies

Square Topology 1.255
Circular Topology 1.15
25 X 4 Rectangular 3.24
Expected Stretch 1.125

x = 1.255 x = 1.15
x = 3.24
State Complexity Analysis/Simulations
GPSR DSDV XYLS ORRP
Node State O(1) O(n
2
) O(n
3/2
) O(n
3/2
)
Reachability High High 100% High (99%)
Name Resolution O(n log n) O(1) O(1) O(1)
Invariants Geography None Global Comp. Local Comp.
ORRP state
scales with
Order N
3/2
ORRP states are
distributed fairly evenly
(no single pt of failure)

Reachability Finer Grained
Directionality (NS2 Simulations)
Observations/Discussions
For sparse networks, reachability
increases dramatically as number
of interfaces increases. This is
due to more node choices to
effectively route paths
Non-complete reachability even
with MAM due to network
fingers
Finer-grained interface
spread have increased
effectiveness in sparse
networks to a point
Finer-grained interface
spread increases reach in
networks with voids
Additional Results (in brief)
MAM increases reachability to almost
100% even in rectangular topologies in
NS2 simulations
Path stretch with MAM stays relatively
constant even with finer granularity of
antenna spread (discounting unreach)
Numerical Simulation of additional lines
yields very little REACH and PATH
STRETCH gain while adding a lot of
additional state

Summary
ORRP achieves high reachability in
random topologies
ORRP achieves O(N
3/2
) state
maintenance scalable even with flat,
unstructured routing
ORRP achieves low path stretch
(Tradeoff for connectivity under relaxed
information is very small!)
Future Work
Mobile ORRP (MORRP)
Hybrid Direction and Omni-directional nodes
More detailed abstraction to 3-D
Route loop prevention
ORRP for peer to peer networks requires the
concept of locally consistent virtual direction

Thanks!
Questions or Comments: chengb@rpi.edu


Affect of Control Packet TTL on
Varying Network Densities (NS2)
Observations/Discussions
Reachability increases heavily when
TTL is increased from 2 to 7 but stays
roughly constantly with continued
increases (Saturation Pt.)
Total States increases dramatically
from setting a TTL of 2 to 7 and then
stays constant
Average path length remains
unchanged with TTL
Reach increases
until Saturation Pt
with increase in TTL
Total States
increases until
Saturation Pt with
increase in TTL
Average Path
Length Remains
constant with
varying TTL
Additional Lines Study
Observations / Discussions
Probability of reach is not
increased dramatically with
addition of lines above 2
Path stretch is decreased
with addition of lines but not
as dramatically as between
1 and 2
Total States maintained is
increased heavily with
increase in number of lines
Motivation Hybrid FSO/RF
MANETs
Current RF-based Ad Hoc
Networks:
802.1x with omni-directional RF
antennas
High-power typically the most
power consuming parts of laptops
Low bandwidth typically the
bottleneck link in the chain
Error-prone, high losses

Free-Space-Optical
(FSO) Communications
Mobile Ad Hoc
Networking
High bandwidth
Low power
Directional secure,
more effective use of
medium
Mobile communication
Auto-configuration
Free-Space-Optical
Ad Hoc Networks
Spatial reuse and angular diversity in nodes
Low power and secure
Electronic auto-alignment
Optical auto-configuration (switching, routing)
Interdisciplinary, cross-layer design

State Complexity Varying Number of
Interfaces (NS2 Simulations)
Observations/Discussions
Total States increases with the
number of nodes in the network
(expected)
Total states is not very
dependent on the number of
interfaces
Increase in Total States
maintained consistent with
increased reachability (more
states = more reachability)
Stretch Average Path Length vs.
Varying Interfaces (NS2 Simulations)
Observations/Discussions
As node density increases, path
length increases as next hop
nodes are chosen at random from
the nodes within the transmission
range + LOS. With more nodes,
there is more choices of closer
nodes
Average Path Length improves
for dense networks with more
interfaces. More interfaces
increases granularity and limits
node selection
ORRP Introduction
Assumptions
Neighbor Discovery
1-hop neighbors
Given direction/interface
to send packets to reach
each neighbor
Local Sense of Direction
Ability to
Transmit/Receive
Directionally
Directional, smart
antennas
FSO transceivers

Deviation Correction:
Multiplier Angle Method (MAM)
Number of Interfaces (|) The angle node received packets from
Received Angle () The angle node received packets from
Deviation Angle (u) The angle to add/subtract that previous node deviated from
desired angle when sending
Desired Angle (o) The desired angle to send out
Found Angle (|) The angle of transceiver found with neighbor closest to desired
angle
Separation Angle (t) The angle of separation between each transceiver
Multiplier (m) The value to multiply t by to find new desired angle
t
o |
u t o
t
| t
u
t
| t
u

=
+ =

|
.
|

\
|

|
.
|

\
|

+ =
m
negative m m
positive m m
, 2 ,
4
) (
min
, 2 ,
4
) (
min
Important Notes:
1. Only corrections outside of antenna
spread considered
2. MAM assumes that relative distances
from one hop to another are relatively
equal
3. All deviation correction done at RREQ
and ORRP Announcement level (not
on each transmission)
ORRP Packet Deviation Issue
Sending in orthogonal
directions increases
likelihood of intersections
(Single line: 69% intersection vs.
Orthogonal Lines: 98% intersection)
Packet deviation potentially
lowers the likelihood of
intersections (ie: if packets end
up traveling in parallel paths)
Question: How can we
maintain straight paths as
much as possible without
adding too much overhead to
the system?

Thanks!
Can directionality be used at
Layer 3? YES!

You might also like