Proposed ad hoc Routing Approaches• Conventional wired-type schemes (global routing,

proactive):– Distance Vector; Link State

• Proactive ad hoc routing:– OLSR, TBRPF

• On- Demand, reactive routing:• DSR (Source routing), MSR • AODV (Backward learning)• AODV-DFR

• Scalable routing :– Hierarchical routing: HSR, Fisheye– OLSR + Fisheye– LANMAR (for teams/swarms)

• Geo-routing: • GPSR, GeRaF, etc• Motion assisted routing

“Direction” forwarding for mobile, large scale ad hoc networks

• In Distance Vector Routing (e.g., Bellman Ford, AODV etc.) node keeps pointer to “predecessor”

• When the predecessor moves, the path is broken • Alternate paths, even when available, are not used

Sink

Source

DV updatePredecessorData flow

Proposed solution: direction forwarding

Distance Vector not robust to mobility

Dest

Source

Primary PredecessorPrimary Path

Direction to Dest

Alternate Data Path

DV Update

Direction Forwarding

• Distance Vector update creates not only “predecessor”, but also “direction” entry

• Select “most productive” neighbor in forward direction

• If the network is reasonably dense, the path is salvaged

How to compute the “direction” Need “stable” local orientation system (say,

virtual compass) to determine direction of update Local (rather than global) reference is

required; Local reference system must be refreshed

fast enough to track avg local motion GPS will do (e.g., neighbors exchange (X, Y)

coordinates) If GPS not available, several non-GPS

coordinate systems have been recently published Sextant [Mobihoc ’05]; beacon DV; RFID’s

etc

Computing the “direction”(cont)

Compute “direction” to a destination when DV updates are received: If a DV update packet with a more recent

Seq # or smaller hop distance is received: New “direction” replaces the old one

The “direction” to the predecessor is used as the “direction” to the destination

If multiple DV updates received from different “predecessors” with same hop distance and seq # for the destination Take vector sum of directions

Computation of the “direction”

)(tan

)()(

12

121

212

212

XX

YY

YYXXr

−−

=

−+−=

−θ

Computation of the “direction”

Where the polar angle is the radian from the x-axis that is used as the direction of the predecessor node.

Suppose node A receives DV update packets from B & C

Compute the “directions” to predecessors node B & C, respectively,

A

C

B),( bb rθ

),( cc rθ

)1,( cθ

)1,( bθ

dθ

“Direction” to a destination

Unit vectors are used to combine the two “directions”

Directions to predecessors

Direction Forwarding vs Geo routing

• Geo-routing:– Direction points to destination– This direction may be unfeasible (holes, etc)– Global geo-coordinates (eg, GPS)– Geo Location Server– Robust to mobility

• Direction Forwarding– Direction of updates (always feasible)– Local (not global) position reference system– Advertisements from destination– Robust to mobility

Robust Ad Hoc Routing for Lossy Wireless Environment

• Challenges for routing in mobile ad hoc network

– Route breakage

– High BER

– Scalability

• The shortcomings of on-demand routing

• Not scalable for mobility

• The shortcomings of proactive routing

• Constant and high routing overhead

• The shortcomings of current Geo-routing

• Need Geo-Location Service, GLS

• “Face routing” is inefficient

Hybrid Routing: AODV-DFR

(AODV with Directional Forwarding Routing) • Combines on-demand and proactive routing

– When a source starts comm, it first finds the destination as in an on-demand fashion

– Once the destination is notified, it initiates periodic routing updates in a proactive fashion

• Utilizing an alternate path instantly based on “direction” to the destination if a path fails– resemblance with Georouting in the update

message– No location server system is required (not

like GPSR)

AODV-DFR

• Source initiates route discovery a la AODV – Destination, or any node that has a route,

replies– The path is set up

• Destination begins proactive advertisements (a la DV) after receiving data pkts from source – Intermediate nodes rebroadcast ads– Only for active connections– Period increases with distance from

destination (Fisheye concept)• Packet routing assisted by Direction Forward• The destination stops advertisement if it does

not receive pkts for some time

Performance Evaluation

• Compare AODV, AODV-DFR, GPSR and ADV (proactive and on-demand Hybrid Routing)– Performance: Delivery ratio, Packet delay,

Routing Overhead– Mobile & lossy network: UDP and TCP traffic

• Mobility Speed• Packet loss: uniformly distributed on a link

• Simulation– 100 nodes randomly moving in 1000x1000m– The traffic pairs are randomly distributed

over the network– UDP flows: pkt size 512 bytes, rate 1pkt/sec– TCP flows: NewReno, pkt size 1460 bytes

