Research Article  Open Access
Shijie Shu, Jing Zhao, Yin Han, "Signal Timing Optimization for Transit Priority at NearSaturated Intersections", Journal of Advanced Transportation, vol. 2018, Article ID 8502804, 14 pages, 2018. https://doi.org/10.1155/2018/8502804
Signal Timing Optimization for Transit Priority at NearSaturated Intersections
Abstract
Transit signal priority is a useful way to improve transit operations in urban networks. Most of the existing studies have been conducted in conditions with low saturation to avoid the detrimental effects of vehicles without priority. However, from the public transit point of view, it is more meaningful to assign transit signal priority when the degree of the saturation intersections is high. This study proposes a signal control model for transit signal priority to minimize the overall delay at nearsaturated intersection. The delay increment is calculated in three scenarios for buses and private vehicles according to the dissipation time of the vehicular queue. A set of constrains are set up to avoid queue overflows and to ensure the rationalization of the signal timing. The proposed control model is tested based on a case study and numerical experiments. The results show that the proposed model can reduce the total person delay at nearsaturated intersections. The length of priority time, degree of saturation, and number of lanes are the three main influencing factors. More than 6% reductions in person delay can be obtained for undersaturated intersections when the priority time is less than 5 s. Moreover, even when the intersection saturation is 0.95, the bus signal priority can be applied if only the priority time is less than 5 s.
1. Introduction
As cities become more and more congested, land for road construction is limited. Therefore, congestion needs to be alleviated by changing the model of travel [1]. The improvement of public transit service can elicit important social benefits. Transit signal priority and exclusive bus lanes are two common methods used to improve the public transit service [2]. Setting exclusive bus lanes is a costeffective approach for providing a highquality transit service, enhancing the reliability of bus services, and increasing the speed of buses [3–6]. Although continuous running ways yield great benefits to transit operations, transit vehicles will be delayed at intersections by traffic signals [7], especially buses that arrive at the end of the green light period. Therefore, to further improve the running efficiency of buses at intersections, the exclusive bus lanes need to be perfected by assigning transit signal priority.
Transit signal priority is a typical strategy used to reduce the delay of public transit, and it has been used extensively to improve transit operations in urban networks. Existing transit signal priority strategies can be divided into three major categories: passive priority strategies, active priority strategies, and realtime priority strategies [8].
Passive priority strategies are developed on the basis of historical data and do not require any detection system. Skabardonis introduced transit priority measures on urban networks [9]. It has been proven that the passive priority strategies are effective in simple network configurations and elicit predictable dwell times and high bus frequencies. However, it was also pointed out that the passive priority strategies were less adaptable when vehicular flow was low. Estrada proposed a passive signal priority simulation model which restricted the maximum car passenger delay increment [10]. The model was applied to a series of trial networks, and the results showed that the model can reduce the bus travel times by 8.5%, while concurrently increasing the car delay by 5%. Correspondingly, the detection system developed numerous active priority strategies which could provide more effective priority to buses.
Active priority strategies include phase extension, phase advance, phase insertion, and phase rotation [11]. These strategies respond to traffic variations in realtime and are therefore more effective than passive priority strategies. Ngan et al. examined the impacts of a set of traffic parameters on the effectiveness of a transit signal priority application [12]. The results show that transit signal priority application would be more effective when the bus volume is moderate to heavy. Jacobson and Sheffi presented an analytical model for the bus preemption analysis [13]. The results showed that the benefits of bus preemption were small when the crosstraffic flow was much lower than the preemption direction. Furth and Muller put forward a method of operational control which can keep buses on schedule [14]. This method has been implemented in the Netherlands. The results showed that the proposed method can improve the bus schedule. In addition, conditional priority elicited minimal delay increments compared to the case where no priority was assigned, while absolute priority caused large delay increments. Zeeshan et al. proposed an analytical model for estimating the delay impacts of several transit priority strategies [15]. Active priority strategies often have detrimental effects on the vehicles that have no priority, so realtime priority strategies were developed which focused on the total vehicular delay or the total passenger delay at signal intersections.
