Abstract—Coupling Vehicular Ad Hoc Networks ( VANETs ) with wired webs such as the Internet via entree points creates a dif?cult mix of extremely nomadic nodes and a inactive substructure. In order to measure the public presentation of typical ad hoc routing protocols—in peculiar. we used Dynamic MANET On Demand ( DYMO ) —in such VANET scenarios. we combined microsimulation of route traf?c and event-driven web simulation. Therefore. we were able to analyse protocols of the Internet protocol suite in VANET scenarios with extremely accurate mobility theoretical accounts. Changing parametric quantities of DYMO for a battalion of traf?c and communicating scenarios helped indicate out attacks for bettering the overall public presentation and revealed jobs with the deployment. It could be shown that in realistic scenarios. even for medium densenesss of active nodes and low web burden. overload behaviour leads to a drastic lessening of the sensed web quality. Cross-layer optimisation of conveyance and routing protocols hence seems extremely advisable.
Recent research in the country of Vehicular Ad Hoc Networks ( VANETs ) was chiefly focused on the development and
the rating of extremely specialised protocols. e. g. for the exchange of place information or jeopardy warnings between autos. Signi?cantly less work dealt with measuring the usage of bing Internet protocols. along with standard hard- and package. to make and keep VANETs and match these webs with the Internet. The Mobile Ad Hoc Network ( MANET ) working group of the Internet Engineering Task Force ( IETF ) develops criterions for routing in dynamic webs of both nomadic and inactive nodes. One protocol presently in the working group’s focal point is Dynamic MANET On Demand ( DYMO ) . It was conceived as replacement to the popular Ad Hoc on Demand Distance Vector ( AODV ) routing protocol. Its usage in the context of VANETs has already been extensively investigated. The DYMO protocol bill of exchange expressly provides for the yoke of a MANET with the Internet. which makes an rating of communicating connexions between nomadic nodes and inactive substructure particularly attractive. A auto taking portion in a MANET scenario could already set up such connexions in range of one of an of all time turning figure of public hot spots while driving in the metropolis. and a deployment of entree points along main roads in the close hereafter seems executable. Apparently. this yoke of MANET and Internet is particularly attractive for route users if it allows the use of virtually all bing resources of the Internet without trusting on expensive dedicated channels provided by a cellular web.
Routing information airing in AODV and DYMO
In this work. the feasibleness. the public presentation. and the bounds of ad hoc communicating utilizing DYMO were evaluated and potencies for optimising the deployed conveyance and routing protocols were investigated. Particular attention was taken to supply realistic scenarios of both route traf?c and web use. This was accomplished by imitating a assortment of such scenarios with the aid of two conjugate simulation tools. A microsimulation environment for route traf?c supplied vehicle motion information. which was so fed into an event-driven web simulation that con?gured and managed a MANET theoretical account based on this mobility informations. The protocols of the conveyance. web. information nexus. and physical beds were provided by welltested executions for the web simulation tool. while MANET routing was performed by our ain execution of DYMO.
II. DYNAMIC MANET O N D EMAND ( DYMO )
DYMO is a new reactive ( on demand ) routing protocol. which is presently developed in the range of the IETF’s MANET working group. DYMO builds upon experience with old attacks to reactive routing. particularly with the routing protocol AODV. It aims at a slightly simpler design. assisting to cut down the system demands of take parting nodes. and simplifying the protocol execution. DYMO retains proved mechanisms of antecedently explored routing protocols like the usage of sequence Numberss to implement cringle freedom. At the same clip. DYMO provides enhanced characteristics. such as covering possible MANET–Internet gateway scenarios and implementing way accretion as depicted in Figure 1. Besides route information about a requested mark. a node will besides have information about all intermediate nodes of a freshly discovered way. Therein lies a major difference between DYMO and AODV. the latter of which merely generates path table entries for the finish node and the following hop.
III. S IMULATION T OOLS
For the choice of a suited traf?c simulation tool. two facets had to be weighed against each other. Clearly. the
underlying traf?c theoretical account was to be every bit simple and comprehendible as possible. so that consistent consequences could be obtained. On the other manus. the simulation theoretical account needed to be complex plenty to bring forth realistic forms. which—as has been shown in related work—greatly in?uence the quality of consequences obtained from overlaid web simulations. Microsimulation of route traf?c was performed by an version of Traf?cApplet 1. an unfastened beginning traf?c microsimulation tool that provides an accurate theoretical account of microscopic driver behaviour. as opposed to the still common simplistic or proprietary behaviour theoretical accounts. It implements the microsimulation theoretical accounts IDM and MOBIL to cipher longitudinal and sidelong motion. severally. The behaviour of fake vehicles can be con?gured with simple parametric quantities like “desired velocity” or “comfortable acceleration” . which were used to pattern two different types of route users.
