Real-time Traffic Estimation And Path Planning Based On Crowdsourced Mobile Phone Data

With the advance of science and technology,the seamless integration of technology innovation and urban management has become an important goal of the smart city development,wherein one of the important parts is the construction of intelligent transportation system(ITS).Real-time traffic estimation and route planning are two major services provided by an ITS system.The key purpose of these two services is to guide people to travel more intelligently and build a safer and faster traffic environment.At present,most of the traffic source data is collected by fixed equipments(e.g.,loops and cameras)and floating vehicles(e.g.,taxi).However,both methods have the disadvantages of high deployment cost and low coverage.Today,mobile smartphones are becoming more and more popular and powerful,which have integrated with various advanced capabilities,such as data collection,processing and transmission.Thus,researchers have proposed to collect the traffic source data by crowdsourcing the GPS data of a large number of smartphones.Comparing with the above two traditional data collection methods,this new data collection method has the advantages of higher coverage,lower deployment cost,and more real-time.In this thesis,we will study the real-time traffic estimation and dynamic route planning based on the crowdsourced GPS data of mobile smartphonesThe real-time traffic estimation mainly consists of three important parts: mobile GPS data preprocessing,data mapping and traffic estimation.For mobile GPS data preprocessing,we adopt the “threshold method” to filter the error data and clean the dirty data.This method can effectively remove the non-vehicle data,redundant data and error data.For the data mapping,we propose a Hausdorff distance matching algorithm based on the road network topology,which can quickly map the high frequency GPS sampling trajectories to a realworld map.Based on the data mapping result,we calculate the road travel time by using the speed integral method and the exponential weighted method,and further tranform it into the descriptive road traffic condition.Finally,we perform simulations with the real GPS data collected from mobile smartphones on Fufei Bei Road in Fuzhou,which show that our proposed real-time traffic estimation method based on the crowdsourced GPS data can correctly estimate the real road conditions.The dynamic path planning mainly consists of two important parts: dynamic hierarchical network partition and dynamic path planning.For the former part,we first use several static road attributes to achieve the static road partition,and then attach the real time road traffic information(obtained in the above)on to the static road partition,hence constructing a dynaic hierarchical road network.After the dynamic road partitioning,we further propose a dynamic hierarchical path planning algorithm based on the multi-attribute dynamic hierarchical network.Simulations and experiments show that the proposed dynamic path planning can achieve the shortest route in terms of the traveling time with a low computational complexity.Moreover,when the road traffic condition cha nges suddenly,our proposed algorithm can re-schedule the route according to the real-time traffic information,so as to avoid the congestion roads.Comparing with the traditional static route planning algorithms(which do not re-schedule the route),our proposed algorithm can achieve a shorter traveling time.