Routing Protocol Performance: Difference between revisions

From
Jump to navigation Jump to search
No edit summary
No edit summary
 
(7 intermediate revisions by the same user not shown)
Line 1: Line 1:
"Monarch" Paper: A Performance Comparison of Multi-Hop Wireless Ad Hoc Network Routing Protocols 1998
"Monarch" Paper
[[DSR]],[[AODV]],[[TORA]],[[DSDV]]
Protocols compared are [[DSR]],[[AODV]],[[TORA]],[[DSDV]]


[[AODV|AODV-LL]]
Special Implementations: [[AODV|AODV-LL]] and [[DSDV|DSDV-SQ]]
[[DSDV|DSDV-SQ]]


In this paper the performance is compared using the freely available ns2 network simulator.
Performance Vergleich mittels ns2


Simulation Parameters (still need to find the ns2 simulation input files)
Performance Unterschiede durch
(maybe here: [http://monarch.cs.cmu.edu/software.html ns2packets?])
Implementierung
* node mobility
Simulationsumgebung (Paketgröße,Netzauslastung,)
* realistic physical layer
* radio network interface
* IEEE 802.11 MAC layer (with link breakage detection - except DSDV !!!)

Each node buffers 50 packets. The protocol buffers additional 50 packets.

50 nodes in a 1500m x 300m area.
simulation time: 900 seconds
pause time (nodes did not move): 0 - 900 seconds
node speed: 1m/s - 10m/s (average) - 20m/s max.

Communication Aspects:

CBR sources: 10,20,30
packets delivered: 1,4,8 per second (!)
packet size: 64byte,1024 byte (smaller than ethernet frame)

The paper discovered that sending packets/second is almost equivalent to the number of traffic sources.
So they only simulate 4 packets per second.

The simulation was further restricted when they discovered that some nodes (1 or 2) drop almost all packets they received when using packet sizes of 1024 bytes (the density of the nodes might be one reason).
So they reduce the packet size to 64 byte - because they wanted to test routing mechanics.

Another important fact: they dont use TCP sources because the packets are send with respect to the network load and that prevents a direct comparison.

Loser:
* DSDV at higher mobility
* TORA at high load

Winner:
* DSR
* AODV


[[AODV]] vs. [[DSR]]
[[AODV]] vs. [[DSR]]
[http://www.cs.ucsb.edu/~ebelding/txt/aodv_infocom.ps AODV 2000]


In this paper the performance is compared using the freely available ns2 network simulator.
Performance Vergleich mittels ns2



Winner:
* AODV


[[AODV]] vs.[[OLSR]]
[[AODV]] vs.[[OLSR]]
[http://sar/teaching/2004%20Ad-Hoc%20Networks/readings/aodv-oslr-comparison.pdf Paper 2004]

In this paper network performance is compared on a theoretical base.

[http://www.tml.hut.fi/Studies/T-110.551/2004/papers/ Homepage]






other papers:
theoretischer Performancevergleich ohne Simulationen
[http://www.cs.ucsb.edu/~ebelding/txt/Perkins_PerfComp.pdf AODV 2001]
[http://menetou.inria.fr/~viennot/postscripts/medhocnet2002sim.ps.gz Comparative Study of Routing Protocols for Mobile Ad Hoc Networks 2002]

Latest revision as of 11:41, 4 November 2004

"Monarch" Paper: A Performance Comparison of Multi-Hop Wireless Ad Hoc Network Routing Protocols 1998 Protocols compared are DSR,AODV,TORA,DSDV

Special Implementations: AODV-LL and DSDV-SQ

In this paper the performance is compared using the freely available ns2 network simulator.

Simulation Parameters (still need to find the ns2 simulation input files) (maybe here: ns2packets?)

  • node mobility
  • realistic physical layer
  • radio network interface
  • IEEE 802.11 MAC layer (with link breakage detection - except DSDV !!!)

Each node buffers 50 packets. The protocol buffers additional 50 packets.

50 nodes in a 1500m x 300m area. simulation time: 900 seconds pause time (nodes did not move): 0 - 900 seconds node speed: 1m/s - 10m/s (average) - 20m/s max.

Communication Aspects:

CBR sources: 10,20,30 packets delivered: 1,4,8 per second (!) packet size: 64byte,1024 byte (smaller than ethernet frame)

The paper discovered that sending packets/second is almost equivalent to the number of traffic sources. So they only simulate 4 packets per second.

The simulation was further restricted when they discovered that some nodes (1 or 2) drop almost all packets they received when using packet sizes of 1024 bytes (the density of the nodes might be one reason). So they reduce the packet size to 64 byte - because they wanted to test routing mechanics.

Another important fact: they dont use TCP sources because the packets are send with respect to the network load and that prevents a direct comparison.

Loser:

  • DSDV at higher mobility
  • TORA at high load

Winner:

  • DSR
  • AODV

AODV vs. DSR AODV 2000

In this paper the performance is compared using the freely available ns2 network simulator.


Winner:

  • AODV

AODV vs.OLSR Paper 2004

In this paper network performance is compared on a theoretical base.

Homepage



other papers: AODV 2001 Comparative Study of Routing Protocols for Mobile Ad Hoc Networks 2002