You are on page 1of 5

JOURNAL OF COMPUTING, VOLUME 2, ISSUE 9, SEPTEMBER 2010, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/ WWW.JOURNALOFCOMPUTING.

ORG

33

Efficient Data Scheduling in VANETs


Vishal Kumar, Narottam Chand
Abstract This paper primarily focuses on scheduling the requests in an optimize manner. We address some challenges in vehicle roadside data access. We propose a priority based scheduling scheme called D* S/P to consider both service deadline and data size along with priority when making scheduling decisions. These requests are queued in different combinations of priorities to apply the scheduling algorithm. Once the combinations are applied, the request with the highest priority is served first. Simulation results show that the scheduling scheme performs better in comparison to other scheduling schemes. Further, the scheduling scheme is adaptive to different workload scenarios. Index Terms Vehicular ad hoc networks (VANETs), RSU, scheduling, queue.

1 INTRODUCTION
ITH the increase in the number of vehicles the possibility of accidents has also increased. Therefore it is required to make such a system which could help in reducing the possibilities of accidents. A vehicular ad-hoc network (VANET) is an effort towards this which adds ability in the vehicles to communicate each other (V2V) or with the infrastructure (V2I). A lot of work has been done to address data dissemination issues in vehicular ad hoc networks (VANETs). Broadly there are two approaches when we consider V2I or I2V dissemination; push based and pull based approach. In the push based approach [4, 5, 19] the info-stations also called roadside units push out the data to everyone. This approach has its application in disseminating traffic alerts, weather alert, etc. The pull based approach [4, 8, 9] is based on requestresponse model as this is based on user specific data. Recently, vehicle-roadside data access has received considerable attention. In vehicle-roadside data access, the RSU can act as a router for vehicles to access the internet. Although this can bring many benefits to the drivers, the deployment cost and maintenance cost are very high. As another option, RSU can also be just used as a buffer point (or data island) between vehicles. In this thesis we focus on the scheduling of the data at these data centers or RSUs. In this paradigm, all data on the RSUs are uploaded or downloaded by vehicles. Many dierent and sometimes competing design goals have to be taken into account for VANETs to ensure their commercial success. When equipped with WAVE (Wireless Access in Vehicular Environment), forms a highly dynamic network. Due to this dynamic nature, the vehicles requests to be serviced create a time limitation. Our work aims at prioritizing and scheduling of the requests arriving at the data center in vehicular ad-hoc network.

Currently, most existing data scheduling techniques consider only two parameters data size and deadline involved in vehicle ad-hoc networks [14, 15]. Their main focus is on specific elements involved in the communication, such as medium access, vehicle mobility, traffic flow, routing, etc. The challenges for VANETs raise the demand for a more perfect data dissemination technique. The goal of this paper is to schedule the data efficiently by considering parameters like deadline, data size and priority of the request. To do this, the following problems are addressed: 1. First, it is essential to be thoroughly familiar with the characteristics of vehicular ad-hoc networks. Thus, the initial task is to analyze data dissemination techniques, routing and scheduling approaches used in vehicular networks. The second challenge is to consider the existing data scheduling techniques in vehicular ad-hoc network. It means to deal how to schedule the data in high mobile environment so as to serve as many requests as possible. Thus, we have to analyze existing data scheduling techniques of VANETs. Third challenge is to analyze the different factors which are considered in existing data scheduling techniques of VANETs. In this way determine the factors which should be included in existing techniques to schedule the data efficiently. Fourth challenge is to apply additional feature like priority to the existing scheme for better scheduling in VANETs. Then determine how much improvement has got after adding these additional features.

2.

3.

4.

Vishal Kumar is with the Department of Computer Science & Engineering, National Institute of Technology, Hamirpur, INDIA-177005. Dr. Narottam Chand is with the Department of Computer Science & Engineering, National Institute of Technology, Hamirpur, INDIA-177005.

