اندازه گیری همزمان از تعادل نفوذ در بازار و میزان انطباق سیستم های پیشرفته اطلاعاتی مسافران

We consider a specific advanced traveler information systems (ATIS) whose objective is to reduce drivers’ travel time uncertainty with recurrent network congestion through provision of traffic information. Since the provided information is still partial or imperfect, drivers equipped with an ATIS cannot always find the shortest travel time route and thus may not always comply with the advice provided by ATIS. Thus, there are three classes of drivers on a specific day: drivers without ATIS, drivers with ATIS but without compliance with ATIS advice, drivers with ATIS and in compliance with ATIS advice. All three classes of drivers make route choice in a stochastic manner, but with different degree of uncertainty of travel time on the network. In this paper we investigate the interactions among the three classes of drivers in an ATIS environment using a multiple behavior stochastic user equilibrium model. By assuming that the market penetration of ATIS is an increasing function of the actual private gain (time saving minus the cost associated with system use) derived from ATIS service, and the ATIS compliance rate of equipped drivers is given as the probability of the actual travel time of complied drivers being less than that of non-complied drivers, we determine the equilibrium market penetration and compliance rate of ATIS and the resulting equilibrium network flow pattern using an iterative solution procedure.

Advanced traveler information systems (ATIS) have experienced a rapid growth in the past decade, as an alternative to alleviate traffic congestion and enhance the performance of road networks. There has been substantial interest in user behavior modeling under ATIS and benefit evaluation of ATIS in order to determine the feasibility, benefits and risks of such technologies. Various theoretical and practical results have been achieved through approaches of laboratory experiments (for example, Mahmassani and Jayakrishnan, 1991; Yang et al., 1993; Reddy et al., 1995), mathematical and computer simulation models (for example, Ben-Akiva et al., 1991; Kanafani and Al-Deek, 1991; Al-Deek and Kanafani, 1993; Hall, 1993; Yang, 1998) and even field deployments (for example, Tsuji et al., 1985). The market penetration of ATIS, defined as the proportion of vehicles (drivers) equipped with ATIS, has been widely recognized as an important factor to determine the actual advantages of ATIS implementations (for example, Yang et al., 1993; Yang, 1999). Most previous studies have focused on the investigation of the advantages or benefits by assuming varied levels of market penetration exogenously. Furthermore, equipped drivers were assumed to be spread homogeneously among all origin–destination (O–D) pairs, regardless of their trip characteristics (for a review, see Yang, 1998; Yang et al., 1999). It is, however, questionable whether these exogenous given ranges of market penetrations will be attainable or sustainable in a true competitive ATIS market. Recently, Yang (1998) developed a theoretical endogenous market penetration model for ATIS services in a road network. Market penetration was determined from the information benefit derived from ATIS using a mixed stochastic and deterministic network equilibrium model. More recently, Yang et al. (1999) addressed some important issues in the evaluation of ATIS benefit when market penetration is an endogenous variable. Both private and system benefits by ATIS have been evaluated with reference to trip lengths, level of traffic congestion and degree of travel time variability as well as annual system usage cost and driver’s value of time. They found that different trip lengths are associated with different travel time savings and hence different market penetrations, ignoring this important factor may lead to misleading results of ATIS performance. In particular, the previous assumption of uniform market penetration may underestimate the expected system benefit achieved from ATIS in terms of total travel time reduction. Besides the determination of final saturation level of market penetration of ATIS services, Yang and Meng (2001) have further modeled the time line of the growth of market penetration to reach such a saturation level based on a modified logistic type growth model. The parametric sensitivity analyses of the growth pattern might be useful for the optimal control of temporal evolution of ATIS products or services (Yang and Huang, 2002). It is obvious that the real benefits of ATIS implementations critically depend on how drivers will respond to these systems. All of the aforementioned studies assumed perfect compliance of equipped drivers with ATIS routing advices. However, although ATIS are intended to provide more accurate real-time information to drivers reducing the degree of travel time uncertainty, it is doubtful whether drivers would ever rely on these computerized guidance systems. Even there is no deliberate attempt being made to sacrifice individual benefits in the interests of a system-optimum, drivers’ compliance with ATIS is still influenced by information attributes such as quality, nature and feedback, traffic conditions, driver characteristic, and prior experience. In recognition of this issue, a few authors investigated ATIS compliance behaviors by addressing how different factors affect drivers’ performance and limit compliance with ATIS advices (Bonsall and Parry, 1991; Boehm-Davis and Fox, 1998; Chen et al., 1999; Srinivasan and Mahmassani, 2000). Recently, Oh et al. (2001) considered compliance rate as an endogenous variable and then developed a framework to parametrically evaluate network performances under user equilibrium route guidance and system optimal route guidance. Nevertheless, the market penetration of ATIS is fixed and given and a behaviorally sound method for determining the compliance rate remains to be explored. In this paper, we consider simultaneous determination of endogenous equilibrated market penetration and compliance rate of ATIS using a multiple behavior equilibrium model. We consider a specific information system whose objective is to reduce variations of the travel time perceptions of drivers with recurrent network congestion through provision of travel information. It is expected that the drivers who are equipped with an ATIS will receive more adequate information, and if they comply with the advices, they will have less perception errors of actual travel times than unequipped drivers. Nevertheless, due to the partial or imperfect information from ATIS, not all equipped drivers would necessarily follow the ATIS advices. Therefore, there are three classes of drivers to be considered: drivers without ATIS, drivers with ATIS but without compliance with ATIS advice, drivers with ATIS and in compliance with ATIS advice. These three classes of drivers interact with each other on a common network in an ATIS environment. We first apply a multiple behavior stochastic user equilibrium (SUE) model to describe the interactions among the complied, non-complied and unequipped drivers for given market penetration and compliance rate of ATIS. In view of the fact that travel time is the most important factor to dictate drivers’ route choice, it is thus rational to assume that the actual travel time saving will ascertain whether or not for a driver to equip an ATIS and, if equipped, whether or not to follow ATIS advice. The market penetration of ATIS, defined as the percentage of drivers equipped with an ATIS, is determined by a binary choice model by assuming that drivers decide to buy an ATIS if the net benefit (actual time saving minus cost associated with system use) exceeds a certain level of threshold. The compliance rate of ATIS, defined as the proportion of complied drivers in total equipped drivers, is given as the probability of the travel times of complied drivers being less than those of non-complied & unequipped drivers who make route choices based on their own perception. With these considerations, we propose an iterative procedure to calculate the endogenous equilibrated market penetration and compliance rate of ATIS. The paper is organized below. In Section 2, we present the multiple user behavior SUE model to describe route choices of compiled and non-complied and unequipped drivers. We propose a binary choice model for determination of ATIS market penetration in Section 3, and further model equipped drivers’ behavior of compliance with ATIS advice in Section 4. An iterative solution algorithm is developed in Section 5 and illustrated with a numerical example in Section 6. Finally, conclusions are presented in Section 7.

