Views
3 years ago

Geo-localized Virtual Infrastructure for VANETs - Sidi Mohammed ...

Geo-localized Virtual Infrastructure for VANETs - Sidi Mohammed ...

Among the various

Among the various choices that influence the design and theanalytical modelling of the GVI is the question related to where toposition the GVI in order to allow for a best-possible support ofVANETs. As we are dealing with the city environment, anintersection sounds suitable as geographic region because of its betterline-of-sight and also because it is a high traffic density area. Hence,the proposed GVI mechanism can periodically disseminate the datawithin a signalized (traffic lights) intersection area, controlled infixed-time and operated in a range of conditions extending fromunder-saturated to highly saturated. Thus, it can be used to keepinformation alive around specific geographical areas [6] (nearbyaccident warnings, traffic congestion, advertisements andannouncements, available parking lot at a parking place, etc.). It canalso be used as a solution for the infrastructure dependence problemof some existing dissemination protocols like UMB [5].The rest of this paper is organized as follows. Section II describesthe GVI scheme. Section III presents the derived analytical formulasthat provide the necessary guidelines for choosing the systemparameters. These are followed by a discussion of simulation andanalytical results in section IV. Finally, Section V concludes thepaper depicting some future research directions.II.GEO-LOCALIZED VIRTUAL INFRASTRUCTUREThe geo-localized virtual infrastructure mechanism consists onelecting vehicles that will perpetuate information broadcasting withinan intersection area. To do so, the GVI is composed of two phases: (i)selecting the vehicles that are able to reach the broadcast area (i.e. asmall area around the intersection center, where an electedvehicle could perform a local broadcast); then, (ii) among theselected vehicles, electing the local broadcaster which will perform alocal single-hop broadcast once it reaches the broadcast area (i.e. atthe intersection center).)LJXUH±6HOHFWLQJYHKLFOHVFDQGLGDWHVLQWKH*9,PHFKDQLVP$ 6HOHFWLQJFDQGLGDWHYHKLFOHVAmong the vehicles which are around the intersection, onlythose which are within the intersection region could participateto the local broadcast. They are selected as candidates if theyare able to reach the intersection center. The intersection regionis an area around the intersection starting at TR/2 m before andextending to TR/2 m beyond the intersection where TR is thetransmission range. Figure 1 illustrates the candidate vehiclesselection where vehicles {A, B, C, D, E, F, G, H} couldparticipate to the GVI mechanism since they are located withinthe intersection region and only vehicles {A, B, D, F} areselected as candidates because they are moving towards thebroadcast area.% (OHFWLQJWKHORFDOEURDGFDVWHUEach vehicle selected as candidate vehicle starts by computing theWLPHSHULRG QHHGHGWRUHDFKWKHintersection center by consideringits geographical location, direction and speed. According to this timeperiod, it computes a weight 3 . This one has to be minimal whenthe expected delay matches the desirable broadcast cycle time 7 ofthe GVI and it increases when we are far from 7. One possiblefunction for computing the weight is given by (1) (V is a constant)but other functions (e.g. triangle) can also be considered.3 ) = . 23 u1 V exp ( -1/2 ( 7 V' )²) (1)After the weight calculation, a waiting time :7(3) = MaxW (1 -3 3 max will be assigned to each candidate vehicle. Thecandidate vehicle with the highest weight 3 will have theshortest waiting time :7 to broadcast a short informativemessage telling other candidate vehicles that it has been electedas the local broadcaster.One may also note that the probability of having a collisionbetween two informative messages is weak. This is due to tworeasons, the length of these messages and the number ofvehicles that may compute similar weight 3 In the unlikelyevent of a collision among two broadcasted messages, the GVIwill have multiple elected nodes which will perform the localbroadcast while arriving at the intersection center instead ofone. So, such collisions will not break the GVI (i.e. nodramatic effect).The reason to choose the intersection region starting at TR/2m before the intersection is that the elected vehicle has toinform the other candidate vehicles. In the worst case, theelected vehicle is TR/2 away from the intersection and it cancover the points up to TR/2 away at the other side of theintersection. An example of vehicle election is illustrated inFigure 2: Candidate nodes, vehicles A, B & C compute thetime period to reach I (the intersection center) consideringtheir position, direction and speed. B will have a long timeSHULRG B since it is stopped at the traffic light. C has a veryVKRUW WLPH SHULRG C since it is very close to I. A requires aWLPH SHULRG A very close to the broadcast cycle time 7.Consequently, A will have the highest value of 3 and theshortest :7(3). A will be the first to send a message tovehicles B and C informing them that it has been elected toperform the local broadcast once it reaches the broadcast areaaround the intersection center I. Once vehicles within thetransmission range of the elected vehicle receive thebroadcasted message, they will participate in the election of thenext local broadcaster.Note that the elected vehicle has always the closest timeduration to 7. Hence, we can ensure that our GVI will perform