JOURNAL OF COMPUTING, VOLUME 2, ISSUE 9, SEPTEMBER 2010, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/ WWW.JOURNALOFCOMPUTING.ORG

34

Based on the above mentioned challenges, we have pro posedthefollowing: A new scheme called D*S/P which considers data size, request deadline and priority of the requests to be served. Study the frequency of various messages based on the priority assigned to the messages. Simulation to study the performance of the proposed scheduling scheme. Compare the proposed scheme with existing scheme. Several mobile computing systems such as MobiEyes and CarTel [20] were designed and implemented by UCLA and MIT, respectively to collect process and deliver data from sensors located on vehicles to roadside infrastructures for analysis. The similar projects include the DieselNet at Umass [25], FleetNet in Germany [19], InternetCAR in Japan [21]. Vehicles are moving at a high speed and they stay in the area of roadside units only for a short period of time. Therefore scheduling of data is very crucial so as to serve as much requests as possible, reducing delay of downloading etc. Scheduling is an important issue for data access in vehicular environment. In the next section, we shall study the work related to the scheduling for vehicle-roadside data access.

ted QCs. This work is also based on point-to-point communication and does not take advantage of broadcasting. All these works mainly focus on responsiveness such as average/worst- case waiting time or fairness without considering the time constraints of the user requests. However, in vehicular networks, time constraint of the request has to be considered. Jiang and Vaidya [10], Rajan et al. [17] and Xu et al. [18] studied the scheduling problem in real-time broadcasting environment and took time constraint into account. The authors in [18] investigated online scheduling algorithms for time critical on- demand data broadcast. However, they ignored the data update issue. They assumed that data are read only or can only be updated by the server. Hence, they only tried to improve the service ratio for download broadcasting. In contrast, our vehicle-roadside data access model is different as both update and download compete for the same bandwidth. Also, missing the update degrades the data quality. The authors in [1] gave the rst understanding of the impact of the vehicles speed, transmission rate, 802.11 bitrate, and packet size on throughput and delay of vehicleroadside communication and illustrated a basic picture of how running vehicles contact with roadside hot spots through a drive-thru data access. In various earlier works deadline constraints of the requests were not considered. However, some works took into account deadline for some requests and found that the requests shall be dropped completely if vehicles move out from the RSU area. In this case if the request is urgent shall not be served and dropped. The authors in [15] proposed a basic scheduling scheme called D*S to consider both deadline and data size when making scheduling decisions. To make use of wireless broadcasting, the authors gave another scheduling scheme called D*S/N to serve multiple requests with a single broadcast. The authors also identified the effects of upload requests on data quality, and proposed a Twostep scheduling scheme to provide a balance between serving download and upload requests. They have compared the three naive schemes; FCFS, FDF and SDF and showed their performance and limitations. In [15] authors proposed a scheduling algorithm in which each file is chopped in several segments. If a vehicle shall not be able to finish downloading all segments from the RSU then the algorithm allows the vehicle to continue downloading the segments from the next RSU. The authors have also investigated the problems of scheduling of file distribution from roadside unit (RSU) to the vehicle in the urban environment. The work in [15] considers the parameters given in [16]. We consider the problem of providing or reducing the delay to those requests which needs to be served immediately. We build our strategy on previous work in this area [15, 16]. We are proposing our scheme that shall consider priority and serves the requests with priority first

2 RELATED WORK
Earlier a large amount of work related to CPU and job scheduling is done in the literature. Wong studied several scheduling algorithms such as rst-come-rst-serve (FCFS), longest wait time (LWT), most requests rst (MRF) in the broadcasting environments [9]. Later, many broadcast scheduling algorithms have been proposed to reduce the waiting time and energy consumption [17]. Acharya and Muthukrishnan [11] addressed the broadcast scheduling problem in heterogeneous environments, where data items have different sizes. The solution is based on a new metric called Stretch, dened as the ratio of the response time of a request to its service time. Based on stretch, they proposed a scheduling algorithm, called longest total stretch rst (LTSF) to optimize the stretch and achieve a balance between the worst case and the average case. However, a straightforward implementation of LTSF is not practical for a large system, as at each broadcast time, the server has to recalculate the total stretch for every data item with pending requests in order to decide which data to broadcast next, and hence the scheduling algorithm becomes a bottleneck due to the high computation overhead. The work in [12] introduced the concept of Quality Contracts (QCs) which combines the two incomparable performance metrics: response time or Quality of Service (QoS), and staleness or Quality of Data (QoD). QCs allow individual users to express their preferences for the expected QoS and QoD of their queries by assigning prot values. They proposed an adaptive algorithm to maximize the total prot from submit-

JOURNAL OF COMPUTING, VOLUME 2, ISSUE 9, SEPTEMBER 2010, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/ WWW.JOURNALOFCOMPUTING.ORG

35

otherwise the scheme behaves similar to the scheme proposed by [15]. In this paper, we have studied the application-layer scheduling for vehicle-roadside data access. We propose and implement our scheme at the application layer and call it D*S/P.

