A Performance Comparison of Multi-Hop Wireless Ad Hoc Network Routing Protocols
by Josh Broch, David A. Maltz, David B. Johnson, Yih-Chun Hu, Jorjeta Jetcheva
url show details
Details
| organization: | ACM/IEEE | | address: | New York | | abstract: | An ad hoc network is a collection of wireless mobile nodes 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 | | month: | oct # " 25-30, | | annote: | Josh Broch (Computer Science Department; Carnegie Mellon University; Pittsburgh , PA 15213); David A. Maltz (Computer Science Department; Carnegie Mellon University; Pittsburgh , PA 15213); David B. Johnson (Computer Science Department; Carnegie Mellon University; Pittsburgh , PA 15213); Yih-chun Hu (Computer Science Department; Carnegie Mellon University; Pittsburgh , PA 15213); Jorjeta Jetcheva (Computer Science Department; Carnegie Mellon University; Pittsburgh , PA 15213); | | type: | misc | | pages: | 85--97 | | year: | 1998 | | publisher: | ACM Press | | booktitle: | The fourth annual ACM/IEEE international conference on Mobile computing and networking |
|
|
You need to log in to add tags and post comments.
It was a very good read about various ad-hoc routing schemes and comparison analysis of their performance under a simulated environment.
I don't have good background in this area, so it might be a bit stupid question. But, what are the evaluation metrics and usage models based on? Do we have a real deployment case for one of these routing protocols now? And, I think the problem of power consumption should be one of the important evaluation metrics, since we're focusing on highly mobile nodes where power consumption is a sensitive issue.
An interesting paper, and a good reading for someone who tries to understand the problems with ad hoc routing.
It seems clear that DSR and AODV are the best choice (TORA does not get the best route and creates too much overhead and DSDV is always outperformed by DSR and AODV).
I agree with Kang that it is important to compare these protocols in terms of power comsumption (although one could suspect that those protocols that create more overhead also consume more power).
The authors claim that other simulations by other authors suffered from some inaccuracy or other problems. Unfortunately it is likely that the simulations used in the paper also dismissed some details and the results may differ from real-world data.
Although I am not very familiar with this research area and some of the terms in this paper were new to me (e.g. virtual carrier sense) this was an interesting read. It gave a concise evaluation and thorough comparison of MANET routing protocols in terms of packet delivery ratio, routing overhead and path optimality.
When implementing the routing protocols for the ns simulator the authors made a few improvements. It is interesting that the synchronization prevention improvement is also adopted by the ETX protocol, which was presented in the other assigned paper. In this paper path optimality is directly related to hop-count. However, according to the ETX paper, this is usually not true for multi-hop wireless networks and it should not be used as a metric for choosing the best route.
I thought it was interesting and wasn't too surprised by the results. TORA, in which "route optimality (shortest-path routing) is considered of secondary importance, and longer routes are often used to avoid the overhead of discovering new routes)" was clearly the worst performer. This is not too surprising, and suggests that putting effort into finding optimal routes is a good idea. DSR seemed to perform the best overall. I found it a little surprising that DSR outperformed and is less expensive than AODV despite the fact that AODV attempts to combine the advantages of both DSR and DSDV.
Compared to other ad hoc networking papers we have been looking at, this is super old - 1998! I'm surprised by how many ad hoc protocols already existed back then, given that I'm not aware of any real ad hoc networks currently. Or are there some?
The paper provides good comparisons, and good information on four routing protocols in wireless ad-hoc network , namely DSDV, TORA, DSR, and AODV.
Just a minor question after reading these two papers, in ad-hoc network why do we need to avoid "synchronization"?
I was somewhat surprised that Dynamic Source Routing (DSR) did so well. I would have thought that the sender, being farther away from the destination than intermediate nodes, would not have as up-to-date information. I guess the lower overhead helped it out a lot, as well as various optimizations that it performed, such as: the initial non-propagating Route Request, hosts sniffing their neighbors' packets for route information, and intermediate nodes repairing a broken route and notifying the sender. It would be interesting to see how much each optimization helped, by simulations with and without the optimization.
This is a very nice written paper; it provides background introduction of ad-hoc network and routing protocols in reasonable details (possibly because it is an "old" paper, as Sue Ann pointed out), so it is a perfect member of the reading list of the course. For Anshul's comment, there is a SIGCOMM'04 paper on comparing routing metrics on static multi-hop wireless networks which are relevant to both of the readings. I also like Arvind's comment a lot, thank you!
ad-hoc wireless networks seem to present an interesting problem, and I wouldn't mind learning more about them. This paper provides a valuable practicle contribution: evaluating many of the protocols proposed in the literature. I would be interested to know how networking folks view simulation results like these: do they take into account enough of the "noise" of the real world to give high-confidence results?
Another interesting question seems to be how these protocols might react to a malicious user: can one guy with a laptop in the middle of an ad-hoc network destroy everyone's connectivity?
This paper gives a very comprehensive comparison for different protocols on ad-hoc networking routing. While the experiment is carried on ns-2, I am wondering if there is any real network experiment result for the comparison.
Does anyone know if these studies have been replicated either in the real world(as opposed to a simulator) or under an emulator? I think Srini mentioned in class that the behavior of these links is often very different from ideal/classical models, so I'm wondering how well the simulator was able to model these non-ideal scenarios.
This is a paper well written and easy to read. I like its style. In the paper, the author said they implemented the complete 802.11 DCF as well as different routing protocols in NS2. I think this work itself is as important as the result of comparing different routing protocols. Does NS2 adopt this implementation of 802.11 MAC eventually?