The Effect Of Routing And Adaptive Traffic Light Strategy In A Manhattan-like Urban System

Realistic traffic network can be seen as a system formed of numerous self-propelled particles. In the process of evolution over time, the system will present many rich and complicated non-equilibrium physical phenomena. As the economy develops, the urban population increases, corresponding with the increase of vehicles. This directly results all kind of traffic jams, especially at a rushing hour or on public holidays. Hence the improvement of urban traffic becomes the focus of attention. Traffic flow theory research is developed on this purpose. By trying new methods like complex network, cellular automata et al., physicists are able to build traffic network models and do simulations in all kind of conditions. This is very helpful to reveal the basic rules of traffic network, and to put forward the optimization method of solving traffic congestions. It may also provide as guidance for the transport departments to better plan, design and improve the traffic network.In this paper, we take a realistic traffic network as the prototype. By studying its basic structure and characteristics, we build our own model and acquire lots of non-equilibrium physical phenomena by doing simulations on it. The results can be a reference to vehicle motions in a traffic network. Based on the predecessors’work, we researched the impact of the routing strategy and self-adaptive traffic light in a Manhattan-like traffic network. The contents of the paper are as follows:1. We have discussed the impact of different routing strategy to the whole traffic system. By running numerical simulations under the same conditions, we get all kinds of figures of the system. Comparison of the results shows that:·random routing strategy: Vehicles pick a direction randomly when they get more than one choices at the intersection. This method is simple and intuitive, and can be accomplished by vehicles in the system automatically. However, the drawback is that vehicles will concentrate in the center of the network. Some intersections have to deal with cars coming from all four directions, forming long queues on the road. It’s very easy to cause a jam and lead to global deadlock.·ATISSR: Vehicles pick up the road that has a higher level of mean velocity when face more than one choices at the intersection. Under this strategy, the system presents a windmill-like density distribution, with peak values on the diagonal. Vehicles from different direction concentrated in different corners. ATISSR just changes density distribution, it hardly does any difference on mean flux and mean velocity. Furthermore, the execution of ATISSR requires additional observations and statistics of road information, and equipment to convey the information to the head vehicle on each road.·ATISMR: Vehicles will find all the shortest passes between the destination and current position and pick the one with the highest mean velocity. The system showed a diamond-like density distribution with peak value on the corner of the diamond. Since vehicles are concentrated on four side of the network, the system gets plenty of room there and is able to put up with more cars, hence improving NOR to a new level. In the meantime, ATISMR is able to improve mean flux and mean velocity. Same with ATISSR, ATISMR requires additional observations and statistics of road information, and equipment to convey the information to the head vehicle on each road. The other challenge is dealing with massive information in a very short time, hence productive algorithms are needed. ATISMR is an optimal option when the other two strategies don’t work.2. Introducing self-adaptive traffic light strategy into the system will relate green time to the number of cars on the corresponding road. Three variables are involved in this strategy. We researched how the variables impact the system by numerical simulations. The results are as follows:·static green light Tstatic:Static green light Tstatic is a fixed value and allocated to every incoming road of an intersection. It’s to ensure that every entry gets some green time during a traffic light cycle. The smaller Tstatic is, the more depended green time is to the density on the road. In a system with self-adaptive traffic light, cars tend to concentrate to the center of the network, with peak value of density drop down at the same time. This effect significantly improves NOR of the system.·variable a: The adjustment of variable alpha can affect the allocation of green time at the intersection, hence affects NOR. Under ATISSR, when a is on a relatively low level, NOR increases correspondingly to a, reaches to a maximum value and then declines. Under ATISMR, when a<5, NOR will maintain on a higher level, and then decreases rapidly with the increase of a. Furthermore, as a increases, vehicles concentrate in the center of the network and density distribution is more uniform.·traffic light cycle T: System is quite sensitive to the change of traffic light cycle T. It can be seen from the fluctuations of NOR when T changes. Different traffic light cycle T can significantly affect the duration of the average speed of stable phase, thus affecting NOR.