OD-Based Partition Technique to Improve Arterial Signal Coordination Using Connected Vehicle Data

Abstract

Maximizing two-way arterial bands often plays a critical role in signal coordination along urban arterials. With an increasing number of signals involved, the arterial bands tend to shrink to some extent resulting in inefficient arterial signal coordination, and not all vehicles fully take advantage of the arterial bands to travel through the entire corridor. In response to that, system partition is a technique for handling arterials with many signals. Rather than designing end-to-end signal coordination, efficient arterial signal coordination is highly reliant on traffic origin and destination (OD) patterns on the arterial, which have been difficult to obtain using conventional data collection methods. The emerging big data sources, such as connected vehicles, provide great potential to gather such invaluable OD information for improving arterial signal coordination. This research proposes an easy-to-implement OD-based partition technique to improve arterial signal coordination by utilizing vehicle trajectory data automatically collected from connected vehicles. The proposed signal timing technique was tested using an arterial with 17 signalized intersections in Orange County, California. The results demonstrated that the OD-based partition technique improved the arterial average travel speed by 2.7% and 12.1% for the eastbound and westbound directions, respectively. At the same time, the proposed technique shortened the arterial average travel time by 2.6% and 11.1% for the eastbound and westbound directions, respectively. The total travel time was shortened both for the main-street and side-street major traffic flows.

Keywords

operations traffic signal systems arterial signal system signal timing signalized intersection traffic signal

Signal coordination has proven to be a cost-effective arterial management strategy to smooth traffic flows by minimizing travel times and the number of stops. To achieve ideal signal progression, maximizing two-way arterial bandwidths has been acknowledged as a popular approach among researchers and practitioners. In reality, arterial bandwidths tend to reduce with an increasing number of signals and maximizing arterial bandwidths only may not guarantee efficient performance for all turning flows because of unknown arterial band utilizations. When dealing with such arterials with many signals, the system partition technique can serve as an effective signal timing strategy, primarily because a vehicle may only traverse a small segment of the entire arterial. Knowing arterial traffic OD patterns assists in partitioning the arterial, but it has been extremely difficult to obtain such OD patterns using conventional data collection methods. The emergence of connected vehicles gives birth to prolific connected vehicle data, which provides a promising source for obtaining vehicle OD information on the arterial. The objective of this research is to develop an easy-to-implement OD-based partition technique to improve arterial signal coordination by utilizing vehicle trajectory data generated from connected vehicles.

The remaining paper is organized as follows: after the introduction, the literature review is presented. Then, commonly used criteria for arterial partition are discussed and the procedure of OD-based partition technique using connected vehicle data is presented next, followed by a case study. Finally, conclusions and discussions are summarized from this research.

Literature Review

Bandwidth-Based Optimization Models

To begin with, several pioneering pieces of research focused on modeling the bandwidth optimization problem with maximizing arterial bandwidths as the objective function, such that the optimal signal coordination parameters, including the optimal cycle length, offsets, and phase sequence could be obtained by solving the model ( 1 – 3 ). In 1966, Little et al. ( 4 ) developed MAXBAND, an extended program to maximize the weighted combination of bandwidths by producing the optimal cycle time, offsets, speeds, and phase sequence. Because of optimization limitations of MAXBAND, Chang et al. ( 5 ) explored the significance of left-turn phase sequence on multi-arterial closed networks and proposed an enhanced program, MAXBAND-86. Considering actual traffic volumes on arterial links, Gartner et al. ( 6 ) developed an enhanced model, MULTIBAND, to optimize volume-weighted bandwidth, and this provided a wider range of design options for traffic practitioners. By incorporating variable-bandwidth progression, Stamatiadis and Gartner ( 7 ) extended MULTIBAND to MULTIBAND-96 to optimize multi-arterial grid networks.