TABLE 1 POSSIBLE VALUES OF P


P 0 1 Meaning Normal Accidental Possibility of Emergency Warning/Alerts

3 PRIORITY BASED DATA SCHEDULING (D*S/P)


All vehicles can send requests to the RSU if they want to access the data. Each request is characterized by a 5-tuple: <v-id, d-id, op, deadline, r-priority>, where v-id is the identifier of the vehicle, d-id is the identifier of the requested data item, op is the operation that the vehicle wants to do (upload or download), deadline is the critical time constraint of the request, beyond which the service becomes useless and r-priority is the priority sent with the request according to the message type. 2.

2 3

The second list called S-List (dataSize-list) is used to record the size (S value) of the data item that is asked by the request. S-List is sorted in ascending order of the data size. The third list called P-list (Priority-list) is used to record the priority of the request.

3.

Algorithm Implementation of D*S/P: 1. 2. The requests arrive at the RSU. The tuples of the request are read which consist of parameters including priority. The P parameter is evaluated if its value lies between 0 P 3, then the request is specially treated on priority. Otherwise, the algorithm runs similar to the D*S scheduling scheme.

Vehicle request data

3.

Is 0P 3

4.

P-list

D-list

S-list

Intuitively,

Given two requests with the same priority, the one asking for a small size data with minimum deadline should be served rst.

Service Processing

We have used four values to the request priorities. A normal request shall have P=0. In other cases, P can be 1, 2 or 3. A accidental help request shall have P=1. A request with P=2 means a possibility of emergency. A P=3 means a requests for alerts or warnings. We shall try to find out the arrival frequency of different types of messages and try to queue them in different queues based on the priority and finally schedule the transmission to increase the service ratio.

Fig.1. Flowchart of D*S/P scheme

4 SIMULATION SETUP
This section describes the simulation of Priority Based data scheduling. The simulation has been done using simulator NCTUns 6.0 [2]. Each vehicle generates requests and sends to the roadside unit (RSU) with 5-tuples in the request. The proposed scheduling scheme is simulated on a 400 m*400 m square road scenario. One RSU server is put at the crossing roads of the area. It is also the intersection of one horizontal road and one vertical road, where road has two lanes as shown in Figure 2.

The proposed scheduling algorithm maintains the sorted list each for deadline (D), data size (S) and priority (P). These three lists are as follows: 1. One list called D-List (Deadline-list) is used to record the deadline (D value) of each request. DList is ordered by the increasing deadline of each request.

JOURNAL OF COMPUTING, VOLUME 2, ISSUE 9, SEPTEMBER 2010, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/ WWW.JOURNALOFCOMPUTING.ORG

36

TABLE 2 SIMULATION PARAMETERS Parameter


Simulation Time Vehicle-Vehicle Space Vehicle Velocity Wireless Coverage Packet Size Data Item size Fig.2. Thesimulationscenariolayout Routing Protocol Ratio Proportional Model Antenna Model Mac Type Traffic Type

Value
100s 10-20m 0 60 m/s 200m 1024 bytes 25 KB DSR Two Ray Ground Omni Antenna 802.11 CBR(UDP)

To simulate the vehicle traffic, we randomly deploy 40 vehicles initially. All vehicles move towards either end of the road. They are moving forth and back during the simulation to mimic the continuous traffic ow in the intersection area. When one vehicle reaches the end of the road, which means the vehicle will move out of the RSU area, its request not serviced will be dropped. Each vehicle issues service requests with a priority P, (0 P 3). When one vehicle is served or reaching the end of the road, it waits some time to issue a new request. The interarrival time of each request follows an exponential distribution with a mean of . Its density function is: f(t) = et The simulation parameters are listed in Table 2. All requests are queued at the RSU; served upon arrival. Based on the scheduling algorithm, the server serves one request and removes it from the request queue.

