Research On Urban VANETs Data Delivery And Intelligent Security Driving

With the rapid development of automobile technology, the number of car ownership grows as the days passed. As a result, the frequency of the traffic accident increases rapidly. According to statistics, there nearly 3400 people die each day and tens of millions of people are injured or disabled each year in traffic accidents, leading the economic loss of 500 billion US dollars. Under modern car manufacturing technology condition,the proportion of the traffic accident caused by car design is becoming smaller and smaller while the driver factor becomes the main cause of a traffic accident. To this end,intelligent transportation system(ITS) is proposed. It is aimed at applying advanced information technology, data communications technology, sensor technology and computer technology into the traffic system to enhance the communication among human, car and road of traffic system, thereby reducing the traffic accident. Compared with the passive security defense technology, ITS can proactively identify dangerous source that threats the road safety, and provide the drive with a warning to avoid traffic accidents. Thus,ITS more satisfies the demand of future development. Research on ITS has theoretical significance and engineering value.Currently, the key technology to achieve intelligent transportation system is the wireless communication technology and the intelligent sensing technology. Vehicular Ad Hoc Networks(VANETs) and smartphone sensing as the typical representative of the two technologies have drawn great attention from industry and academia. Based on a systematical summary of relevant works on vehicular ad hoc networks, this dissertation focuses on the data delivery of VANETs. In addition, based on a systematical analysis of the characteristic of smartphone, this dissertation also makes a deep research and exploration on the value of the smartphone application of intelligent transportation system.To solve these problems, several achievements are gained in this dissertation. The major contributions of this dissertation are as follows:1. The mode of urban VANETs data delivery and factors that affect network connectivity are analyzed, and a Traffic-Aware data Delivery Scheme(TADS) for urban VANETs is proposed. In TADS, it proposes an improved geographical greedy routing algorithm for the data delivery in the straightway mode. For the data delivery in the intersection mode, by calculation a utility function for the candidate road, TADS chooses the road with the maximum utility function value as the forwarding path. The utility function is consisted of three components which are the vehicular density, the variance of vehicular spatial distribution, and the Euclidean distance between the candidate intersection and the destination, respectively. The vehicular density and the variance of vehicular spatial distribution can determine the network connectivity. The Euclidean distance between the candidate intersection and the destination can determine the direction of the transmission of data. The larger value of utility function denotes the better of network connectivity and the closer between the candidate intersection and the destination. Furthermore, in order to reduce the network overhead caused by the information collected, according to the movement characteristic of the vehicle, TADS gives a traffic prediction model which largely reduces the information collection period. The simulation results show that TADS achieves a higher delivery ratio with lower routing overhead and average delivery delay.2. In view of VANETs data distribution, especially for the large content files, such as video and image, vehicular data distribution using COllaborative Urban Parking clusters scheme(COUP) is proposed. It mainly leverages lots of parking clusters as the natural infrastructures to maintain the data and provide data distribution services for moving users through the parking cluster collaborative. Specially, for each content downloading request, the parking cluster head first estimates the data amount that is provided for requester by the local parking cluster according to the principle of first come, first served.For the remained content, based on the requester’s historical travel records, the parking cluster head builds its travel prediction model and then selects a destination parking cluster to which the remained content will be transferred. When the requester passes through the parking cluster, it can continue the unfinished content downloading. Theoretical results verify the effectiveness of our approach and simulation results based on the real traffic data show that compared with the other two data distribution algorithms,COUP has a higher downloading rate, especially in sparse traffic and multiple requests conditions.3. Motivated by the fact that unsafe driving behaviors seriously threaten the traffic safety, a smartphone auto-calibration based dangerous driving behavior identification system(DrivingSense) is proposed. In DrivingSense, it uses the smartphone sensors to sense the parameters of vehicle moving, monitoring the moving states of vehicles. It can identify three dangerous driving behaviors: speeding, irregular driving direction change and abnormal speed control. To achieve this objective, DrivingSense first analyzes theoretically the impact of the sensor error on the vehicle driving behavior estimation. Secondly,it proposes the smartphone sensor error distribution determination method under vehicle moving condition and then an improved Kalman Filter based smartphone sensor error correction algorithm is proposed to obtain data that can reflect the vehicle driving state more accurate. Finally, it proposes vehicle driving state estimation method by using the corrected data. Specially, in order to estimate the vehicle speed, a novel speed estimation method is proposed based on the kinematic knowledge. In order to identify the irregular driving direction change, a two-stage vehicular turn signal detection method is proposed:(1) noise filter is proposed by analyzing the frequency of turn signal sound;(2)for the noise filtered audio, a sound cross-correlation algorithm is used to detect turn signal.In order to identify the abnormal speed control, a threshold based scheme is proposed.Based on the realistic environments experiment, the results show that DrivingSense can effectively identify the dangerous driving behaviors.4. Motivated by the fact that due to the sight of the driver is restricted during nighttime and traffic accidents are frequent, a smartphone based Driver Nighttime Assistance System(DNAS) is proposed. In DNAS, it mainly uses the smartphone camera to sense the driving state of vehicles that are behind the host vehicle, monitoring the dangerous vehicle with speeding or tailgating and providing a warning for drivers to obtain more reaction time to deal with the risk event. For the unique problem of the system implementation, this dissertation provides the corresponding solutions. To be specific, it first proposes the sensing distance of camera determination method. Secondly, based on the characteristic of the bright vehicle headlight and their geometric distance, a vehicle identification method for the nighttime road image is proposed. In addition, according to the temporal and spatial characteristic of moving vehicle, an identified vehicle tracking method is proposed, which can be used to estimate the speed of the monitored vehicle. Finally, based on the imaging principle, a relative distance estimation method is proposed.Based on the realistic environments experiment, the results show that DNAS can effectively detect the event of speeding and tailgating.