Traveler Oriented Dynamic Traffic Assignment Model And Distributed Path Finding Algorithm

The contradiction between the traffic supply and demand surges increasingly due to therapid development of the social economic. In order to assign traffic reasonably within limitedresources, it becomes more important to study the dynamic traffic assignment problem whichis one of the most important parts in the intelligent transportation system. Current dynamictraffic assignment models are mostly confined to the theory and performance analyzing, butless focus on the models’ manageability on the real road network, which makes it hard to beapplied directly to the traffic management. What’s more, the path finding problem which isthe key issue of the model is usually studied separately. Except for the limitation ofalgorithms due to different objectives, it is also not easily to incorporate algorithms into thedynamic traffic assignment model because of its separation from the transportation research.The problems are studied and discussed in this paper, considering the rapid development ofrelated technologies of the Internet of Things and Cloud Computing。First of all, this paper proposes a model called TOM-CTM which incorporates thetraveler dynamics into the cell transmission model (CTM) by decoupling the travel demandfrom the traffic demand, in order to simulate traffic dynamics in a more controllable way. Themodel takes travelers as the source of the traffic load and formulates the relationship betweenthese two demands mathematically. With the proper design, the model can be used as anenhancement to any existed cell transmission model. For implementation in this paper, anopen source macro-simulation tool, called Aurora Road Network Modeler (AuroraRNM), isadopted. The experimental results show the traveler oriented model can reflect the trafficdynamics properly and reasonably, and also provides a way to find out the manageable statusin a nonequilibrium traffic. Although user optimal or system optimal solutions present optimalequilibrium traffic, this new model is more practical as such analytical optimal solutionsremain hard to apply to the real world management due to limited resources or other reasons.Secondly, this paper proposes a distributed algorithm called BBFS_SDTD to solve thetime dependent shortest path problem. The algorithm process all related data of the shortestpath distributedly, and break the limitation existed in previous algorithms. The algorithm iswritten on the MapReduce framework in Hadoop from the perspective of the data intensiveprocessing with respect to the computing intensive processing before. Therefore, thisalgorithm is inherently suitable for processing historical traffic data offline. BBFS_SDTD iscompared with the other two algorithms in this paper, and all algorithms are tested on threenetworks with different scales. The results show when the network is small, the algorithm proposed needs relatively long time to run, and most of time are spent by the framework onscheduling, data transfer, etc. But when the network becomes large enough, BBFS_SDTD’sadvantage in running time is evident. And whatever the scale of the network is, it can getmore information of shortest paths in the minimum amount of time.Finally, the algorithm aforementioned is also applied to the dynamic traffic assignment inthis paper, which comes out a novel model called HTIGM which takes the historical trafficinformation as the guide. It is designed for the deviation from the dynamic traffic assignmentmodel during the optimization of path finding problems, and clears the relationship amongmodules and their processing logic. The model takes the results of the distributed algorithm asthe input, studies the dynamics of traffic with different proportion of travelers accepting thehistorical traffic information. The corresponding experiments show that under the roadnetwork and demand specified in this paper, with the increased proportion, the delay of thenetwork presents downward trend at first then remains relatively stable; however, the vehiclehours traveled first increases and then declines. Therefore, when appropriate proportions oftravelers choose the time dependent shortest path, the traffic can reach a reasonable andmanageable status. Besides, the model also can be used to find out the bottleneck of currentroad network.