Transcript Communication Systems and Networks

Ad-hoc Routing
Michalis Michaelides
Dept. of Electrical and Computer Engineering
University of Cyprus
ECE 654 – Computer Networks Seminar
November 12, 2004
Presentation Overview









Motivation
Ad-hoc Networks
Ad-Hoc Routing Challenges
Ad-Hoc Routing protocols
DSR- Dynamic Source Routing
AODV- Ad-Hoc on Demand Distance Vector
DSR vs. AODV comparison
Sensor Networks
Conclusions
References










Josh Broch, David Johnson, and David Maltz. The Dynamic Source Routing protocol for mobile ad
hoc networks. Internet draft (work in progress), Internet Engineering Task Force, October 1999.
Josh Broch, David A. Maltz, David B. Johnson, Yih-Chun Hu, and Jorjeta Jetcheva. A performance
comparison of multi-hop wireless ad hoc network routing protocols. In Proc. ACM/IEEE MobiCom,
pages 85–97, October 1998.
David B. Johnson. Routing in ad hoc networks of mobile hosts. In Proc. of the IEEE Workshop on
Mobile Computing Systems and Applications, pages 158–163, December 1994.
Charles Perkins, Elizabeth Royer, and Samir R. Das. Ad hoc On demand Distance Vecor (AODV)
routing. Internet draft (work in progress), Internet Engineering Task Force, October 1999.
Charles E. Perkins and Pravin Bhagwat. Highly dynamic Destination-Sequenced Distance-Vector
routing (DSDV) for mobile computers. In Proc. ACM SIGCOMM Conference (SIGCOMM ’94),
pages 234–244, August 1993.
E. M. Royer and C-K. Toh. "A review of current routing protocols for ad-hoc mobile wireless
networks." IEEE Personal
Communications, April 1999.
D. Braginsky, D. Estrin, Rumor Routing Algorithm For Sensor Networks. WSNA 2002
Karp, B., and Kung. H. T. "GPSR: Greedy Perimeter Stateless Routing for Wireless Networks." In
Proceedings of the 6th Annual International Conference on Mobile Computing and Networking
(MobiCom 2000), 243-254, 2000.
W. R. Heinzelman, A. Chandrakasan, and H. Balakrishnan, "Energy-efficient communication
protocol for wireless micro-sensor networks" In Proc. IEEE Hawaii Int. Con. on System Sciences,
pages 4-7, January 2000.
J. E. Wieselthier, G. D. Nguyen, and A. Ephremides, "On the construction of energy-efficient
broadcast and multicast trees in wireless networks" In Proceedings of IEEE INFOCOM, pages
585-594, March 2000.
Motivation



Mobile hosts such as notebook computers featuring
powerful CPUs, large main memories and disk space,
multimedia capabilities and colour displays are now quite
common in everyday business and personal life.
At the same time network connectivity options for use
with mobile hosts have increased dramatically, including
support for a growing number of wireless networking
products based on radio and infrared.
Natural desire and ability to share information between
mobile users:




Employees in a conference room
Friends in an airport terminal
Search and rescue teams
Military data acquisition operations in inhospitable terrain
Ad-hoc Networks



A collection of wireless mobile hosts dynamically
forming a temporary network without the use of
any existing network infrastructure or centralized
administration.
Due to the limited transmission range of wireless
network interfaces, multiple network “hops” may
be needed for one node to exchange data with
another across the network.
Need a dynamic routing protocol that can
efficiently find routes between two nodes.
Ad-hoc Routing
Source
Destination
Ad-hoc Routing Challenges




Correct and efficient route establishment between a pair
of nodes so that messages may be delivered in a timely
manner.
Keep up with high degree of node mobility.
Conserve power- power aware routing
Wireless medium challenges:





Links may not be bi-directional
High error rates (Hidden Terminal Problem)
Redundant paths
Limited BW
Multicast
Ad-hoc routing protocols history


DARPA (Defence Advanced Research Projects) packet radio
networks in early 1970s was the first attempt.
Since then numerous protocols have been developed:

Table-driven




Source-initiated On-Demand




Attempt to maintain consistent up-to-date routing information from each
node to every other node in the network.
DSDV, CGSR, WRP
Substantial traffic and power consumption caused by periodic updates but
routes are always available.
When a node requires a route to destination it initiates a route discovery
process within the network.
AODV, DSR, TORA, ABR, SSR…
MANET (Mobile ad hoc networking) has been formed within the
IETF (Internet Engineering Task Force) to develop a routing
framework for IP-based protocols in ad hoc networks.
2 primary candidates: DSR and AODV
Dynamic Source Routing (DSR)



Source routing - The sender knows the complete hop-byhop route to the destination.
Route cache - Nodes may learn and cache multiple
routes to any destination.
Composed of 2 mechanisms:





Route Discovery
Route Maintenance
Requires no periodic packets of any kind at any level
within the network- purely on demand.
Allows uni-directional links.
Supports internetworking between different types of
wireless networks and mobile IP.
Route Discovery

RREQ (Route Request packet)
 Is
broadcast when node S needs do send a packet to
D and does not already know a route.
 Each RREQ includes source and destination address,
unique request id and complete route record of all
intermediate nodes.

