張郁昌 R91921076 NTUEE Computer Science Group Sophomore
Summary:
An ad hoc network is a collection of wireless mobile nodes dynamically forming a temporary network without drawing on the existing network infrastructure or central control. 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.
Over the past few years, a variety of new routing protocols targeted specifically at the ad hoc networking environment have been proposed, but little performance information on each protocol and no detailed performance comparison between the protocols has previously been available.
They extended the ns -2 network simulator to model node mobility, realistic physical layer including a radio propagation model, Radio network interfaces, and The IEEE
802.11 Medium Access Control (MAC) protocol. This new simulation environment provides a powerful tool for evaluating ad hoc networking protocols and other wireless protocols and applications. Then, using this simulation environment, they present the results of a detailed packet-level simulation comparing four recent multi-hop wireless ad hoc network routing protocols. These protocols, DSDV, TORA, DSR, and AODV, cover a range of design choices, including periodic advertisements vs. on-demand route discovery, use of feedback from the MAC layer to indicate a failure to forward a packet to the next hop, and hop-by-hop routing vs. source routing. They simulated each protocol in ad hoc networks of 50 mobile nodes moving about and communicating with each other, and presented the results for a range of node mobility rates and movement speeds.
Each of the protocols studied performs well in some cases yet has certain drawbacks in others. DSDV performs quite predictably, delivering virtually all data packets when node mobility rate and movement speed are low, and failing to converge as node mobility increases. TORA, although the worst performer in our experiments in terms of routing packet overhead, still delivered over 90% of the packets in scenarios with 10 or 20 sources. At 30 sources, the network was unable to handle all of the traffic generated by the routing protocol and a significant fraction of data packets were dropped. The performance of DSR was very good at all mobility rates and movement speeds, although its use of source routing increases the number of routing overhead bytes required by the protocol. Finally, AODV performs almost as well as DSR at all mobility rates and movement speeds and accomplishes its goal of eliminating source routing overhead, but it still requires the transmission of many routing overhead packets and at high rates of node mobility is actually more expensive than DSR.

Evaluation:
Before this paper, a variety of new routing protocols targeted at ad hoc network have been developed, but little performance information on each protocol and no realistic performance comparison between them is available. The authors did the comparison well and we could easily get the ideas of the conclusions.
Despites of the result, it’s interesting to note that their implementation made some changes to the original design of the four protocols. I am not very sure if these changes are necessary for the original designs, or if there are still room for further improvements.
I am sure that different models and different input parameters may lead to very different results, needless to say the protocols designed with intuition and without modeling. We are not wise enough and real experiments are not always practical, here come simulations for rescue.
This paper is good at clarity and significance. And according to the claim, they model the physical layer properties much better than others, the results will be correct. Finally, I think there isn’t much originality in this kind of comparison paper.
Future directions:
There may be two directions. First, almost all of the ad hoc routing protocols design for pure ad hoc environment only. As more hotspots come around us, more chances will be mixed type of routing environment—fixed access points sitting there with ad hoc routing to extend the Internet accessibility. And the usage pattern may be lots of traffic between fixed AP and mobiles nodes. I think it’s important to use a routing protocol that works properly both on pure ad hoc environment and the mixed type of routing environment.
Second, there are many wireless protocols today. It may be good to use a single protocol suit over Bluetooth, 802.11, Zig Bee, and Ultra Wide Band. We could simulate this kind of environment by modeling the physical layer properties.