Realtime priority strategies optimize signal timings based on performance criteria, such as passenger delays, vehicular delays, and combinations of vehicular delays. Ling and Shalaby put forward an algorithm which is based on reinforcement learning and the control policy is learned by the algorithm [16]. The best duration of each phase is determined by the reinforce learning. The experimental results show that the proposed algorithm can reduce the transit headway deviation and disrupt cross street traffic smaller. Christofa and Skabardonis presented a realtime signal control system for signal priority on conflicting transit routes that also minimized the negative effects on vehicular traffic [17]. To solve the issue of multiple priority requests, Head proposed a model which could analyze complex controller behaviors [18]. The analyzed results showed that the proposed model for serving priority requests resulted in fewer delays than a firstcome, firstserved policy. From the perspective of intersection groups, Ma presented an optimization model which could generate the optimal combination of priority strategies for intersection groups [19]. Christofa et al. presented a realtime signal control system for assigning signal priority on conflicting transit routes that also minimized the passenger delay on arterials [20]. Ma et al. proposed a priority signal model for multiple bus requests [21]. The optimal priority serving sequence is generated by the proposed model, so the bus can maximize the utilization of available green times. With the development of communication technologies, optimization methods were proposed in a connected vehicle environment. Hu presented an optimization method for an intelligent transit signal priority logic that enables bus signal cooperation among consecutive signals [22]. The results showed that the intelligent TSP logic can efficiently reduce the bus delay compared to the conventional TSP. Li et al. proposed an adaptive transit signal priority optimization model that optimizes green splits for three consecutive cycles to minimize the vehicle delay [23]. The numerical example results show that the bus delay and the traffic delay along the bus movement direction were all reduced. He and Head presented a requestbased, mixedinteger, linear program, which can coordinate multiple priority requests from pedestrians and different modes of vehicles [24]. Lin presented a control model that minimized bus passenger waiting time at downstream bus stops but did not increase the delay for vehicular passengers [25]. The test results show that the proposed model can reduce the total intersection delay and the bus passenger waiting time. Altun and Furth found that the optimal performance that used conditional priority under an aggressive schedule can lead to obvious reductions in running time and headway irregularity [26]. A novel method was also proposed by Guler and Gayah that provided priority to buses at signalized intersections using singlelane approaches [27]. The method used addition signals to stop cars on the opposing travel lane so that the bus could use the travel lane in the opposite direction. The results showed that the presignal strategy could reduce the bus and the total passenger delays. The genetic algorithm was used to optimize four basic signal timing parameters and transit priority settings. The results showed that the genetic algorithm elicited a good performance in the optimization of the transit priority setting [28]. Ma et al. assumed that the bus speed is available to adjust in realtime [29]. A set of integrated operational rules which integrated operation of signal timings and bus speed were developed to provide priority to bus at isolated intersections. The experimental analyses show that the proposed integrated operational rules perform better than the priority strategies which do not adjust the bus speed.
Most of the existing transit priority studies have been conducted in lowsaturation conditions. It was argued that the intersection under nearsaturation conditions is difficult to adopt transit signal priority, owing to the detrimental effects on the nopriority vehicles, which will lead to long vehicular queue dissipation times and long queue lengths [30]. Accordingly, serious traffic jam will be easily caused by assigning transit signal priority. However, from the public transit point of view, it is more meaningful to adopt transit signal priority when the intersection saturation is high. It can improve the level of service of public transit systems which would encourage more travelers to choose the transit mode for their travels.
To remedy this deficiency, this study aims to create a signal control model for transit priority at intersections with high degree of saturation. A passengerdelaybased optimization method is used, in which both the performance of buses and private vehicles are considered. The rest of this study is organized as follows. In Section 2, a green extension model is presented for signal priority on buses that minimizes the total passenger delay at intersections with increased saturations. In Section 3, a case is presented, whereby the signal control model is tested in a nearsaturated intersection in Shanghai, China. In Section 4, the sensitivity analysis of the proposed model is presented. In Section 5, conclusions are outlined.
2. Methodology
2.1. Basic Idea
To reduce bus delays, especially when these arrive at the end of the green traffic light periods, the green extension strategy is a common priority strategy that has used in numerous optimization models. Owing to the increase in the time period the green light lasts for buses, the corresponding periods for which the green light is on for other conflicting phases should be shortened. Therefore, the green light period for other phases should be resplit. Correspondingly, three commonly used ways to achieve this include the distribution of all the increased redlight times to a certain phase [31, 32], distribution of the increased redlight period to subsequent phases according to the phase splits [33], and the resplitting of the green light period according to the degree of saturation. However, these redlight time distribution strategies may not lead to optimal signal timing results. This study proposed a signal timing optimization model for transit priority at intersections with high degree of saturation to minimize the total intersection passenger delay. As shown in Figure 1, the green light period of movement j, which has the same direction as that of the bus, is extended to . The delay of movement will increase owing to the green extension . To decrease the delay of movement , the green light period of movement will be extended to, and the same strategy will be adopted in the following movements.
