Location-Based Routing for Vehicular Ad-Hoc Networks原码、反码、补码的表示范围
Holger F ¨
ußler,Hannes Hartenstein,Dieter Vollmer
Martin Mauve
Michael K¨asemann
University of Mannheim
NEC Europe Ltd.
DaimlerChrysler AG
{fuessler,mauve }
@informatik.uni-mannheim.de
{hannes.hartenstein,
evaluate函数excel里哪个版本有
michael.kaesemann }@de
dieter.vollmer@daimlerchrysler
1.INTRODUCTION
Communication between vehicles is likely to be one key area where mobile ad-hoc networks will be used in the near future.Cur-rently there are several projects ([1,3])that investigate this appli-cation area,while car manufacturers and their suppliers aim at the development of products within the next few years.This raises the question what kind of routing algorithm is suited for communica-tion between vehicles.Existing performance studies for various ad-hoc routing protocols do not use movement patterns which re-semble the movement of cars and are thus not well suited to answer this question.
We have conducted an extensive simulation study based on re-alistic vehicle movement patterns.Our main aim was to investi-gate how a topology-based approach compares to a location-based routing scheme when applied to vehicular networks.As repre-sentatives we have chosen DSR [4]as a topology-based approach and GPSR [5]complemented with a simple location service as a location-based scheme.For a general discussion of topology-based and location-based ad-hoc routin
g please refer to [2]and [6].
Vehicular ad-hoc networks have several unique characteristics.The main challenge is the high speed with which nodes move in respect to each other.If oncoming traffic is included in the for-warding of packets,then relative speeds of 250km/h to 300km/h are common.This implies a very high rate of topology changes.Also it is not clear if one single partition spans sufficient distance to enable meaningful applications such as emergency warnings or vehicle-to-vehicle entertainment.On the other hand it is safe to assume that battery power is not an issue for vehicular ad-hoc net-works.Location-based approaches further benefit from the fact that vehicles are aware of their geographical positions through the use of on-board navigation systems.
2.PRELIMINARY RESULTS
2.1Real-World Movement Scenarios
Generating realistic vehicular movement patterns is a direction of research which is of fundamental interest to vehicle ,to determine how certain parts of a vehicle will last until they have to be replaced.As the vehicle movements are generated by a ‘pre-process’and complexity is therefore a minor concern,we decided to use a Driver Behavior Model for the traffic simula-tion.Such
a model not only takes the characteristics of the vehi-cles into account but it also includes a model of the driver’s behav-ior,like lane changing and passing decisions,traffic regulation and
Copyright is held by the author/owner.
MOBICOM’02,September 23–28,2002,Atlanta,Georgia,USA.ACM 1-58113-613-7/02/0009.
traffic sign considerations,or decreasing speed in curves,to name only a few.Driver Behavior Models are known to be highly accu-rate.As a simulator we used the well validated DaimlerChrysler-internal driver behavior simulation tool called FARSI.In particu-lar FARSI simulations show realistic speeds,distances,and macro-scopic properties like traffic flow and lane usage.Thus,FARSI guarantees that the vehicle movement patterns forming the basis of our experiments are as realistic as possible.
With FARSI we generated movement patters for (German)high-way traffic as well as traffic within parts of Berlin.The key contri-bution of our work is to study these patterns and the behavior of the routing protocols analytically and by means of discrete simulation using the ns-2network simulator.An example for an analytical graph of a highway scenario can be found in Figure 1.This fig-ure displays the connectivity of the ad-hoc network depending on the radio range and whether or not to use onco
ming traffic for the forwarding of packets.
An important observation of this analysis is the need to use on-coming traffic if the radio range is smaller than about 500meters.This leads to high relative speeds and low topological stability.
2.2Comparison of Routing Strategies
For the comparison of location-based and topology-based rout-ing algorithms,we choose GPSR [5](location)and DSR [4](topol-ogy),since they are well known representatives of the respective classes.The comparison done in [5]unrealistically assumed that each node knows the position of each other node.Furthermore the movement patterns of the nodes were random and did not allow an assessment for vehicular ad-hoc networks.