RREP (Route Reply packet)
 If
a node receives an RREQ for which it is either the
destination or it has a route to the destination in its
route cache it responds with a RREP.
DSR
N1-N2-N5-N8
N2
RREP
N1-N2-N5-N8
N1-N2-N5-N8
N8
N5
N1-N2
N1
N1
N1-N2-N5
N1-N3-N4
Source
N1-N3-N4
N4
N1
Destination
N1-N3-N4-N7
N7
N1-N3-N4
N1-N3-N4-N6
N1-N3
N3
N6
RREQ
Route Maintenance

Hop-by-hop acknowledgement




Link-level acknowledgement IEEE 802.11
Passive acknowledgement (Overhearing)
DSR specific acknowledgement
RERR (Route Error packet)


Informs the source of any broken link.
Source removes any routes containing broken link from route
cache.
Additional Optimizations

Packet Salvaging


An intermediate node can use an alternate route from its own cache in case of a
failed link.
Gratuitous route repair

A source node receiving an RERR packet piggybacks the RERR in the following
RREQ.
 Helps clean up cashes of other nodes in network.

Promiscuous listening

When a node overhears packet checks to see whether it could be routed via
itself to gain a shorter route and sends a gratuitous RREP to source.
 Learn different routes without participating in routing process.
Ad-hoc On-Demand Distance Vector
Routing (AODV)






Discovers routes on-demand.
Uses traditional routing tables, one entry per destination
that are dynamically established at each intermediate
node.
Use ‘hello’ messages for local connectivity management.
Sequence numbers maintained at each destination to
determine freshness of routing information and to
prevent rooting loops.
Timer-based states in each node regarding utilization of
individual routing table entries.
Expanding ring search optimization.
Path Discovery

Every node maintains two separate counters:
Node sequence number – Maintain freshness information of route
 Broadcast id – Incremented for every new RREQ


RREQ (Route Request packet)

<Source and destination address ,source and destination sequence
number, broadcast id, hop count>
 Each node that cannot satisfy the RREQ rebroadcasts to its own
neighbours after increasing hop count.
 Each node keeps expiration timers to remove old RREQ and routes
from its cache.

RREP (Route Reply packet)


Unicast back to the neighbour from which it received the first RREQ.
<Source and destination address, destination sequence number, hop
count, lifetime>
AODV
RREP
N2
N5
N8
Destination
N1
Source
N4
N3
N7
N6
RREQ
Path Maintenance

Detecting link failures
 Periodic
‘hello’ messages
 Link Layer acknowledgements (LLACKS)
 Attempts to forward packet to next hop fail

RERR (Route Error packet)
 Created
when next-hop link breaks.
 Propagated to all predecessors until all sources using
the failed link are informed.
 Sources restart discovery process if they still need the
route to destination.
DSR

vs.
Source routing


Route caching
Only next-hop information
 Periodic “hello” messages for
local connectivity





Supports uni-directional links
Only broadcast
No mechanism to expire stale
routes or prefer ‘fresher’ routes
RERR backtracks the data
packet
One entry per destination

More routing information
 Fast recovery from failure
 More RREP

Dynamic routing tables

More routing overhead
 No periodic routing
advertisements

AODV
Limited routing information
 New RREQ for every failure
 More RREQ




Only bi-directional links
Multicast capability
Expiration timers remove stale
routes and sequence numbers
RERR informs all predecessor
nodes of link failure
DSR-AODV performance comparison

Simulation model








ns-2 model for simulating multihop wireless networks
Virtual carrier sense with RTS/CTS to reduce hidden terminal problem.
802.11 Link layer acknowledgements for each packet transmission.
Send buffer with FIFO and routing packets getting higher priority than
data packets.
Continuous bit rate (CBR) sources and destinations randomly spread in
rectangular field.
Nodes move with randomly chosen speed, pause and then move again.
Vary number of sources, node mobility, traffic load
Performance metrics

Packet delivery fraction (Throughput)


Routing load


Ratio of packets delivered to destination to those generated by the CBR
Routing packets transmitted per data packet delivered
Path optimality or Delay (QoS)
DSR-AODV performance comparison
simulation results

DSR authors
 DSR
has the lowest routing overhead.
 Both DSR and AODV exhibit similar good
performance compared to DSDV,TORA.

AODV authors
 DSR
outperforms AODV in less stressful situationssmall number of nodes and lower mobility. AODV
outperforms DSR in more stressful situations.
 DSR consistently generates less routing load than
AODV.
Sensor Networks
Sensor networks are a special class of Ad-Hoc Networks
DIFFERENCES
 Number of nodes- scalability
 Densely deployed
 Prone to failure (power outage)
 Topology may change frequently
 Limited power, computational capability and memory
 Broadcasting vs. point-to-point
 Sensor nodes may not have global identification ID
because of the large amount of overhead and large
number of sensors

What are sensor networks?
Sink
Internet or
Satellite
network
Task Manager
Node
sensor
field
sensor
nodes
Ad-hoc routing for sensor networks

Geographic routing



Probability routing



GPSR (Greedy perimeter stateless routing) uses positions of immediate
neighbours and a packet’s destination to make greedy forwarding
decisions.
Outperforms DSR and demonstrates scalability on densely deployed
networks.
Gossip- Each node forwards a message with some probability to reduce
overhead of routing protocol.
Adding gossiping to AODV results in significant performance
improvement.
Minimum Energy

LEACH (Low Energy Adaptive Clustering Hierarchy)- Clustering based
protocol that uses randomization to distribute energy dissipation.
 BIP (Broadcast Incremental Power)- Uses Wireless Multicast Advantage
Conclusions



Mobile Ad hoc Networks are becoming part of
our everyday lives.
DSR and AODV both provide simple on-demand
routing solutions for ad-hoc networks.
They fail however to scale to large networks with
limited energy resources so further
enhancements are possible using position,
probability, clustering, power management…