Research On Traffic Sensor Optimization Problem Under Uncertainty

It is of great importance to collect accurate and reliable traffic information for both transportation management department and travelers. Traffic sensor is an effective means of collecting various information of transportation network. It provides large amount of information for many transportation applications including traffic flow detection, Origin Destination(OD) estimation, travel time calculation, etc. On one hand, the optimization of traffic sensor network will largely affect the performance of collecting information and subsequent computation. On the other hand, the uncertainty is very common in transportation network. For example, the change of flow will have an effect on the travel time, or the traffic incident will have an effect on the network topology. In these situations, the original scheme under the priori information may not function well in fact. In addition, the uncertainty is also reflected in time and space. From the perspective of time, there are differences in transportation network status in different time in a day. From the perspective of space, in different time different links are of different importance for collecting information. Therefore, the optimization of traffic sensor network under uncertainty is a problem worthy of studying.This article focuses on how to place traffic sensors to attain the optimized traffic application performance under uncertainty. This article mainly study two ways to solve the problem:one is to use a mobile traffic sensor to improve the ability to collect information of traffic network, in this problem, it mainly aims to optimize the route of mobile sensors to estimate the OD; the second is to employ the time-space relationship in designing traffic network. In this condition, we assume that flow on one link can be obtained from related links. Therefore, there is no need to get the network fully covered. At last, the paper gives a summarize and the future research direction.