We ported the original implementation of GPSR to ns-2.1b8a and extended it with a simple reactive location service (RLS),in-spired by DSR route discovery:whenever the position of a node is required,the node looking for position information floods a request containing the ID of the node it is looking for.The request contains the ID and position of the requesting node.When a node receives a request with its own ID,it replies to the node looking for its posi-tion.In order to reduce the range of the flooding an expanding ring search is performed:the flooding starts with a range of 2hops and is re
peated with exponentially increasing range when no response is received during a certain time.For DSR we used the standard distribution which is part of ns-2.As MAC layer IEEE 802.11was employed,while the communication partners were chosen such that there is always a valid path from the source to the destination.Figure 2shows the main results of the comparison:the rate of successfully delivered packets for DSR diminishes when the max-imum communication distance becomes larger.This is caused by the fact the DSR needs to maintain a route from the sender to the receiver which becomes harder when the length of the route in-creases.The location-based approach stays close to the perfect
5
10
15
20050010001500
20002500300035004000
[m ]
[m]
(a)Connectivity-snapshot ignoring oncoming traffic 05
1015202530
35100
200300400
5006007008009001000
a v g  n o  o f  p a r t i t i o n s  i n  10k m对数函数公式大全及图解
radio range [m]
both directions one direction
(b)Average number of partitionsoverride中国
Figure 1:Network partitioning in a highway scenario
5060708090100500
1000
150020002500300035004000
4500
p a c k e t  d e l i v e r y  r a t i o  [%]
maximum communication distance [m]
802.11
正则表达式数字检验
Figure 2:Packet Delivery Ratio of DSR and GPSR
packet delivery rate of 100%for all distances.This can be ex-plained by the properties of location-based approaches:packet drops can occur only for one of the following three reasons:(1)if a local maximum is reached.This is extremely unlikely in our scenario.(2)If the information about the position of the local neighbors is inaccurate.Again this is very unlikely since the flooding of the lo-cation service in combination with piggy-backed beacons will pro-vide nearly perfect information about the neighbors.(3)If the in-formation about the position of the destination is inaccurate.This is also very rare,since the reply containing the position of the des-tination requires only minimal time to reach the sender,thus it is very accurate when the data packet is transmitted.
We are currently in the process of comparing the performance of both routing protocols in city enviro
nments.The generation of traffic patterns is complete and the obstacle modeling is also done.At this time simulations are underway.
3.CONCLUSIONS AND OUTLOOK
In this work we have shown that location-based routing is likely to be an appropriate scheme for routing in vehicular ad-hoc net-works,at least for highway environments.The direct usage of ex-isting topology-based mechanisms seems less appropriate.How-ever,it may be possible to tune topology-based by preferring routes over vehicles driving in the same direction.Cur-rently we investigate city scenarios to allow a broader view on the suitability of the distinct routing approaches.In addition to the an-alytical and theoretical evaluation we are going to conduct exper-iments using a fleet of 10DaimlerChrysler Smart-class vehicles.Although the multihop-capability of such a small network may not be highly representative,we still expect valuable insights by con-necting theory and the real world.
Acknowledgments
This work has been carried out within in the framework of the ‘FleetNet’project as part of BMBF (German Ministry for Educa-tion and Research)contract no.01AK025.
4.REFERENCES
position和location的区别
[1]R.Morris ,J.Janotti ,F.Kaashoek ,J.Li ,and D.S.
DeCouto.Carnet:A scalable ad hoc wireless network system.In Proc.of 9th ACM SIGOPS ,page 127ff,Kolding,Denmark,September 2000.
[2]E.M.Royer and C.-K.Toh .A review of current routing
protocols for ad-hoc mobile wireless networks.IEEE Personal Communications ,pages 46–55,April 1999.[3]The FleetNet project.www.fleetnet.de.
[4]D.B.Johnson and D.A.Maltz.Dynamic source routing in ad
hoc wireless networks.In T.Imielinske and H.Korth,editors,Mobile Computing ,volume 353.Kluwer Academic Publishers,1996.
[5]B.Karp and H.T.Kung.GPSR:Greedy perimeter stateless
routing for wireless networks.In Proc.of MOBICOM’00,pages 243–254,Boston,MA,U.S.A.,August 2000.[6]M.Mauve,J.Widmer,and H.Hartenstein.A Survey on
Position-Based Routing in Mobile Ad-Hoc Networks.IEEE Network ,15(6):30–39,November/December 2001.

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。