The proposed model aims to minimize the total passenger delay at the intersection by changing the lengths of the green light periods for each movement j. Transit vehicles are assumed to travel on exclusive transit lanes. The dualring, eightphase structure is considered herein, because it is universally applied in the traffic control industry [34]. The advantage of the dualring concept is that it allows the first (usually leftturn) phases in Rings 1 and 2 to terminate independently after their respective demands have been satisfied, which can enhance the flexibility of the signal timing [34]. A fourlegged intersection with eight movements is shown in Figure 2(a), and each ring in the controller contains four phases, as depicted in Figure 2(b).
(a) Intersection layout showing the associated phases
(b) Dualring phase structure
The model will be established in three steps: (1) calculation of the benefits (decreased delay) for private vehicles with the same movements as those for buses, and negative effects (increased delay) for other private vehicles when the green light extension strategy is adopted, (2) establishment of a minimized optimization model that considers the total delay at the intersection, and (3) use of enumeration methods to solve the model. For an easier understanding, a summary of the model notations is presented in Table 1.

2.2. Passenger Delay Increment Calculation
In this study, the total passenger delay increment was used as the indicator and comprised two parts: (1) the passenger delay increment of private vehicles at the intersection and (2) the passenger delay increment of buses at the intersection. The total passenger delay increment was defined as the delay in instances when the bus priority strategy was adopted at intersections minus the delay in instances when no bus priority was allocated at intersections. Therefore, the value of the total passenger delay increment should be negative when the transit signal priority strategy was adopted. Since the proposed model was expected to be used at nearsaturated intersections, the application of the transit signal priority may lead to the oversaturation of other traffic movements. For a fair comparison, the analysis included all the cycles until the queue lengths of all the movements were restored to their original conditions. According to the vehicular queue dissipation time, three scenarios are considered herein: (1) scenario 1, the queue length of movement j can be restored to zero before the end of the green light period that can serve in the first signal cycle, as shown in Figure 3; (2) scenario 2, the queue length of movement can be restored to zero before the end of the green light period that can serve in the second signal cycle, as shown in Figure 4; (3) scenario 3, the queue length of movement can be restored to zero before the end of the green light period that can serve in the third signal cycle, as shown in Figure 5. When more than three cycles are required for the queue length of movement to be restored to zero, the total passenger delay increment will be greater than zero. These cases are beyond the scope of the current analyses and are not discussed.
2.2.1. Passenger Delay Increment of Private Vehicles
(1) Scenario 1. In scenario 1, the queue length of movement can be restored to zero before the end of the green light period that can serve in the first cycle. Figure 3 shows the cumulative number of vehicles present at an intersection for cycle T and cycle . Without loss of generality, the beginning of the phase that adopts the bus signal priority is set to be the beginning of the signal cycle. The area of the yellow part is the delay increment caused by the extension of the green light, . The area of blue part is the delay decrement caused by the extension of the green light, . Then, the passenger delay increment of private vehicles on the movement in scenario 1 can be calculated by the following:Equation (1) is subject to the constraint:Variables in (1), and , can be calculated by (3) and (4), respectively. Intermediate variables, , , and , are the cumulative number of vehicles of movement at points A–D, which can be calculated by (5)(8), respectively. Intermediate variables, and , are the time of queue dissipation of movement j, which can be calculated by (9) and (10), respectively. Intermediate variable, , is the largest queue length of movement in scenario 1, which can be calculated by (11).