System Partition Technique in Signal Coordination Optimization

Although numerous mathematical optimization models have been proposed to optimize arterial signal coordination, that arterial bandwidths tend to shrink with an increasing number of signals does hold true in engineering practice. To maintain ideal signal progression on a long corridor, Tian and Urbanik ( 8 ) proposed a system partition technique to maximize two-way signal progression by dividing a large system into several subsystems and maximizing signal progression for each subsystem to improve arterial progression. The application of the partition technique opened a new era for designing arterial signal coordination. Fan and Tian ( 9 ) compared and evaluated three different network partition techniques: coupling index, strength of attraction, and coordinatability factor. They found that the three techniques functioned similarly and could produce the same solutions in all scenarios. Clustering analysis, such as hierarchical distance-based clustering ( 10 ) and K-clustering ( 11 ), was applied to partition an arterial to improve operational performance. In addition, a lot of research centered on proposing mathematical models to partition an arterial into multiple subsystems to achieve enhanced arterial signal progression ( 12 – 14 ). Mixed-integer linear programming (MILP) was widely used to formulate the model for optimization ( 15 , 16 ). Rather than considering main-street signal progression only, prioritization of major turning traffic flows from and to cross streets on the arterial ( 17 – 19 ) and multi-arterial network ( 20 ) was greatly emphasized to produce efficient arterial signal coordination.

Connected Vehicle Applications in Traffic Signal Control

Along with bandwidth-based optimization models and partition techniques used for arterial signal coordination, the emergence of connected vehicle technology fuels potential applications in traffic signal control. In 2011, AASHTO presented its view that connected vehicle technology, combined with on-board equipment, roadside network service, and in-vehicle systems, provided a wide range of opportunities in adaptive signal control, traffic signal prioritization, and arterial network signal coordination ( 21 ). By leveraging connected vehicle data, existing studies were conducted to estimate traffic volumes ( 22 ), detect queue spillback ( 23 ), optimize intersection offsets by extracting arrival profiles ( 24 ), perform adaptive signal control ( 25 , 26 ), design network-level signal coordination ( 27 ), and benefit transit signal priority control ( 28 ) under the low penetration rate of the connected vehicle data. Argote-Cabanero et al. developed estimation methods for various measures of effectiveness (MOEs) and proposed a methodology to determine the minimum connected vehicle penetration rate for accurate MOE estimation. They found that 15% penetration rate was appropriate for use ( 29 ).

Contribution of This Research

As referenced above, partition technique is functioning as a heuristic approach for maximizing two-way signal progression especially on a signalized arterial with many signals. Meanwhile, booming development of connected vehicle technology potentially provides revolutionary solutions to enhance traffic signal coordination. In reality, not every vehicle travels the entire arterial with a great number of signals and traffic OD information plays an essential role in revealing arterial flow patterns, which provides invaluable information for arterial partitioning to design more effective arterial signal coordination. In light of that, this paper aims to propose an easy-to-use OD-based partition technique to improve arterial signal coordination by leveraging vehicle trajectory data generated from connected vehicles.

Commonly Used Criteria for Arterial Partition

It is not uncommon to know that arterial bandwidths shrink with an increasing number of signals in an arterial system, which reduces potential opportunities for vehicles to travel through the arterial and leads to inefficient arterial signal coordination. System partition, an effective signal timing strategy, divides a signalized long arterial into several subsystems and maximizes bandwidths within each subsystem. Band connections are built between each subsystem to achieve the overall enhanced signal progression. Several commonly used partition criteria and their applications are demonstrated in Table 1.

Table 1.

Commonly Used Arterial Partition Criteria

Partition criteria	Applications
Land use type	Intersections including residential, recreational, and commercial land use types become potential candidates for partition points.
Intersection distance	An empirical value of 0.5 mi (2640 ft) is used to coordinate adjacent signals or as the partition boundary.
Traffic volumes	Intersections with heavy left-turn traffic both on the main street and on side streets can be selected as partition points to regroup traffic flows.
Degree of saturation	Intersections with a high degree of saturation are potential partition points to avoid queue spillback.
Origin and destination data	Starting from the main-street through movement at the first intersection, intersections with continuously heavy left-turn flows are supposed to be grouped and the partition point occurs at the following one with obvious trip disappearance.

