Research On Routing Protocol And Route-decision Model Of Vehicular Ad-hoc Network

These years, China’s automobile industry has been progressing with astonishing rate. Due to the strong promotion of the ’China’s 12 th five-year plan’ of the industry of Web of Things, Internet of Vehicles(IOV)has been booming and become the support and direction of future Intelligent Transportation System(ITS). The technical breakthrough and the application of the Internet of Vehicles will effectively mitigate the pressure from environment and resources, thu s improve the efficiency of the resource. VENET is the foundation of the IOV, and the mobile model, routing protocol, routing decision and the model recommendation has become the core technology of VENET.Firstly, VANET has fast node movement, and the movement mode is constrained by the road. The mobile model is the core and the basis of the IOV. A dynamic path planning algorithm based on Dijkstra algorithm has been proposed in this paper, which combines the map module, transport module, vehicle motion control module. 1) Based on the road section and intersection, the map module could well simulate the traffic environment; 2) The mobility control could be achieved through communication among the nodes. And the road condition and the map information can be obtained using the real-time information of surrounding nodes, which makes the path decision more accurate; 3) Using the dynamic path planning algorithm based on Dijkstra to choose optimal path, realize the recommend efficient path selection. Quantitive research on the model proposed has been performed from both micro and macro view, in unobstructed, crowd and blocking scenarios. Experiments show that Compared with FTM, GBMM, IDM_IM and IDM_LC, the mobile model has better connection duration, linking stabilit y and protocol performance. Compared with random turning and the shortest path model, this model can make vehicle reach the destination faster and safely in different traffic conditions.Secondly, as the future intelligent transportation infrastructure, VANET constitute a unified wireless communication network by communication between car and car, car and roadside node, which is used to transfer real-time information of the auxiliary driving and avoiding accidents, providing convenience of safe driving, which is critical for real-time. First, the performance of three MANET classic routing protocol: AODV, DSDV and DSR has been studied quantitively in the environment of IOV. Analysis result indicates that low PTR, high NRL and large AEL are the critical factor that will impact the real-timing of the VANET routing protocol. Taking this into account, on-demand routing protocol DT-AODV based on the delay tolerant network has been proposed. At the same time, the VANET is modeled as a directed multigraph model to simulate the time variability. Experiments show that compared with PRAODV and PRAODV-M which are based on delay tolerant network protocol, DT-AODV has better effect in packet delivery ratio and average end-to-end delay. Therefore, DT-AODV is more adaptable to VANET.In addition, when obtaining the basic data through the collaboration of corresponding mobility models and routing protocols in VANET, data mining on the massive data generated by VANET becomes the future VANET direction. Route optimization is one of the important information services of VANET, travel time estimation method is an essential basis for route optimization, and real-time performance, reliability and high precision are the basic elements of prediction models. Firstly, in order to get more accurate vehicle speed at corresponding time period in the region, massive real taxi GPS historical data has been clustered in accordance with regional and density. Then, vehicle travel time estimation model has been proposed based on artificial neural network, where the reciprocal reverse of the vehicle speed has been taken as path weights to measure crowded conditions in the road sections and design the improved A-Dijkstra algorithm to achieve the recommendation service on most time-saving route. Experiments show that the model proposed in this paper is more accurate for the estimation of the travelling time, compared with the historical average time and the regression model based on the support vector.Finally, taxi probability at passengers’ current location and taxi waiting time prediction model in the VANET environment has been established for the issues of the difficult taxiing, unknown taxiing probability and waiting time. Combinin g with the road networks of the city, the PTM probability model of getting a taxi has been proposed according to data mining on the massive historical trajectory data from taxis’ GPS. A unique passenger waiting time prediction model PTWTM has been proposed according to data mining on the massive historical trajectory data. The maximum likelihood estimation has been used in this model, considering the arrival rate function as a piecewise linear function, thus the prediction of taxi probability and waiting time could be achieved. Experiments show that compared with the Computer Society of Beijing’s artificial statistical data, taxi probability prediction accuracy rate of the recommendation algorithm proposed in this paper in different time periods in weekdays and weekends is more than 90%, and accuracy rate of wait time prediction is close to 85%. The taxi probability and the wait time within 10 minutes can be calculated within 10 seconds using the recommendation algorithm. Prediction model proposed in this thesis is able to help the passengers choose the appropriate travel time, place and arrange travel plans.