2.1 Traffic data collection and traffic state recognition
Traditionally, real-time traffic information collection is categorized into three schemes [LZ05]: site-based, sensor-based and probing vehicle-based data collection schemes.
Site-based measurement collects vehicle license plate characters and arrival times at various checkpoints through automatic vehicle identification (AVI) technologies, matches the license plates between consecutive checkpoints, and computes travel times from the difference between arrival times. Vehicle-based methods analyze the raw data collected from fleet of probe vehicles by matching the vehicle tracks with geographical information system (GIS).
Sensor-based scheme collects raw data from the stationary sensors like loops detectors, transponders or radio beacons installed at arterial roads. However, each traffic information collection method has some drawbacks and limitations. For example, site-based and sensor based methods have the spatial coverage problem due to the fixed and limited sensors or AVI devices. Vehicle-based scheme [NT04, YAN05, CH+03] has cost, spatial and temporal coverage problems due to the very high cost for maintaining a dedicated fleet of urban network traffic probing vehicles. Besides, the cost of real-time transmission for the whole traffic network in each data collection scheme is also very high [LTT09].
15
2.2 Traffic patterns mining and traffic bottlenecks discovering
In the literature, many researches aimed to find out the traffic patterns[ZZ+04, TLT05, TSA06, CHU03, CMC07] or to identify traffic states[KD+05] in traffic network, and some researches worked on predicting the travel time to provide drivers route suggestion[LTT09, KKS05]. Kerner [KD+05] proposed FCD (Floating Car Data) method to recognize traffic state (e.g., congested or not) by FCD vehicles in urban network, but still cannot identify the locations of the bottlenecks. Till now, most of previous studies tried to locate and control congestion patterns on highway bottlenecks [KER05, KER07] which are usually static, clear and always located around the gateway, but locating the bottlenecks on the urban network is more difficult than that on the freeway because there are no intersections and traffic signal control on the freeway. In other words, the task of analyzing traffic patterns in urban network and finding out traffic bottlenecks is a complex and difficult mission due to the following reasons: first, traffic network in urban area is more complex than that in freeway or simple arterial network; second, bottlenecks in urban network which are spatiotemporal dynamic varied with spatial or temporal environment and varied with traffic demands; third, not only more traffic factors but also more non-traffic factors have to be concerned in urban network than in freeway, such as traffic signal, social event, and traffic incidents, etc. Recently, [LG+08] tried to recognize urban traffic congestion propagation and identify bottleneck based on Cell Transmission Model (CTM), which discretizes each roadway into homogeneous section (cell) and discretizes time into intervals. Given the network objects capacities, it tries to identify the network bottlenecks by simulating the traffic demands of the urban network. In this dissertation, we try to not only discover the statistical and congestion propagation types of network bottlenecks, which similar to [LG+08], but also define and discover several types
16
of traffic patterns including object level patterns and area level patterns, which are then transformed to spatiotemporal traffic bottlenecks by our proposed heuristic methods.
2.3 Dynamic traffic assignment (DTA)
DTA researches have evolved substantially since the pioneering work of Merchant and Nemhauser [MN78A, MN78B], which are typically classified into two broad categories:
analytical models and simulation-based heuristic models [ZW+04]. The analytical models can be further classified by three groups: mathematical programming, optimal control, variational inequality [PZ01]. Efforts in the analytical models include mathematical programming approaches by [JAN91] and [ZIL00], optimal control theory based formulations by [FL+90], and variational inequality approaches introduced by [FB+93] and [SMI93]. The evolutions and literature review of the related analytical as well as simulation based approaches was done by Peeta and Ziliaskopoulos [PZ01]. Most analytical formulations are extensions of their static formulations and seem to have two main disadvantages: 1) they cannot adequately capture the realities of street network due to simplifications, and 2) they tend to be intractable for realistic size networks [ZW+04]. Besides, most analytical researches obtain the O-D data by assumption, estimation or simulation, which cannot capture the spatiotemporal traffic demand of the traffic network. So that the assignment suggestions based upon the assumption, estimation or simulation won’t have good effect for network performance enhancement.
2.4 Travel time prediction (TTP)
Study in [IK01] found that level of reduction in congestion depends on the complexity of
17
the road network. Vehicular flows on freeways are often treated as uninterrupted flows; flows on urban network are conceivably much more complicated since vehicles traveling on urban network are subject to not only queuing delays but also signal delays and turning delays. Thus, TTP for an urban network is more challenging than predicting the travel time for freeway or single arterial. Besides, the routing and path selection problems should be solved in TTP for urban network, i.e., the TTP model has to suggest a shortest travel time path on a given OD (origin, destination) pair as a request. Many models had been proposed for travel time prediction in these decades, but most of them focused on predicting the travel time on freeway [WHL04, CK03] or simple arterial network [JZ03, LKM04].
In the past, many ITS studies and transportation agencies use the traffic data from dual-loop detectors which are capable of archiving with traffic count (the number of vehicles that pass over the detector in that period of time), velocity, and occupancy (the fraction of time that vehicles are detected) and readily available in many locales of freeways and urban roadways [LZ05]. Nowadays, traffic data collecting techniques have made great progress and evolved to real-time collecting in order to improve traffic management efficiency. In [LZ05], traffic information collection and travel time measurement can be divided into three categories: site-based, vehicle-based and sensor-based measurement. Site-based measurement collects vehicle license plate characters and arrival times at various checkpoints through automatic vehicle identification (AVI) technologies, matches the license plates between consecutive checkpoints, and computes travel times from the difference between arrival times.
Vehicle-based methods make TTP by analyzing raw data collected from fleet of probe vehicles. Sensor-based methods make TTP measurement by collecting raw data from the stationary sensors like loops detectors, transponders or radio beacons installed at arterial roads.
However, each traffic information collection method used for TTP has some drawbacks and
18
limitations. For example, site-based and sensor based TTP methods have the spatial coverage problem because the sensors or AVI devices are fixed and limited. Vehicle-based TTP methods [NT04, YAN05, CS+03] have cost, spatial and temporal coverage problems because the total cost is very high if a dedicated fleet of urban network traffic probing vehicle is maintained.
There are numerous previous TTP approaches based on the historical traffic data analysis in the literatures, which can be categorized as follows [LZ05]: regression method (mathematical model) [WHL04], time series estimation method, hybrid of data fusion or combinative model [WLC05] and artificial intelligence method like neural network [MSR04].
In [NT04], auto regression (AR) model and state space model for time series modeling were used to predict travel time. The Kalman filtering provides an efficient computational (recursive) in many TTP researches [LKM04, CHU03, YAN05], because it is very powerful in several aspects: it supports estimations of past, present, and even future states even if the precise nature of the modeled system is unknown. In [WHL04], the support vector regression model was used to predict travel time for highway users. In [BCK04], pattern matching technique was used for TTP. Traffic patterns similar to the current traffic are searched among the historical patterns, and the closest matched patterns are used to extrapolate the present traffic condition. Chung et al. [CH+03] developed an O-D estimation method to make more accurate estimation of traffic flow and traffic volume in congestion traffic status. Moreover, the data fusion models of TTP integrated grey theory [TTI03] and neural network-based.
Yang [YAN05] developed some hybrid models toward data treatment and data fusion for traffic detector data on freeway.
19