(2) Scenario 2. In scenario 2, the queue length of movement can be restored to zero before the end of the green light period that can serve in the second cycle. Figure 4 shows the cumulative number of vehicles present at an intersection for cycle T and cycle . The areas of the yellow parts, and , are the delay increments caused by the extensions of the green light periods and , respectively. The areas of the blue parts, and , are the delay decrement caused by the extension of the green light . Then, the passenger delay increment of private vehicles on the movement in scenario 2 can be calculated by (12). In (12), the condition implies that the queue length is larger than or equal to the queue length when the movement does not adopt the extension of the green light period . The condition implies that the queue length is less than the queue length in the instance when the movement does not adopt the extension of the green light period . Equation (12) is subject to the restrictions:Variables in (12), , , , and , can be calculated by (15)(18), respectively. The intermediate variable denotes the number of vehicles when the green light period is completed, which can be calculated by (19). The intermediate variables , , , , , and are the cumulative numbers of vehicles for movement at points A–F, which can be calculated by (20)(25), respectively. The intermediate variables , , and are the queue dissipation time for movement j, which can be calculated by (26)(28), respectively. The intermediate variable is the largest queue length of movement j, which can be calculated by (29).
(3) Scenario 3. In scenario 3, the queue length of group can be restored to zero before the end of the green time period that can serve in the third cycle. Figure 5 shows the cumulative number of vehicles present at an intersection for cycles T, , and . The area of the yellow part, , , , is the delay increments caused by the extensions of the periods when the green light was on. The area of the blue part, , is the delay decrement caused by the extension of the green light period. Then, the passenger delay increment of private vehicles on the movement in scenario 3 can be calculated by (30).Equation (30) is subject to the restrictions:Variables in (12), , , , and , can be calculated by (33)(36), respectively. Intermediate variables and are the number of vehicles when the time period of the green light is completed, which can be calculated by (37) and (38), respectively. The intermediate variables , , , , , and are the cumulative number of vehicles of movement at points A–F, which can be calculated by (39) and (44), respectively. The intermediate variable is the time of queue dissipation of movement j, which can be calculated by (45). The intermediate variable is the largest queue length of movement j, which can be calculated by (46).In summary, the passenger delay increment of buses at the intersection on movement in scenarios 1–3 can be expressed as (47). For the scenario 1, 2, and 3, (2), (13)(14), and (31)(32) should be satisfied, respectively. The passenger delay increment of private vehicles, , equals the sum of the passenger delay increment for private vehicles of all movements, as shown in (48).
2.2.2. Passenger Delay Increment of Buses
The passenger delay increment of buses, , is the negative value of the length of the remaining red light of movement 1, as shown in the following:
2.2.3. Validation of the Delay Model
The delay increment calculation should be validated through the microscopic simulation package VISSIM. The basic idea of delay increment calculation validation is as follows: as shown in (48), the passenger delay increment of private vehicles, , equals the sum of the passenger delay increment for private vehicles of all movements, so the delay calculation of each movement is calculated by the same equation (47). In other words, the validation of the passenger delay increment calculation is actually to validate (47).
In the following validation process, a test movement is selected, the basic parameters of the test movement and the green light extension time are input, and, after the simulation, the total vehicular delay of the movement is output. Because the correctness of the delay increment calculation under different scenarios needs to be tested, different green light extension times are input, which can result in shorter green times assigned to the vehicle and the queue length needing multiple cycles to return to zero. Finally, the total vehicular delay of the movement under the three scenarios is recorded and compared with the calculated values of the model.
The value of the input parameters are as follows: the cycle length is 100 s, the green time of the test lane is 25 s, the arrival rate and the saturation rate of the test lane are 360 and 1800 veh/h respectively, and the green extension time will be from 1 to 10. If the green extension time is 1,2,3, the queue length of test movement will be restored to zero in first cycle, so the delay increment calculation in scenario 1 will be validated. If the green extension time is 4,5,6, the queue length of test movement will be restored to zero in second cycle, so the delay increment calculation in scenario 2 will be validated. If the green extension time is 7,8,9,10, the queue length of test movement will be restored to zero in third cycle, so the delay increment calculation in scenario 3 will be validated. The simulation time of VISSIM is set to 4000s. The delay increment calculation validation results are presented in Figure 6. As depicted in Figure 6, the delay increment calculation value is close to the simulation value, and the maximum difference does not exceed 16%, the delay increment calculation is proved to be valid.
2.3. Optimization Model
2.3.1. Objective
The transit signal priority strategy is to improve the operational efficiency. Minimizing the total passenger delay was selected as the optimization objective, which is equivalent to minimizing the total passenger delay increment caused by the transit signal priority strategy, as shown in (50), in which the passenger delay increment of private vehicles and buses can be calculated by (48) and (49), respectively.