a periodic local broadcast. To avoid a too high variabilitybetween two successive broadcast messages, we define amargin 0, such as an elected vehicle may have an estimatedtime period to reach the center within theinterval > 7 0 ; 7 0 @.,)LJXUH±(OHFWLQJWKHORFDOEURDGFDVWHULQWKH*9,PHFKDQLVPA critical question that arises is how to choose the broadcast cycletime in order to achieve a good trade-off between the probabilityto inform a vehicle (that is a measure of quality of service) andthe number of copies of the same message received by avehicle (that is a measure of cost to provide the service). Thisis closely related to the time spent by a vehicle to go throughthe intersection area. Furthermore, the GVI is emulated byvehicles residing in the intersection area: a vehicle that enters theintersection region of a GVI attempts to participate in the mechanism;a vehicle that leaves the geographic region ceases to emulate the GVI.So, the other question concerns the risk that the GVI breaks,especially when the vehicular traffic density within theintersection area is very low. In other words, what is theprobability to fail during the election of the next localbroadcaster? In the following section, we present an analyticalmodel allowing the study of these issues.III. ANALYZING THE GVI PERFORMANCEIn this section, we present an analytical model to study theGVI mechanism. First, we want to determine the disseminationcapacity of our mechanism and thus, it is necessary to estimatethe delivery ratio and the mean number of reception of thebroadcasted message. When considering the broadcast cycletime 7, the optimal value of such parameter has to be chosen sothat all the vehicles receive once the broadcasted message.Consequently, the periodicity of the mechanism is closelyrelated to the sojourn time of the vehicles within the receptionarea (radius TR). Intuitively, this parameter has to be of thesame magnitude order than the minimum sojourn time of avehicle within this area. Accordingly, this sojourn timecorresponds to the travel time a vehicle would haveexperienced in the absence of traffic signal control (i.e. trafficlights).Furthermore, the GVI mechanism is based on inter-vehiclead-hoc communications. Consequently, when a message isbroadcasted, it is necessary that at least one vehicle is stillwithin the intersection area (radius TR/2) and located beforethe traffic signal so that it could rebroadcast the message lateron. The impact of the margin has also been considered. Thesecond performance criterion is thus the probability S that novehicle can be elected within the considered interval. Note thatthis situation does not mean that the mechanism fails. Indeed,the vehicles within the reception area (radius between TR/2 andTR) may broadcast later on the message.$ 1RWDWLRQDQGV\VWHPDVVXPSWLRQVThe sojourn time at signalised intersections constitute a verysignificant part in the GVI mechanism since it has an importantimpact on the system parameters and performance. It can berepresented by two parts: the on-move sojourn time and thequeuing sojourn time. Note that many models have alreadybeen proposed for the queue length and the delay analysis attraffic signals [7][8][9][10], but none of them answers ourneeds. This is what motivated the following model.In order to develop our model, the following assumptionswill be adopted:- The arrival process of vehicles constitutes a Poisson processwith parameter O.- Without lake of generality (i.e. it has no impact on themodel), the possibility to turn left or right is not considered.- The queuing sojourn time : is computed as the differencebetween the travel time actually experienced by a vehicle whilegoing across the intersection and the travel time this vehiclewould have experienced in the absence of traffic signal control.- The moving speed of a vehicle is constant. Therefore, the onmovesojourn time can be represented by the sum of:* -: the required time to join the queue* /: the necessary time to leave the intersection.* &: the time needed to cross the intersection.The sojourn time of a vehicle within the reception area 6 is thusequal to: 6 - : & /- The amber period is modeled as follows [8], when a vehiclewants to enter the intersection, if the residual green period islower than the necessary time & to cross the junction, thevehicle stops and wait for the following green period.Table 1 gives the parameters which have been considered:7DEOHParameters of the analysis6&(1$5,2Green, amber, red Intervals (7J7D7U) (38s,2s,40s)Capacity of the intersection = 1/& 0.5 veh/sTransmission range200 mOn-move sojourn time (J,L,C)(10s,10s,2s)Broadcast time cycle = 740sCycle Duration 7F=7J+7D+7U80sVehicle velocity (city)30 km/h

Virtualized IT Infrastructures and Their Management
VIRTUAL DESKTOP INFRASTRUCTURE - Getronics
Creating a Dynamic Infrastructure through Virtualization
SnapManager for Virtual Infrastructure - VMware Communities
HP virtual desktop infrastructure
The Virtualized Infrastructure Services Opportunity and Parallels ...
Creating a Dynamic Infrastructure through Virtualization
Scaling Virtual Infrastructure: Planning for Growth, High ... - VMware
The Impact of Virtualization on Enterprise Infrastructure - Profoss
Surviving Regulatory Compliance in the Virtual Infrastructure - VMware
Reliability Support in Virtual Infrastructures - salsahpc
Toward Measurement and Analysis of Virtualized Infrastructure - Caida
The Virtualized Infrastructure Services Opportunity and Parallels ...
Virtualizing IT infrastructure in all directions - enterpriseinnovation.net
Building a virtualised environment Virtual infrastructure ... - Softcat
Creating an agile infrastructure with Virtualized I/O - Informatix ...
Virtual Infrastructure 3: Beta to Production - VMware
Value of Virtual Infrastructure - HITS Nova Scotia
How do I ensure my virtual server infrastructure seamlessly ...
Vehicular Networks - Sidi Mohammed Senouci Home Page
Senouci_UBIROADS07_2.. - Sidi Mohammed Senouci Home Page
Vehicular Network Self-Organizing Architectures - Sidi Mohammed ...
Data Collection in Vehicular Networks - Sidi Mohammed Senouci ...
Ambience Project ³Find a Meeting´ Video - Sidi Mohammed Senouci
Data Collection in Vehicular Networks - Sidi Mohammed Senouci ...
A Panorama on Wireless Mesh Networks - Sidi Mohammed Senouci ...
Current Trends in Vehicular Ad Hoc Networks - Sidi Mohammed ...
Persistent Localized Broadcasting in VANETs
Multi-Hop-Aware Cooperative Relaying - Sidi Mohammed Senouci ...
Call Admission Control in Cellular Networks - Sidi Mohammed ...