earlier work, and then select the smallest one with the highest priority based on the value in the P-list. Here also we consider the same data structure as in [16] to reduce the computation complexity to O(logm). A large number of vehicles retrieve (or upload) their data from (or to) the RSU when they are in the communication range. The RSU (server) maintains a service cycle, which is nonpreemptive; i.e., one service cannot be interrupted until it finishes. When one vehicle enters the RSU area, it listens to the wireless channel. All vehicles can send requests to the RSU if they want to access the data.

5 SERVICE RATIO
The service ratio evaluation is based on the possibility of as many requests that can be served. It can also be termed as throughput of the algorithm. We have analyzed the service ratio of both the scheduling algorithms. The graph on X-axis has the simulation time. The comparison is between D*S and D*S/P. Therefore, the result shows that the service ratio is improved when considering the priority of the requests. Figure 3 shows the comparison of service ratio of D*S and D*S/P scheduling algorithms. In this figure, we are comparing requests using priority and requests not using priority.

Fig.3. Service Ratio of D*S and D*S/P

7 CONCLUSION 6 SIMULATION RESULTS


A straightforward implementation of the D * S/P scheme is to compute the DS_values of all requests as done in This paper primarily focuses on scheduling the requests in an optimize manner. We addressed some challenges in vehicle roadside data access. We proposed a priority based

JOURNAL OF COMPUTING, VOLUME 2, ISSUE 9, SEPTEMBER 2010, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/ WWW.JOURNALOFCOMPUTING.ORG

37

scheduling scheme called D* S/P to consider both service deadline and data size along with priority when making scheduling decisions. These requests are queued in different combinations of priorities to apply the scheduling algorithm. Once the combinations are applied, the request with the highest priority is served first. Simulation results show that the scheduling scheme performs better in comparison to other scheduling schemes. Further, the scheduling scheme is adaptive to different workload scenarios.

[15] M. Shahverdy, M. Fathy, S. Yousefi, Scheduling Algorithm for Vehicle to Roadside Data Distribution, ICHCC-ICTMF 2009, China, Communications in Computer and Information Science (CCIS Book Series), Vol. 66, pp. 30-38, Springer Verlag Berlin Heidelberg, 2010. [16] Y. Zhang, J. Zhao, G. Cao, On Scheduling Vehicle-Roadside Data Access, in Proceedings of the fourth ACM international workshop on Vehicular ad hoc networks, ACM/ VANET07 pp. 918, 2007. [17] T. Nadeem, P. Shankar and L. Iftode A Comparative Study of Data Dissemination Models for VANETs,in Proceedings of the 3rdAnnualInternationalConferenceonMobileandUbiquitousSys tems(MOBIQUITOUS),SanJose,California,pp.110,July2006. [18] Kevin C. Lee, Uichin Lee, Mario Gerla, Survey of Routing Protocols in Vehicular Ad Hoc Networks, in Advances in Ve hicularAdHocNetworks:DevelopmentsandChallenges,IGIGlobal, Oct,2009. [19] The FleetNet Project, http://www.fleetnet.de. [20] The CarTel Project, http://cartel.csail.mit.edu/doku.php. [21] Internet ITS, http://www.internetits.org. [22] J. Ott and D. Kutscher, Drive-thru Internet: IEEE 802.11b for Automobile Users, in Proceedings of the INFOCOM04, pp. 3342, 2004. [23] Y. Zhang, J. Zhao, G. Cao, Service Scheduling of VehicleRoadside Data Access in ACM/Springer Journal on Mobile Networks and Applications, Vol. 15, No. 1, pp. 83-96, Feb. 2010. [24] D. Hadaller, S. Keshav, T.brecht, et al. Vehicular Opportunistic Communication Under the Microscope, in Proceedings of the 5th International Conference on Mobile Systems, Applications, and Services (MobySys07), pp. 206-219, 2007. [25] http://prisms.cs.umass.edu/diesel

ACKNOWLEDGMENT
The authors wish to thank Mohammad Bilal, Mobile and Satellite Communications Research Centre, School of Informatics, University of Bradford, BRADFORD, (UK) for his cooperation in conducting the simulation.