2.3.2. Constraints
To avoid vehicle queue spillovers, the queue length cannot be greater than the maximum queue length limitation, , as shown in the following:The green light extension for movement is less than or equal to the green light extension for movement j, as shown in (52). For , they are at the endperiods of the ring and cannot extend the green light period to the next movement. Therefore, they are equal to zero, as shown in (53). Moreover, in the dualring, all phases in one group must be terminated before the beginning of any of the phases of the next group. Thus, should be equal to , as shown in (54).The green time under the dualring phase plan should meet the constraint (55) due to the setting of the barriers shown in Figure 2(b). Moreover, the minimum green time period should be larger than or equal to the minimum green light period needed for the pedestrians to cross the street, as shown in (56).
2.3.3. Solution
The proposed model was coded in MATLAB and tested on an Intel i5, 2.5 GHz processor and 4.0 GB RAM, running Windows, and was solved using an enumeration method. The program running interface is presented in Figure 7.
3. Case Study
The proposed model is applied to a realworld intersection, namely, the intersection of the Huashan and Middle Yan’an roads located in Shanghai, China. The intersection’s layout is presented in Figure 8(a). As shown in this figure, the main through movement is via Yan’an road. The exclusive bus lane is set on Middle Yan’an road. The traffic volumes during the morning peak hours (8 to 9 a.m.) were obtained based on a field survey. The real data acquired from the Huashan and Middle Yan’an roads are presented in Table 2. The intersection is operated on a fixed fourphase cycle with a cycle length of 190 s, as presented in Figure 8(b). The saturation of the intersection is 85.2%. The average passenger occupancy of private vehicles is 1.5 per/veh; the passenger occupancy of buses is 20 per/veh.

(a) Original layout
(b) Original signal timing
Now, there is an application of the bus priority. The required bus priority time is 18 s. In order to better reflect the advantages of the proposed model, the basic timing plan will be adjusted, the existing timing plan will be optimized according to actual traffic flow. In this way, the problem of unreliable model results due to the irrationality of the basic signal timing can be solved. The adjusted signal timing is presented in Figure 9. The proposed model is used to optimize the adjusted signal timing. The running time in MATLAB is 1.72 s. The optimization results include the extension time for the green light period , , , . The optimized signal timing is presented in Figure 10.
The performance of the proposed model is compared with two traditional methods, namely, traditional methods A and B as shown in Tables 3–5. The traditional method A is the method which distributes redlight time to the following phase according to the phase splits. The traditional method B is the method which resplits the green light time period according to the degree of saturation.
 
Note: M = movement. 
 
Note: M = movement. 

As Table 3 shows, the queue length of movement will be restored to zero in two cycles. As Table 4 shows, the largest queue length of movement is less than 150 m, which is less than the road length. Thus, the proposed method is suitable for this intersection. Moreover, the proposed method outperforms the traditional methods in terms of the recovery time and queue length. All the movements can be recovered within 2 cycles for the proposed method, while it requires 3 cycles for recovery. The average and longest queue length for the proposed method is shortened by 1.7% and 3.4%, respectively, with the comparison of the traditional methods.
Table 5 compares the three methods from the perspective of the delay variation in three signal cycles. As Table 5 shows, the proposed method can reduce the total passenger delay and the percentage of delay decrement by 4.5%. Conversely, the traditional method will increase the total passenger delay. In addition, the percentage of the autodelay increment is more than 9.1% after the optimization of the traditional method, which is larger than 3.8%.
4. Sensitivity Analysis
A sensitivity analysis was conducted in order to analyze the influence of the intersection’s geometry and traffic conditions on the efficiency of the proposed model. The length of the priority time for buses, the intersection saturation, and the number of lanes were selected as the three key parameters for the sensitivity analysis. The value of the input parameters are as follows: in Figures 11, 12, and 14, the cycle length is 80 s, there are four phases and the time of each phase is 20 s, the arrive rate of each movement is the same, which is from 315 to 472 veh/h depending on the saturation, the saturation rate of each movement is 1800 veh/h, the average passenger occupancy of private vehicles is 1.5 per/veh, the passenger occupancy of buses is 20 per/veh, and the number of lanes is three. In Figure 13, the arrival rate of each movement is 382 veh/h, and other parameters are the same as above.