Land Use Type

The destination of an individual trip is highly associated with travel activities and such activities mostly occur in residential, recreational, and commercial areas. In other words, intersections including the above land use types are more likely to be major destinations for most trips. Thus, on a signalized arterial, it is reasonable to smooth traffic platoons by developing signal coordination up to those intersections. However, such a partition criterion is highly reliant on empirical engineering judgment.

Distance

In practice, adjacent signalized intersections within 0.5 mi are usually recommended to be coordinated in the same arterial system. On long arterials, platoon dispersion may exert unexpected impacts, such as vehicles falling out of the green band because of speed variations, vehicle interactions, and roadside interferences. The empirical value, 0.5 mi, can be applied as a threshold for arterial partition but such a single factor cannot accurately determine partition points without considering traffic flow patterns and traffic operations. It becomes invalid when adjacent intersections are close to each other on a long arterial.

Volume

Intersections with heavy left-turn flows both on the main street and on side streets can be regarded as partition boundaries to regroup traffic platoons. However, traffic turning volumes mostly indicate traffic demands at each individual intersection rather than reflecting the arterial traffic OD patterns because of innumerable possible traffic OD combinations from upstream intersections. Additionally, inappropriate detector layouts in the field may become a barrier for accurate volume counting. Alternatively, it is a costly and laborious task to hire workers to collect traffic turning volumes on a signalized arterial with many signals at the same time during various periods of the day.

Degree of Saturation

To avoid spillback caused by queueing vehicles, intersections with a high degree of saturation are inclined to be partition points when developing signal coordination. Similar to the previous criterion, not only do volume collections present a great challenge, but the degree of saturation does not reflect traffic flow patterns on the arterial.

Origin and Destination Data

Different from all of the above criteria, OD data is able to provide a comprehensive view of the traffic flow patterns in an arterial system, but yet is regarded as one of the most challenging sorts of data to collect. On a signalized arterial, starting from the through movement at the first intersection, intersections with continuously heavy left-turn flows on the main street should be grouped and the partition point occurs at the following one with obvious trip disappearance. Signal coordination should be developed up to those critical intersections and traffic flows regrouped to the next subsystem. This research fully utilizes traffic OD information extracted from vehicle trajectory data, which was generated from connected vehicles for arterial partition.

OD-Based Partition Technique

This research proposes an easy-to-implement OD-based partition technique, shown in Figure 1, to improve arterial signal coordination integrated with connected vehicle data.

Figure 1.

Flow chart of OD-based partition technique.

Step 1: Arterial Locating

To locate the signalized arterial with two end intersections becomes the initial step, which paves the way for filtering vehicle trajectory data on the arterial based on the geographical information.

Step 2: OD Extractions from Connected Vehicle Trajectory Data

Once the arterial is determined, the next step is to obtain the OD information from connected vehicle trajectory data during a specific period of the day, such as AM (morning peak) and MD (midday). Vehicle trajectory data used in this research was provided by Ticon International, a leading data service company. Table 2 presents the aggregated information of one vehicle trajectory point and it demonstrates that the vehicle of assigned ID Ticon77d558 was captured at 5:02:27 a.m. on July 15, 2019 at the location with latitude 33.687311 and longitude −117.919556 (within the range of City of La Habra in California). Its instant speed was 35.52 mph and the travel direction was eastbound (EB) (heading refers to the angle moving clockwise from the north direction). The vehicle’s engine was running at that moment.

Table 2.

