Dissertation > Excellent graduate degree dissertation topics show

Urban Vehicular Mobility Patterns for Driving Route Prediction

Author: LiZhongWei
Tutor: LiMingLu；XueGuangTao
School: Shanghai Jiaotong University
Course: Applied Computer Technology
Keywords: Car ad hoc networks Characteristics of motor behavior Variable-length multi - order Markov model Path prediction
CLC: TN92
Type: Master's thesis
Year: 2010
Downloads: 144
Quote: 4
Read: Download Dissertation

Abstract

The development of wireless communication technology has brought profound changes to people's work and life. Vehicle self-organizing network technology as an important branch of the wireless communication technology, will achieve security, comfort, and provide technical support for intelligent urban traffic environment. The complexity of the urban traffic environment to the car self-organizing network research has brought great difficulties. Traditional wireless network simulation and verification methods are difficult to apply to the car self-organizing network. How to build a car close to the real self-organizing network environment is an important research topic. This article will be collected based on the Shanghai grid project to the real driving data of more than 4,000 taxis motion feature on-board self-organizing network analysis, and board self-organizing network protocols and mechanisms to optimize the characteristics of these movements. In this paper, we study the following question: First, build the experimental platform on-board self-organizing network research. Accurate and complete data is the basis of the experimental analysis. Vehicle travel data collected based on the GPS device is discrete, inaccurate. These data can not be directly used in the experimental analysis, the need for some preprocessing. Analysis of GPS data to abnormal reasons exist to solve the problem of vehicle location information sections do not match, and put forward the vehicle path selection algorithm, and ultimately by the complete and continuous data interpolation to generate experimental data. Second, the analysis of experimental data and vehicle motion pattern extraction. Social behavior of people accustomed to a certain amount of time and space regularity. Through the analysis of a large number of vehicle history running track, we found that the movement of vehicles similar regularity. These regularities vehicle movement patterns exist in some sections. To extract these movement patterns, we apply a variable-length multi-order Markov model to model the movement patterns of vehicles. Compared with the traditional Markov model, the variable-length multi-order Markov model can be easily extracted vehicle motion mode of different order numbers. Third, the vehicle path prediction. Movement of the vehicle mode, the traveling path of the vehicle a short distance can be used for prediction. If the vehicle is currently located on the section there is a motion mode, then passes under the front of the vehicle sections, we can with a high probability to predict to the next road sections. For experimental verification, according to the prediction algorithm of the traveling path of the vehicle motion model, it is possible to obtain a good predictor of performance. Traffic conditions on the vehicle path prediction. Urban road traffic conditions change dynamically. Order to assess the impact of the traffic conditions on the movement pattern of the vehicle, traffic conditions as an important training parameters of the variable-length multi-order Markov model, and extracting the movement pattern of the vehicle under different traffic conditions. The prediction process of the traveling path of the vehicle, according to the different traffic conditions to apply a different motion mode for prediction. Experimental results show that the path prediction algorithm based on traffic conditions to get better prediction accuracy. Five predictable application by the traveling path of the vehicle. The characteristics of the underlying movement of the vehicle self-organizing network has an important impact on its performance. Movement pattern of the vehicle as the vehicle an important part to the self-organizing network movement characteristics, can be used to optimize the design of the vehicle self-organizing network. In our work, we will be the vehicle path prediction algorithm used in the design of the vehicle self-organizing network routing protocols and data transfer mechanisms. Vehicle path prediction algorithm is applied in the same car under the self-organizing network environment, network protocol showed better performance.

Related Dissertations

CLC: > Industrial Technology > Radio electronics, telecommunications technology > Wireless communications
© 2012 www.DissertationTopic.Net Mobile