In order to avoid the irrationality of the basic signal timing results to unreliable model results, the signal timing and the intersection conditions are set as above described, the signal timing has been set to be optimal. Figure 11 shows the relationship between the delay increment and the three key parameters. Figure 11 is composed of 9 subfigures. The ordinates and abscissa of Figure 11 are priority time and number of lanes, respectively. The ordinates and abscissa of each subfigure are saturation and delay increment, respectively. The nine subfigures of Figure 9 are arranged such that the priority time gradually increases from the bottom to the top, and the number of lanes gradually increases from the left to the right.
From the perspective of each of these subfigures, the vehicular delay increases as the saturation increases; however, the delay of the bus will not change. Additionally, comparison of these subfigures from the bottom to the top indicates that the vehicular delay increases as the priority increases. In contrast, the bus delay decrement becomes smaller. Comparison of the data in the subfigures from the left to the right indicates that the vehicular delay increases as the number of lanes increase, while the bus delay does not change. The reason for which the vehicular delay increases as the number of lanes increases is attributed to the fact that as the number of lanes increases, the number of vehicles will increase if the saturation of the intersection remains constant.
Figure 12 further shows the combination effects of the saturation of intersection and the priority time. As indicated, the total delay will increase as the priority time and the saturation increase. The delay increment increases rapidly and becomes very large as the saturation of intersection becomes larger than 0.9. Similarly, the delay increment will increase rapidly and will become very large as the priority time becomes larger than 10 s.
As Figure 12 shows, the delay increment is sensitive to the priority time. Thus, a better optimization model is needed which is less sensitive to the priority time. As Figure 13 shows, the two traditional methods get the same results, the reason is that the basic signal timing is optimal, the splits and degree of saturation of each phase is the same. Figure 13 compares the proposed model and the traditional model, based on which several inferences can be drawn: (1) the proposed model is less sensitive than the traditional model to the priority time; (2) the difference of the effect of optimization is not obvious between the proposed model and traditional method when the priority time is less than 5 s, and (3) when the priority time is larger than 5 s, the difference of the effects of the optimization between the two methods will increase, as the priority time will increase. When the priority time becomes larger than 9 s, the proposed model is 15% better than the traditional models A and B.
Figure 14 presents a table listing the recommended values for bus signal priorities, which can provide suggestions for practical applications. As Figure 14 shows, the percentages in this figure are the percentage reductions of the total passenger delays at the intersection when bus signal priorities are assigned. Figure 14 contains three parts, namely, the green part that is suitable for application of the bus signal priority strategy, the red part that is unsuitable for application of the bus signal priority strategy, and the yellow part that indicates that the percentage reduction of the total delay is larger than zero but less than 10%. In the yellow highlighted part, the effect of the bus signal priority strategy is minor.
5. Conclusions
This paper presented a signal control model for assigning bus signal priorities that also minimized the total passenger delay increment at intersections with the nearsaturation. The proposed model was tested with simulations at a nearsaturated intersection in Shanghai, China. Several meaningful conclusions can be drawn from the sensitivity analyses:
(1) Bus signal priority can be applied in nearsaturated intersections, whereby the length of priority time, the saturation of the intersection, and the number of lanes were the three main factors that influenced whether the intersections should have adopted the bus signal priority strategy.
(2) The method on resplitting the green light time period affected the benefits of the bus signal priority considerably. Simple distribution of the added redlight time to the following phases according to the phase splits or resplitting the green light time period according to the degree of saturation was shown to be unsuitable.
(3) A table with recommended bus priority values was presented according to the saturation of the intersection and the length of the priority time. More than 6% reductions in person delay can be obtained for undersaturated intersections when the priority time was less than 5 s. Moreover, even when the intersection saturation was 0.95, the bus signal priority could be applied if only the priority time was less than 5 s.
The signal control model proposed in this study can be applied in nearsaturated intersection. However, the issue on the multiple bus requests was not discussed and constitutes a research topic for our ongoing and future research endeavors.
Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.
Conflicts of Interest
The authors declare that they have no conflicts of interest.
Acknowledgments
The research is supported by the National Natural Science Foundation of China under Grant no. 51608324.