Nodes of type Truck traveled at a maximal velocity of 22. 2 m/s ( approx. 80 kilometers per hour. 50 miles per hour ) and made up 20 % of the vehicles simulated. The staying 80 % of vehicles were of type Car and traveled at velocities of up to 33. 0 m/s ( approx. 120 kilometers per hour. 75 miles per hour ) . All simulations were performed at a denseness of 4. 2 vehicles per kilometre and lane. stand foring every night traf?c. every bit good as at a denseness of 28. 0 vehicles per kilometre and lane. which modeled rush-hour traf?c. Sample velocity hints recorded in both scenarios are shown in Figure 2. Obviously. utilizing a smaller figure of fake vehicles allowed the autos to travel about unimpaired by trucks or other autos and to go at or near top velocity. The scenario therefore maximized velocity differences between nodes. so links between autos of different lanes. between autos and trucks. every bit good as between vehicles and roadside substructure were while DYMO shops routes for each intermediate hop. This is illustrated in Figure 1. When utilizing AODV. node A knows merely the paths to nodes B and D after its path petition is satis?ed. In DYMO. the node to boot learned a path to node C. To ef?ciently cover with extremely dynamic scenarios. links on known paths may be actively monitored. e. g. by utilizing the MANET Neighborhood Discovery Protocol or by analyzing feedback obtained from the informations link bed. Detected nexus failures are made known to the MANET by directing a path mistake message ( RERR ) to all nodes in scope. informing them of all paths that now became unavailable. Should this RERR in bend invalidate any paths known to these nodes. they will once more inform all their neighbours by multicasting a RERR incorporating the paths concerned. therefore efficaciously ?ooding information about a nexus breakage through the MANET.
DYMO was besides designed with possible hereafter sweetenings in head. It uses a generic MANET package and message format and offers ways of covering with unsupported elements in a reasonable manner. Speed samples of fake autos at different traf?c densenesss extremely unstable. A larger figure of fake vehicles forced autos and trucks into a stop-and-go gesture. cut downing the cars’ top velocity to that of trucks. This stabilised links between vehicles and decreased velocity differences between vehicles and roadside substructure. but caused big oscillations of local node densenesss.
Realistic communicating forms of MANET nodes were modeled utilizing OMNeT++ 3. 2p1. a simulation environment free for non-commercial usage. and its INET Framework 20060330. a set of simulation faculties released under the GPL. OMNeT++ runs distinct. event-based simulations of pass oning nodes on a broad assortment of platforms and is acquiring progressively popular in the communications community. Scenarios in OMNeT++ are represented by a hierarchy of reclaimable faculties written in C++ . Their relationships and communicating links are stored as Network Description ( NED ) ?les and can be modeled diagrammatically. Simulations are either tally interactively in a graphical environment or executed as command-line applications. The INET Framework provides a set of OMNeT++ faculties that represent assorted beds of the Internet protocol suite. e. g. the TCP. UDP. IPv4 and ARP protocols. It besides provides faculties that allow the mold of spacial dealingss of nomadic nodes and IEEE 802. 11 transmittals between them. IV. S IMULATION M ODEL The DYMO routing protocol was implemented as an application-layer faculty of the INET Framework faculty set. Following the speci?cation [ 2 ] . it employs UDP to pass on with other cases of DYMO. Additionally. it uses two assistant faculties to back up DYMO operation on the web bed. The complete protocol stack is shown in Figure 3. The ?rst assistant faculty is able to line up outbound packages before routing in the web bed occurs. so that a path can be set up by DYMO. The waiting line can so be signaled to let go of buffered packages for a given destination—either in order to hold them routed to the ?rst hop or to hold them discarded by the web bed because no path could be found. The 2nd assistant faculty is installed as a hooking map in
DYMO and support faculties in the protocol stack the inbound package way. It noti?es DYMO of the reaching of packages. This manner. routing table entries can be refreshed and route mistakes can be sent. severally. DYMO and its assistant faculties are assembled together with assorted constituents of the INET Framework to organize fake MANET nodes. Mobile nodes are represented by faculties of type Car. which perform DYMO along with TCP or UDP applications that generate application speci?c traf?c. Communication with other nodes takes topographic point via an IEEE 802. 11 faculty. The roadside substructure is provided by faculties of type AccessPoint. which execute DYMO merely to route between the radio and the wired web. i. e. the Internet. Internet connectivity is modeled by a node of type CSTMGateway that is besides running DYMO. It sends back delayed response messages to petitions
via TCP or UDP. i. e. it simulates the application waiters that are used by the clients ( the Cars ) .