Mobile Network: Delivery Ratio

80 UDP flows

Mobile Network: Packet delay

80 UDP flows

Mobile Network: Routing Overhead

80 UDP flows

Mobile & Lossy Network: Delivery Ratio

UDP Flow number: 80 Mobility Speed: 10 m/s

Mobile & Lossy Network: Routing Overhead

UDP Flow number: 80 Mobility Speed: 10 m/s

TCP in Mobile Network

40 TCP flows

TCP in Mobile & Lossy Network

TCP flow number: 40 Mobility: 10 m/s

AODV-DFR Contributions

• A hybrid routing: proactive + on-demand• Robust to mobility and packet loss• Utilize location information for directional

forwarding with only local updates.• Low overhead• Provide better performance than AODV and

GPSR• Enhances AODV• Competitive with GPSR (does not require

“global” positioning such as GPS)• Ongoing work: local coordinate system;

integration of local and global coordinates (indoor+outdoor)

Robust Ad Hoc Routing for Lossy Wireless Environment

CS 218 Fall 08

UCLA: Jiwei Chen, Yeng-Zhong Lee, Mario Gerla

TJU: He Zhou, Yantai Shu

Milcom 2006

Introduction

• Challenges for routing in mobile ad hoc network– Route breakage– High BER– Scalability

• The shortcomings of on-demand routing• Not scalable for mobility

• The shortcomings of proactive routing• Constant and high routing overhead

• The shortcomings of current Geo-routing• Need Geo-Location Service, GLS• “Face routing” is inefficient

ENTER AODV-DFR!

Related Work• Proactive Routing

– OLSR, DSDV

• On-demand Routing– AODV, DSR

• Geographic Routing– GPSR and several others– greedy + face algorithm– GLS (geo location service)

• Hybrid Routing– ZRP, SHARP, LANMAR zone and group concept– ADV, proactive and on-demand– DREAM, proactive and geo-routing– AODV-DFR

AODV-DFR

• Source initiates route discovery a la AODV – Destination, or any node that has a route, replies– The path is set up

• Destination begins periodic advertisements (a la Distance Vector) after receiving data packets from source – Intermediate nodes rebroadcast advertisements - interval increases

with distance from destination (Fisheye concept)• Packet routing assisted by Direction Forwarding• The destination stops advertisement if it does not receive packets for

some time.

Proactive Advertising in AODV-DFR

• Only for active connections • “Fisheye” updating frequency

Dest

Source

Primary PredecessorPrimary Path

Direction to Dest

Alternate Data Path

DV Update

Direction Forwarding

• Distance Vector update creates not only “predecessor”, but also “direction” entry

Sink

• “Predecessor only” forwarding fails• DFR selects “most productive” neighbor in right direction• If the network is reasonably dense, the path is salvaged

Source

DV updatePredecessorData flow

“Direction” to Sink

Directional Forwarding

• Each node remembers the local advertised directional

• Only local coordinates needed • Direction is computed by the

aggregation of local updates• “Multiple hop” direction can

also be available.• Next node with min-hop and

minimal deviation from the direction is selected.

A

C

B

A

C

B

(XC, YC)

(XB, YB)

(XA, YA)|AB|

AB

|AC|

AC→

+

→

Routing Tables at a Node

• Neighbor Table – Direction to all

neighborsNB ID

Seq

Num

Direction to NB

Dest ID

Hop Count

• Routing table– Routes to all active

destinations

Dest ID

Direction to Dest

Next Hop

Hop Count

Performance Evaluation

• Compare AODV, AODV-DFR, ADV, and GPSR (without charge for GLS overhead).

• Performance in mobile network– Delivery ratio– Routing Overhead

• Performance in mobile & lossy network– Packet loss: uniformly distributed on a link– UDP and TCP traffic

Simulation Environment

• 100 nodes randomly moving in a 1000x1000m space

• The traffic pairs are randomly distributed over the network.

• UDP flows: pkt size 512 bytes, rate 1pkt/sec

• TCP flows: data pkt size 1460 bytes, NewReno

Mobile Network: Delivery Ratio

80 UDP flows

Mobile Network: Routing Overhead

80 UDP flows

Mobile & Lossy Network: Delivery Ratio

UDP Flow number: 80 Mobility Speed: 10 m/s

TCP in Mobile Network

40 TCP flows

AODV-DFR Contributions• A hybrid routing: proactive + on-demand• Robust to mobility and packet loss• Utilize location information for directional forwarding with

only local updates.• Low overhead• Enhances AODV• Competitive with GPSR:

– not affected by GLS and by face routing issues; – does not require “global” positioning such as GPS

• Ongoing work: local coordinate system; integration of local and global coordinates (indoor+outdoor)