Sample Vehicle Trajectory Data Format

Property	Value
Vehicle ID	Ticon77d558
Time	2019-07-15T05:02:27
Latitude	33.687311
Longitude	−117.919556
Speed (mph)	35.52
Heading	90
City	La Habra
State	California
Engine status	Running

OD information is represented by left-turn traffic volumes starting from the through movement at the first intersection to each intersection on the main street. As depicted in Figure 2, $N_{MLT i}$ and $N_{MLT 1}^{'}$ represent main-street left-turn traffic volumes for both directions starting from the through movements at the intersection 1(i) to intersection i (1) (i = 1,2,3…). $N_{MTH i}$ and $N_{MTH 1}^{'}$ represent traffic flows traveling the entire arterial for both directions, respectively. $N_{SLT i}$ and $N_{SLT i}^{'}$ stand for side-street left-turn volumes for both directions at intersection i. Distance between adjacent intersections is denoted by $D_{i - 1 i}$ . When processing the vehicle trajectory data, $N_{MLT i}$ , $N_{MLT i}^{'}$ , $N_{MTH i}$ , $N_{MTH 1}^{'}$ , $N_{SLT i}$ , and $N_{SLT i}^{'}$ need to be extracted for the following steps.

Figure 2.

OD extractions from vehicle trajectory data.

Step 3: Arterial Partition

Based on the extracted $N_{MLT i}$ and $N_{MLT i}^{'}$ , intersections with continuously noticeable left-turn flows should be grouped and, at the following intersection with obvious left-turn trip disappearance, partitioned separately for both directions. To eliminate the negative impacts caused by platoon dispersion, the next intersection should be automatically included in another subsystem if $D_{i - 1 i}$ exceeds the empirical value, 0.5 mi. Results for both directions should then be partitioned and adjustments made to finalize the arterial partition. The details of this step will be discussed in the case study.

Step 4: Subsystem Signal Coordination Design

According to the partitioned subsystems, signal coordination is then designed to maximize two-way signal progression within each subsystem. Then offsets and phase sequence are adjusted at boundary intersections to achieve ideal signal progression between each subsystem. This process is similar to what had been proposed by Tian and Urbanik ( 8 ). In this research, the same cycle length (or multiples of the cycle length) was applied to each subsystem to maintain signal progression.

Step 5: Subsystem Side-Street Signal Timing Adjustment

The extracted $N_{SLT i}$ and $N_{SLT i}^{'}$ are able to provide invaluable insights into side-street left-turn traffic demands at each intersection, such that performance of major left-turn flows can be improved by fine-tuning intersection offsets and adjusting side-street phase sequence. A phase sequence change on the side-street has the possibility of creating potential left-turn bands onto the arterial and benefits the side-street platoon by them slipping into designed bands on the main street.

Case Study

Applications of OD-Based Partition Technique

The proposed partition technique was tested on a signalized arterial located in City of La Habra, California. It is a 3.72 mi segment with 17 signalized intersections on West Imperial Highway from Brass Lantern Drive to North Berry Street. Figure 3 shows the layout of the arterial. This location was selected because of an ongoing signal retiming project where traffic volumes and vehicle trajectory data from connected vehicles were already available to the research.

Figure 3.

West Imperial Highway signalized arterial.

Step 1: Arterial Locating

The east–west signalized arterial, West Imperial Highway from Brass Lantern Drive to North Berry Street, was located with the longitude ranging from −117.972200 to −117.905958 and the latitude ranging from 33.917182 to 33.917582.

Step 2: OD Extractions from Connected Vehicle Trajectory Data

Ticon International high-resolution data covers 95% of roads in the US and transmits every 1 to 3 s, providing up to 15% to 20% penetration rates. Vehicle OD information on the main-street and side-street left-turn volumes was extracted and averaged during the MD period (9:30–15:30) from July 15 to July 19, 2019 (Figure 4). In Figure 4a, main-street left-turn volumes starting from Brass Lantern Drive and North Berry Street to each intersection for both directions are indicated by left-turn arrows connected by dashed lines. No left turn is permitted at EB Village Dr and WB (westbound) Bonita Place in the field. In Figure 4b, side-street left-turn volumes for both directions at each intersection are demonstrated by left-turn arrows as well. Although the numbers only indicate a small portion of the overall traffic volumes, they still serve as a good representation of the actual traffic OD patterns.