For all communications. the complete web stack. including ARP. was used and wireless faculties were con?gured to closely resemble IEEE 802. 11b web cards conveying at 11 Mbit/s with RTS/CTS disabled. The TCP protocol execution follows the TCP Reno speci?cation. Therefore. consequences can be readily compared with bing Linux executions of DYMO. e. g. NIST DYMO or DYMOUM. For the simulation of wireless moving ridge extension. a apparent free-space theoretical account was employed and the transmittal ranges of all nodes adjusted to a ?xed value of 180 m. a tradeoff between changing realworld measurings described in related work [ 17 ] . [ 18 ] . All simulation parametric quantities used to parameterize the faculties of the INET Framework are summarized in Table I.
In order to guarantee realistic application bed traf?c. the undermentioned three different communicating scenarios were modeled: 1 ) Vehicles polled traf?c information from an Internet host. At 5 minute intervals. get downing at a random point in clip no more than 5 proceedingss from the start of a simulation. a vehicle tried to direct a 256 Byte UDP package to the gateway. which. upon response of the package. answered with a 1024 Byte response package. 2 ) Mobile nodes checked a POP3 letter box ( utilizing TCP ) for new messages. con?gured with a maximal section size of 1024 Byte and an advertised window size of 14 336 Byte. to direct eight 16 Byte bids. each triping a 32 Byte response. As in the ?rst instance. the letter box cheque was repeated 5 proceedingss after directing the Fake MANET scenario ?rst bid and the maximal session length limited consequently. 3 ) Vehicles requested RSS provenders from a web waiter ( besides utilizing TCP ) . This was represented by altering the 2nd case’s parametric quantities. so that nodes would merely direct a individual. 256 Byte petition message and have a individual. 65 536 Byte response message. with atomization and reassembly taking topographic point in lower beds. The sculptural nodes were so farther combined to make the MANET scenario shown in Figure 4. a fake main road with two lanes in each way organizing a 10 kilometer long closed ring with equally separated entree points at distances of 2 kilometers. 5 kilometer or 10 kilometer. depending on the scenario.
V. P ERFORMANCE A NALYSIS
Perceived public presentation of the VANET was estimated by entering the overall success rate. i. e. the chance of
successful response of a UDP information package. the last POP3 response. or a complete RSS provender. at the requesting
vehicle depending on the traf?c form in usage. Performance was measured at four different node densenesss of 0. 42. 2. 8. 4. 2. and 28 vehicles per kilometre and lane. matching to the two chosen traf?c densenesss and fractions of 10 % and 100 % DYMO-equipped vehicles. severally. Three different minimal path life-times of 1 s. 3 s. and 10 s were tried for each of the fake scenarios. A value of 1 s proved to equilibrate the sum of path petition and path mistake messages in the web best. Besides. puting the DYMO Network Size parametric quantity to 50 hops alternatively of to the default 10 hops proved to be bene?cial when entree points were spaced more than 2 km apart and node densenesss did non transcend 28 vehicles per kilometre and lane. All consequences are shown as boxplots. For each information set. a box
is drawn from the ?rst quartile to the 3rd quartile. and the median is marked with a thick line. Extra beards extend from the borders of the box towards the lower limit and upper limit of the information set. but no further than 1. 5 times the interquartile scope. Data points outside the scope of box and beards are considered outliers and drawn individually.