REFERENCES
[1] S.Y. Wang and C.L. Chou, NCTUns Tool for Wireless Vehicular Communication Network Researches, Simulation Modelling Practice and Theory, Vol. 17, No. 7, pp. 1211-1226, August 2009. http://nsl.csie.nctu.edu.tw/nctuns.html, 2010 S. Jiang, N. Vaidya, Scheduling Data Broadcast to im-patient Users, in Proceedings of MobiDE99, pp. 5259, 1999. J. Zhao and G. Cao, VADD: Vehicle Assisted Data Delivery in Vehicular Ad Hoc Networks, in Proceedings of IEEE INFOCOM06, pp. 112, 2006. S. Acharya, S. Muthukrishnan, Scheduling On-Demand Broadcasts: New Metrics and Algorithms, in Proceedings of MobiCom 98, pp. 43 54, 1998. J. Xu, X. Tang, W. Lee Time-critical on-demand Data Broadcast: Algorithms, Analysis, and Performance Evaluation, in IEEE Transactions on Parallel Distributed Systems, Vol. 17, No. 1, pp. 314, Jan. 2006. D. Rajan, A. Sabharwal, B. Aazhang, Power Efficient Broadcast Scheduling with Delay Deadlines, in Proceedings of the rst international conference on broadband networks (BROADNETS04), Washington, DC, pp. 439448, 2004. J. Zhao, Y. Zhang, G. Cao Data Pouring and Buffering on the Road: A New Data Dissemination Paradigm for Vehicular Ad Hoc Networks, in IEEE Transactions on Vehicular Technology , Vol 56, No. 6, pp. 32663277, Nov. 2007. Y. Wu and G. Cao, Stretch-optimal Scheduling for OnDemand Data Broadcasts, in Proceedings of the 10th International Conference on Computer Communications and Networks (ICCCN 01), pp. 500-504, 2001. D. Aksoy, M. Franklin, R*W: A Scheduling Approach for Large-scale On-demand Data Broadcast, in IEEE/ACM Transaction on Networking-99, pp. 846860, 1999. C. Suthaputchakun, Priority-Based Inter-Vehicle Communication for Highway Safety Messaging Using IEEE 802.11e, in International Journal of Vehicular Technology, vol. 2009. J. Mi, F. Liu, S. Xu and Qi Li, A Novel Queue Priority Algorithm for Real-Time Message in VANETs, in IEEE International Conference on Intelligent Computation Technology and Automation, pp. 919-923, 2008. W. Xiang, J. Gozalvez, Z. Niu, O. Altintas, and E. Ekici, Wireless Access in Vehicular Environments, EURASIP Journal on Wireless Communications and Networking, Vol. 2009, Article ID 576217, 2009. J. Blum, A. Eskandarian, L. Hoffman, Challenges of Intervehicle Ad Hoc Networks, in IEEE Transactions on Intelligent Transportation Systems, Vol. 5, No. 4, pp. 347351, 2004.

[2] [3] [4] [5] [6]

AUTHORS PROFILE Vishal Kumar is a research scholar (M.Tech) in Department of Computer Science & Engineering at National Institute of Technology, Hamirpur, India. He received his Bachelor of Engineering in Computer Science & Engineering from Kumaon Engineering College, Dwarahat (Almora), India in 2005. His area of interest is vehicular ad-hoc networks. He has published several papers in National/ International conferences. Dr. Narottam Chand is an Associate Professor & Head in Department of Computer Science & Engineering at National Institute of Technology, Hamirpur, India. He received his PhD degree from IIT Roorkee in Computer Science and Engineering. Previously he received M. Tech and B. Tech degrees in Computer Science and Engineering from IIT Delhi and NIT Hamirpur respectively. His current research areas of interest include mobile computing, mobile ad hoc networks and wireless sensor networks. He has published more than 50 research papers in International/National journals & conferences and guiding six PhDs in these areas. He is a member of ISTE, CSI, International Association of Engineers and Internet Society.

[7]

[8]

[9]

[10] [11]

[12]

[13]

[14]

You might also like