Figure 4.

OD extractions on West Imperial Highway: (a) main-street OD information and (b) side-street left-turn volumes.

Step 3: Arterial Partition

For the EB direction starting from Brass Lantern Drive, $N_{MLT i}$ decreased sharply at Market Place while rebounding at South Idaho Street. Considering the platoon dispersion caused by the long distance of 3962 ft between South Idaho Street and South Euclid Street, the first partition point was selected at South Idaho Street for EB. The EB left-turn vehicle demands did not show obvious fluctuations for the remaining intersections except at South Harbor Boulevard with 11 vehicles. South Harbor Boulevard may become a potential partition point. After South Euclid Street, most of the remaining vehicles traveled through North Berry Street.

For the WB direction starting from North Berry Street, the first partition point was selected at South Puente Street based on the long distance of 2631 ft between Palm Street and South Puente Street. $N_{MLT i}^{'}$ tended to increase until South Harbor Boulevard and then declined dramatically at Walmart. As a consequence, South Harbor Boulevard served as the second partition point for WB. From Walmart to South Euclid Street, a good signal progression was needed to serve the heavy WB left-turn flows up to South Euclid Street. Logically, South Euclid Street became the third partition point. After South Euclid Street, it could be seen that major platoons traveled through the remaining intersections.

In light of considerations of the potential partition points for both directions, the entire arterial was divided into four subsystems: Brass Lantern Drive to South Idaho Street (five intersections), South Euclid Street to Walmart (five intersections), South Harbor Boulevard to Palm Street (four intersections) and South Puente Street to North Berry Street (three intersections). These are depicted in Figure 5.

Figure 5.

Partitioned subsystems on West Imperial Highway.

Step 4: Subsystem Signal Coordination Design

Once the subsystems have been defined, it is of the essence to maximize signal progression within each subsystem and build band connections between them to achieve arterial signal coordination. While numerous traffic signal timing packages have become available, TranSync was used in this research because of its powerful bandwidth optimization and easy visualization of the timing plans ( 30 ). During the MD period, West Imperial Highway was running coordination with the common cycle length of 130 s except at Brass Lantern Drive and Village Drive which had 65 s. The designed progression speed was 45 mph for both directions based on posted speed limits in the field.

Figure 6 depicts time–space diagrams (TSD) of the designed signal coordination for each subsystem. In the TSD, the solid green and blue bars indicate phase splits of main-street through phase for EB and WB, respectively. Downward hatched green and upward hatched blue bars represent phase splits of the main-street left-turn phases for EB and WB accordingly. Red bars display the side-street timings. The solid green and blue parallelograms represent the EB and WB link bands between each intersection and arterial bands within each subsystem are outlined by two white parallel lines. The TSD of the combined arterial signal coordination is illustrated in Figure 7.

Figure 6.

Time–space diagrams of signal coordination design for the four subsystems: (a) subsystem 1, (b) subsystem 2, (c) subsystem 3, and (d) subsystem 4.

Figure 7.

The time–space diagram of arterial signal coordination.

Step 5: Subsystem Side-Street Signal Timing Adjustment

Utilizing the extracted $N_{SLT i}$ and $N_{SLT i}^{'}$ , intersections with heavy side-street left-turn flows should be prioritized to minimize their delay and number of stops by adjusting side-street phase sequence and intersection offsets to create left-turn progression bands onto the main street. According to the processed data, side-street left-turn flows exceeding 10 vehicles were considered and existed at the following intersections: both northbound (NB) and southbound (SB) left-turns at South Beach Boulevard, Hills Drive, Market Place, South Idaho Street, South Euclid Street, and South Harbor Boulevard; SB left turn at Brass Lantern Drive, South Cypress Street, Lakeview Avenue, Palm Street, South Puente Street, and North Berry Street; and NB left turn at Walmart.