References
 J. Shi, Z. Wu, and J. Jin, “Reform Beijing to a public transit oriented city  From the view of transportation equity,” Journal of Advanced Transportation, vol. 45, no. 2, pp. 96–106, 2011. View at: Publisher Site  Google Scholar
 K. M. N. Habib, L. Kattan, and T. Islam, “Model of personal attitudes towards transit service quality,” Journal of Advanced Transportation, vol. 45, no. 4, pp. 271–285, 2011. View at: Publisher Site  Google Scholar
 M. Eichler and C. F. Daganzo, “Bus lanes with intermittent priority: strategy formulae and an evaluation,” Transportation Research Part B: Methodological, vol. 40, no. 9, pp. 731–744, 2006. View at: Publisher Site  Google Scholar
 W. J. Ma, K. L. Head, and Y. H. Feng, “Integrated optimization of transit priority operation at isolated intersections: a personcapacitybased approach,” Transportation Research Part C: Emerging Technologies, vol. 40, pp. 49–62, 2014. View at: Publisher Site  Google Scholar
 M. Mesbah, M. Sarvi, and G. Currie, “New methodology for optimizing transit priority at the network level,” Transportation Research Record, vol. 2089, no. 12, pp. 93–100, 2008. View at: Google Scholar
 S. I. Guler and M. J. Cassidy, “Strategies for sharing bottleneck capacity among buses and cars,” Transportation Research Part B: Methodological, vol. 46, no. 10, pp. 1334–1345, 2012. View at: Publisher Site  Google Scholar
 A. H. Alomari, H. AlDeek, A. Sandt, J. H. Rogers, and O. Hussain, “Regional evaluation of bus rapid transit with and without transit signal priority,” Transportation Research Record, vol. 2554, no. 6, pp. 46–59, 2016. View at: Publisher Site  Google Scholar
 K. N. Balke, C. L. Dudek, and T. Urbanik II, “Development and evaluation of intelligent bus priority concept,” Transportation Research Record, vol. 1727, no. 1543, pp. 12–19, 2000. View at: Google Scholar
 A. Skabardonis, “Control strategies for transit priority,” Transportation Research Record, vol. 1727, no. 1584, pp. 20–26, 2000. View at: Google Scholar
 M. Estrada, C. Trapote, M. RocaRiu, and F. Robusté, “Improving bus travel times with passive traffic signal coordination,” Transportation Research Record, vol. 2111, no. 9, pp. 68–75, 2009. View at: Google Scholar
 J. Lee, A. Shalaby, J. Greenough, M. Bowie, and S. Hung, “Advanced transit signal priority control with online microsimulationbased transit prediction model,” Transportation Research Record, vol. 1925, no. 1, pp. 185–194, 2005. View at: Google Scholar
 V. Ngan, T. Sayed, and A. Abdelfatah, “Impacts of various parameters on transit signal priority effectiveness,” Journal of Public Transportation, vol. 7, no. 3, pp. 71–93, 2004. View at: Publisher Site  Google Scholar
 J. Jacobson and Y. Sheffi, “Analytical model of traffic delays under bus signal preemption: theory and application,” Transportation Research Part B: Methodological, vol. 15, no. 2, pp. 127–138, 1981. View at: Publisher Site  Google Scholar
 P. G. Furth and T. H. J. Muller, “Conditional bus priority at signalized intersections: better service with less traffic disruption,” Transportation Research Record, vol. 1731, no. 1, pp. 23–30, 2000. View at: Google Scholar
 Z. R. Abdy and B. R. Hellinga, “Analytical method for estimating the impact of transit signal priority on vehicle delay,” Journal of Transportation Engineering, vol. 137, no. 8, pp. 589–600, 2011. View at: Publisher Site  Google Scholar
 K. Ling and A. Shalaby, “Automated transit headway control via adaptive signal priority,” Journal of Advanced Transportation, vol. 38, no. 1, pp. 45–67, 2003. View at: Publisher Site  Google Scholar
 E. Christofa and A. Skabardonis, “Traffic signal optimization with application of transit signal priority to an isolated intersection,” Transportation Research Record, vol. 2259, no. 1, pp. 192–201, 2011. View at: Publisher Site  Google Scholar
 L. Head, D. Gettman, and W. Zhiping, “Decision model for priority control of traffic signals,” Transportation Research Record, vol. 1978, no. 1, pp. 169–177, 2006. View at: Google Scholar