In this paper, an attempt is made for simultaneous modeling of endogenous market penetration and compliance rate of ATIS in a network with recurrent congestion using a multiple behavior equilibrium model. Market penetration of ATIS is determined from the private gain (time saving minus the cost associated with system use) and formulated as a binary choice model. Compliance rate of ATIS advice is related to the quality of information and is derived from the probability of the actual travel time of complied drivers being less than that of non-complied & unequipped drivers. The proposed fixed-point iterative algorithm is applied to a numerical example and demonstrated to have in general a fast convergence. It is observed that the quality of information affects both market penetration and compliance rate. In contrast, the level of congestion affects market penetration, but has little impact on the compliance rate of ATIS. It should be noted that the actual response or reaction of drivers to travel information might be much more complex than that assumed in this study. Nevertheless, our study does make a good attempt in simultaneous determination of the market penetration and compliance rate, which are two important issues to be explored in actual ATIS implementation. The proposed approach can be applied to improve the adequacy of many efforts made previously on ATIS modeling, such as examination of the temporal evolution of adoption of ATIS products made by Yang and Meng (2001). With better consideration of users’ response to ATIS services, enhancement of the performance of these models could be expected. Another extension of our model is toward the evaluation of ATIS benefit under non-recurrent congestion. To this end, dynamic multiple user behavior traffic assignment with endogenous market penetration and compliance rate of ATIS is required.