Abstract—
One major goal of the vehicular ad hoc network(VANET) is to improve driving safety. However, theVANET may not guarantee timely detection of dangerous roadconditions or maintain communication connectivity when thenetwork density is low (e.g., in rural highways), which maypose as a big threat to driving safety. Towards addressingthe problem, we propose to integrate the VANET with theinexpensive wireless sensor network (WSN). That is, sensornodes are deployed along the roadside to sense road conditions,and to buffer and deliver information about dangerous conditionsto vehicles regardless of the density or connectivity of theVANET. Along with the concept of VANET-WSN integration,new challenges arise and should be addressed. In this paper, weinvestigate these challenges and propose schemes for effectiveand efficient vehicle-sensor and sensor-sensor interactions. Prototypeof the designed system has been implemented and testedin the field. Extensive simulations have also been conductedto evaluate the designed schemes. The results demonstratevarious design tradeoffs, and indicate that satisfactory safetyand energy efficiency can be achieved simultaneously whensystem parameters are appropriately chosen.
Keywords-Vehicular Ad Hoc Networks, Wireless Sensor Networks,IntegrationI.
INTRODUCTION
Driving is an indispensable part of the life of manypeople. The past years have witnessed substantial efforts onimproving driving safety. Among them, the most prominenttechnological one might be the emerging vehicular ad hocnetwork (VANET) and the safe driving-targeted applicationsbuilt atop the VANET. The VANET is composed of highlymobilevehicles and sparsely-deployed roadside stations,each equipped with wireless communication devices andoptionally with sensing devices. Wireless communicationcan be conducted between vehicles and/or between vehiclesand roadside stations. On top of the VANET, applicationshave been developed to collect, process, share and deliverreal-time information about road conditions.These systems sometimes help in accident prevention,but they are not always effective since the underlyingVANET does not provide guaranteed real-time detectionof road conditions or communication connectivity. Firstly,the VANET only opportunistically monitors road conditions.That is, only when there exists a vehicle or a roadsidestation detecting or being notified of some conditions, canthe information be shared within the VANET. Secondly, theVANET can be disconnected due to high mobility and unpredictablemovements of vehicles and the sparse deploymentof roadside stations. If the VANET is disconnected, criticalinformation about road conditions known by one partitionof the VANET cannot be shared timely with vehicles thatneed to know it but are in other partitions.(a) (b)Figure 1. Examples for VANET-WSN IntegrationDeploying more roadside stations appears to be a solution.This, however, may significantly increase the investmentcost; also, lack of power supply is a big obstacle to do soin rural areas. Wide area wireless networks (such as cellularnetworks) could be used to connect disconnected segmentsof the VANET. This approach may achieve communicationconnectivity, but it does not solve the problem of lackingguaranteed real-time sensing of road conditions.Towards addressing the above problems, we propose tointegrate the VANET with the WSN to provide timely detectionof road conditions and to help connect partitioned segmentsof the VANET. Wireless sensor nodes, for example,MicaZ motes [1], are much cheaper than roadside stations.Besides, some inexpensive and low-power sensing modules,for example, the WiEye passive infrared sensors [2], havebeen commercialized and can be installed on the motes tosense road conditions with low cost.These sensor nodes can be deployed along roadside [3]–[5] with higher density than current roadside stations toform a connected network together with the VANET. Thesensor nodes can sense the road conditions, collect andprocess the sensing data to find out information usefulfor safe driving, and deliver the information to vehicles that need it. The sensor nodes also can buffer the safetyrelatedinformation generated by vehicles, and forward theinformation to vehicles in different partitions of the VANET.Following are some examples showing that deployingWSNs can greatly help in preventing road accidents:_ Example I. Deploying WSN along rural roads can helpprevent vehicle-animal collision accidents. As shown inFig. 1 (a), the WSN nodes can detect a deer roamingon the road and propagates the information within thenearby area. Approaching vehicles will get the warningbeforehand. The advantage brought by the deploymentof WSN is significant. It may help to avoid 1.5 millionvehicle-deer collisions happening every year (according toauto insurer State Farm) which result in about 150 deathsand $1.1 billion losses [6]._ Example II. Fig. 1 (b) shows that, bad road conditions(e.g., slippery surface) detected by an isolated vehicle canbe told to nearby roadside WSN nodes, and the WSNnodes can then collaborate with each other to propagatethe information to other vehicles approaching this dangerousarea. Note that, this cannot be accomplished if onlyVANET can be used since the VANET is not connected.To realize the proposed VANET-WSN system, severalimportant issues should be investigated. Firstly, the systemshould be viable in the real scenarios. The impacts of interference,noises and other environmental factors on systemperformance should be investigated. Secondly, the systemshould be scalable, considering the large scale of highwaysystem in the world. As the scale of deployment increases,the difficulty in deploying and maintaining the system shouldnot increase much, and the quality of service and the energyefficiency of the system should remain stable. Thirdly, thesystem should be flexible to changes in the real world. WSNnodes may fail or lose time synchronization, the highwaysmay be extended or reshaped, and traffic pattern may changefrom time to time. It is desired that the deployment andthe working parameters of VANET-WSN system can beadjusted with low overhead as the above changes happen.Fourthly, energy efficiency should be maximized for theroadside WSN. Although WSN nodes can be deployed andredeployed by humans and their batteries can be replacedmanually when necessary, it is still important to minimizethe energy consumption and maximize the network life timeto reduce energy and maintenance costs. Finally, satisfactoryquality of service should be attained. Dangerous road conditionsshould be detected and the information about thedangers should be delivered to related vehicles in a timelyfashion to ensure driving safety.Towards tacking the above issues, this paper makes thefollowing major contributions:_ We adopt the idea of group-based modular design toachieve scalability and flexibility. In our design, the roadsideWSN is made up of sensor groups. Each groupworks autonomously and asynchronously, and neighboringgroups interact with each other through a gatewaynode shared by them. Deployment or redeployment ofa group does not affect others; topology and workingparameter adjustments conducted within each group donot affect others, either._ The objectives of energy efficiency and quality of serviceare achieved by (i) an event-driven duty cycle schedulingstrategy which leverages the VANET to minimize energyconsumption in the WSN, and (ii) low-contentionand low-delay communication protocols which ensurecontention-less intra-group transmission and can reduceinter-group contentions with certain coordination costs._ A prototype of our system has been implemented andtested in the field to study the viability of the system.Based on realistic vehicle traffic traces and roadsidesensor-to-sensor communication traces, extensive simulationshave also been conducted. The results demonstratevarious design tradeoffs, and indicate that desired qualityof service and energy efficiency can be achieved simultaneouslywith appropriately chosen system parameters.To the best of our knowledge, this is the first work thatproposes, implements and evaluates an integrated VANETWSNsystem for driving safety.In the rest of the paper, Section II presents an overviewof our proposed system, which is followed by the designdetails in Section III. Section IV and Section V report ourimplementation and simulation results. Finally, Section VIconcludes the paper.

Download full report
http://cs.iastate.edu/~qinhua/PerCom10.pdf