Figure 5 shows the overall chance of a fake UDP session being successfully completed for different node densenesss and entree point distances. As can be seen. even low node densenesss of 4. 2 nodes per kilometre and lane. every bit good as thin entree point deployment of one node per 5 km main road. suf?ced to allow the exchange of UDP packages in approx. 50 % of all attempts. Consequences for other communicating scenarios are shown in Figure 6. which plots the overall chance of a session being successfully completed for a ?xed entree point distance of 5 kilometer. Due to the retry mechanisms offered by the TCP protocol. POP3 Sessionss ever had a signi?cantly higher opportunity of being completed than obviously UDP Sessionss. even though completion of a POP3 session required the exchange of more packages. Besides seeable is a quickly diminishing chance of Sessionss being successfully completed when node densenesss increased to above 4. 2 nodes per kilometre and lane or when larger messages were to be delivered. While at 0. 42 nodes per kilometre and lane. the chance of RSS Sessionss finishing was about at par with that of POP3 Sessionss. at 2. 8 nodes per kilometre and lane already merely half as many RSS Sessionss ?nished successfully—approx. 20 % compared to approx. 40 % POP3 Sessionss. Figures 7 and 8 illustrate a ground for this lessening. With a lifting figure of pass oning nodes. web traf?c on the shared medium was progressively dedicated to DYMO packages until. at 28. 0 vehicles per kilometre and lane. the MANET was about entirely busy interchanging routing messages. Reducing the figure of actively take parting nodes to 10 % signi?cantly improved ?gures—even for node densenesss every bit low as 2. 8 vehicles per kilometre and lane. To gauge the impact of overload effects on the quality
of paths established by DYMO in the VANET. the relation between the length of a path in figure of hops and the
entire distance bridged between vehicle and entree point was examined. As can be seen in Figure 9. the bridged distance is closely related to the figure of hops and it is increasing linearly by approx. 150 m per hop—not much less than the nodes’ communicating scope of 180 m. In order to cut down the emphasis imposed on the web due to changeless nexus breakages and subsequent ?ooding of path mistake and new path petition messages. a promising mechanism was implemented for gauging the possible path stableness by taking motion waies into history. When comparing two paths to ?nd the shortest way. DYMO now added a Malus of 0. 1–5. 0 hops for each clip a package was sent to a vehicle going in the opposite way. Information about a vehicle’s comparative travel way was assumed to be estimable by the physical bed. Figure 10 shows the consequences of this version. Success rates of Sessionss could so be signi?cantly improved by adding such a Malus. but the version failed to bring forth the immense effects observed by other groups when wholly disregarding oncoming traf?c
for path choice.
VI. C ONCLUSION AND F UTURE WORK
Evaluation of the feasibleness and the expected quality of VANETs operated with the routing protocol DYMO showed
that for little sums of warhead informations to be transported. ad hoc webs of vehicles and inactive main road substructure can be successfully setup. maintained. and used with wellknown protocols from the Internet protocol suite entirely. Even low node densenesss and thin entree point deployment suf?ced to back up everyday polling of information via an Internet gateway. e. g. the checking of a POP3 letter box. Larger sums of web traf?c to be transported over the ad hoc web. nevertheless. induced overload effects that perceptibly destabilized the VANET. Particularly at higher node densenesss. which normally occurred in micro-jams. the routing and conveyance protocol behaviour led to a drastic addition in web burden. When the web became congested and new connexions could non be established. simple retry mechanisms merely furthered congestion. Simulation consequences hence seem to promote an version of the protocols in usage. so jobs perceived by lower beds are reacted to in a reasonable manner and application demands are taken into history when the web becomes overloaded. Cross-layer optimisation might maintain nodes from utilizing potentially unstable paths for low-priority messages in favour of a decrease of web burden. Besides. the consequences of the conducted simulations make a simple ?ooding of messages through the VANET and the choice of paths without taking node place and mobility into history. as proposed in the current bill of exchange of DYMO. look uneconomical. An experimental modi?cation of DYMO. which penalized paths across the lanes when measuring the quality of possible paths. proved bene?cial. but failed to bring forth the predicted addition in overall web quality that was claimed in related work.
[ 1 ] L. Wischhof. A. Ebner. H. Rohling. M. Lott. and R. Halfmann. “SOTIS – a self-organizing traf?c information system. ” in Proceedings of the 57th IEEE Vehicular Technology Conference ( VTC 03 Spring ) . 2003. [ 2 ] I. Chakeres and C. Perkins. “Dynamic MANET On-Demand
( DYMO ) Routing. ” Internet-Draft. draft-ietf-manet-dymo-06. txt. October 2006. [ Online ] . Available: hypertext transfer protocol: //tools. ietf. org/wg/manet/ draft-ietf-manet-dymo/draft-ietf-manet-dymo-06. txt
[ 3 ] C. Perkins. E. Belding-Royer. and S. Das. “Ad hoc On-Demand Distance Vector ( AODV ) Routing. ” RFC 3561. July 2003. [ Online ] . Available: hypertext transfer protocol: //www. ietf. org/rfc/rfc3561. txt
[ 4 ] C. Perkins and E. Royer. “Ad hoc On-Demand Distance Vector Routing. ” in 2nd IEEE Workshop on Mobile Calculating Systems and Applications. New Orleans. LA. February 1999. pp. 90–100.