 W. Ma, X. Yang, and Y. Liu, “Development and evaluation of a coordinated and conditional bus priority approach,” Transportation Research Record, vol. 2145, no. 6, pp. 49–58, 2010. View at: Publisher Site  Google Scholar
 E. Christofa, K. Ampountolas, and A. Skabardonis, “Arterial traffic signal optimization: a personbased approach,” Transportation Research Part C: Emerging Technologies, vol. 66, pp. 27–47, 2016. View at: Publisher Site  Google Scholar
 W. Ma, Y. Liu, and X. Yang, “A dynamic programming approach for optimal signal priority control upon multiple highfrequency bus requests,” Journal of Intelligent Transportation Systems: Technology, Planning, and Operations, vol. 17, no. 4, pp. 282–293, 2013. View at: Publisher Site  Google Scholar
 J. Hu, B. B. Park, and Y.J. Lee, “Coordinated transit signal priority supporting transit progression under connected vehicle technology,” Transportation Research Part C: Emerging Technologies, vol. 55, pp. 393–408, 2015. View at: Publisher Site  Google Scholar
 M. Li, Y. Yin, W.B. Zhang, K. Zhou, and H. Nakamura, “Modeling and implementation of adaptive transit signal priority on actuated control systems,” ComputerAided Civil and Infrastructure Engineering, vol. 26, no. 4, pp. 270–284, 2011. View at: Publisher Site  Google Scholar
 Q. He, K. L. Head, and J. Ding, “Multimodal traffic signal control with priority, signal actuation and coordination,” Transportation Research Part C: Emerging Technologies, vol. 46, pp. 65–82, 2014. View at: Publisher Site  Google Scholar
 Y. J. Lin, X. F. Yang, G.L. Chang, and N. Zou, “Transit priority strategies for multiple routes under headwaybased operations,” Transportation Research Record, vol. 2356, no. 5, pp. 34–43, 2013. View at: Publisher Site  Google Scholar
 S. Z. Altun and P. G. Furth, “Scheduling buses to take advantage of transit signal priority,” Transportation Research Record, vol. 2111, no. 7, pp. 50–59, 2009. View at: Google Scholar
 S. I. Guler, V. V. Gayah, and M. Menendez, “Bus priority at signalized intersections with singlelane approaches: A novel presignal strategy,” Transportation Research Part C: Emerging Technologies, vol. 63, pp. 51–70, 2016. View at: Publisher Site  Google Scholar
 J. Stevanovic, A. Stevanovic, P. T. Martin, and T. Bauer, “Stochastic optimization of traffic control and transit priority settings in VISSIM,” Transportation Research Part C: Emerging Technologies, vol. 16, no. 3, pp. 332–349, 2008. View at: Publisher Site  Google Scholar
 W. Ma, Y. Liu, and B. Han, “A rulebased model for integrated operation of bus priority signal timings and traveling speed,” Journal of Advanced Transportation, vol. 47, no. 3, pp. 369–383, 2013. View at: Publisher Site  Google Scholar
 F. Dion and B. Hellinga, “A rulebased realtime traffic responsive signal control system with transit priority: application to an isolated intersection,” Transportation Research Part B: Methodological, vol. 36, no. 4, pp. 325–343, 2002. View at: Publisher Site  Google Scholar
 M. Janos and P. G. Furth, “Bus priority with highly interruptible traffic signal control: simulation of San Juan's Avenida Ponce de Leon,” Transportation Research Record, vol. 1811, no. 1, pp. 157–165, 2002. View at: Google Scholar
 J. Hu, B. Park, and A. E. Parkany, “Transit signal priority with connected vehicle technology,” Transportation Research Record, vol. 2418, no. 3, pp. 20–29, 2014. View at: Publisher Site  Google Scholar
 J. Li, W. Wang, H. J. van Zuylen, N. N. Sze, X. W. Chen, and H. Wang, “Predictive strategy for transit signal priority at fixedtime signalized intersections case study in Nanjing, China,” Transportation Research Record, vol. 2311, no. 12, pp. 124–131, 2012. View at: Google Scholar
 Transportation Research Board, “Highway capacity manual 2010. Washington D C: National Research Council,” 2010. View at: Google Scholar
Copyright
Copyright © 2018 Shijie Shu et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.