[ 5 ] R. Baumann. “Vehicular ad hoc webs ( VANET ) – technology and simulation of nomadic ad hoc routing protocols for VANET on main roads and in metropoliss. ” Master’s thesis. ETH Z? rich. 2004.
[ 6 ] C. Sommer. I. Dietrich. and F. Dressler. “Realistic Simulation of Network Protocols in VANET Scenarios. ” in 26th Annual IEEE Conference on Computer Communications ( IEEE INFOCOM 2007 ) : Mobile Networking for Vehicular Environments ( MOVE 2007 ) . Poster Session. Anchorage. Alaska. USA: IEEE. May 2007.
[ 7 ] T. Clausen. C. Dearlove. J. Dean. the OLSRv2 Design Team. and the MANET Working Group. “MANET Neighborhood Discovery Protocol ( NHDP ) . ” Internet-Draft. draft-ietf-manet-nhdp-00. txt. June 2006. [ Online ] . Available: hypertext transfer protocol: //tools. ietf. org/pdf/draft-ietf-manet-nhdp-00. pdf
[ 8 ] T. H. Clausen. C. M. Dearlove. J. W. Dean. and C. Adjih. “Generalized MANET Packet/Message Format. ” Internet-Draft.
draft-ietf-manetpacketbb-02. txt. July 2006. [ Online ] . Available: hypertext transfer protocol: //tools. ietf. org/pdf/ draft-ietf-manet-packetbb-02. pdf
[ 9 ] A. Mahajan. “Urban mobility theoretical accounts for vehicular ad hoc webs. ” Master’s thesis. Department of Computer Science. Florida State University. 2006. [ 10 ] A. K. Saha and D. B. Johnson. “Modeling mobility for vehicular adhoc webs. ” in Proceedings of the 1st ACM international workshop on Vehicular ad hoc webs. 2004.
[ 11 ] C. Lochert. A. Barthels. A. Cervantes. M. Mauve. and M. Caliskan. “Multiple simulator complecting environment for IVC. ” in Proceedings of the 2nd ACM international workshop on Vehicular ad hoc webs. 2005.
[ 12 ] M. Treiber. A. Hennecke. and D. Helbing. “Congested traf?c provinces in empirical observations and microscopic simulations. ” Physical Review E. vol. 62. p. 1805. 2000.
[ 13 ] M. Treiber and D. Helbing. “Realistische Mikrosimulation von Stra?enverkehr Massachusetts Institute of Technology einem einfachen Modell. ” in ASIM 2002. Tagungsband 16. Symposium Simulationstechnik. 2002.
[ 14 ] O. Kaumann. K. Froese. R. Chrobok. J. Wahle. L. Neubert. and M. Schreckenberg. “On-line simulation of the expressway web of north rhine-westphalia. ” in Traf?c and Granular Flow ’99. D. Helbing. H. Herrmann. M. Schreckenberg. and D. Wolf. Eds.
Springer. 2000. pp. 351–356.
[ 15 ] B. Tilch and D. Helbing. “Evaluation of individual vehicle informations in dependance of the vehicle-type. lane. and site. ” in Traf?c and Granular Flow ’99. D. Helbing. H. Herrmann. M. Schreckenberg. and D. Wolf. Eds. Heidelberg: Springer. 2000.
[ 16 ] A. Varga. “The OMNeT++ distinct event simulation system. ” in Proceedings of the European Simulation Multiconference ( ESM2001 ) . 2001. [ 17 ] F. Hui and P. Mohapatra. “Experimental word picture of multi-hop communications in vehicular ad hoc web. ” in Proceedings of the 2nd ACM international
workshop on Vehicular ad hoc webs. 2005. [ 18 ] H. Wu. M. Palekar. R. Fujimoto. J. Lee. J. Ko. R. Guensler. and M. Hunter. “Vehicular webs in urban transit systems. ” in Proceedings of the 2005 national conference on Digital authorities research. 2005.