The feature of displaying side-street left-turn bands (light green and light blue parallelograms for SB left turn and NB left turn, respectively) in TranSync allowed easy visualization of the left-turn progression, which can be achieved by adjusting the side-street phasing sequence and fine-tuning intersection offsets. Figure 8 demonstrates TSDs of side-street left-turn bands within each subsystem.

Figure 8.

Side-street left-turn bands in each subsystem: (a) subsystem 1, (b) subsystem 2, (c) subsystem 3, and (d) subsystem 4.

Simulation Analysis

To evaluate the effectiveness of the proposed partition technique integrated with connected vehicle data, another arterial-band-based timing plan, which guarantees two-way arterial bands traveling through the entire arterial, was produced by optimization functions in TranSync. Figure 9 shows TSDs of the mentioned timing plans with Figure 9a produced by OD-based partition and Figure 9b being the arterial band-based plan. It is noticeable that two-way 10 s arterial bands, depicted by two white parallel lines for both directions, exist in Figure 9b while there are no arterial bands for the proposed OD-based partition timing plan (Figure 9a).

Figure 9.

Comparison of time–space diagrams of two timing plans: (a) OD-based partition timing plan and (b) arterial band-based timing plan.

The effectiveness of the two timing plans was evaluated using traffic simulation for two main reasons: (1) the managing agency would not allow testing the timing plans in the field; (2) the expected MOEs could only feasibly be obtained from simulations. TransModeler was selected as the simulation tool because of its ease of use and capability of producing the desired MOEs. To ensure validity of the analysis, the authors did go through a basic model calibration process, by comparing the trajectory data from both simulation and Ticon International based on the timing plan, when vehicle trajectories were gathered. When evaluating two timing plans, all calibrated parameters were kept the same; thus the results were still valid in a relative sense.

In addition to the basic signal timing parameters such as cycle length, offset, and phase splits, minimum recall was applied to intersections with heavy main-street leading left-turn phases under the OD-based partition timing plan to avoid a phase gapping out. The intersections were EB left turn at South Beach Boulevard, Market Place, South Idaho Street, and South Harbor Boulevard; WB left turn at Hills Drive, South Idaho Street, Lakeview Avenue, Village Drive, Mervyn Drive, Palm Street, and South Puente Street. The minimum green time followed the existing value of 10 s for the above left-turn phases.

Batch simulation in TransModeler was conducted with a total of 10 runs under various random seeds for each designed plan. Each run was simulated for 1 h.

Simulation Results

Figure 10 shows simulation results by comparing average travel speed and average travel time of vehicles traveling the entire arterial for both directions under the two timing plans. It was found that the arterial average travel speed under the OD-based partition timing plan outperformed the one under the arterial-band-based timing plan. More specifically, the average speed for the EB and WB increased by 2.7% and 12.1%, respectively when the OD-based partition timing plan was implemented. Meanwhile, the arterial average travel time was reduced by 2.6% and 11.1% for the EB and WB, respectively, under the OD-based partition timing plan.

Figure 10.

Arterial-level MOE comparison: (a) arterial average travel speed and (b) arterial average travel time.

Several vehicle trajectories traveling the entire arterial for both directions under the two timing plans were extracted from simulations and imported into TranSync for validating the result. Vehicle stops were identified when the instant speed was less than 5 mph and highlighted as red points on each trajectory. From Figure 11a, it could be seen that EB vehicles traveling the entire arterial fell into the designed arterial band and some of them experienced stops at South Beach Boulevard under the arterial-band-based timing plan. Under the OD-based partition timing plan, vehicles experienced an extra short stop at Bonita Place but drove within the band as well in each subsystem as expected. Different from EB, Figure 12 demonstrates a sharp contrast for WB vehicles traveling the entire arterial under the two timing plans: Although there existed a 10 s arterial band for the WB, uncertainty of early return from side streets and speed fluctuations caused vehicles to fall out of the designed arterial band and stop multiple times under the arterial-band-based timing plan. By contrast, the OD-based partition timing plan produced fewer stops and reduced travel times for vehicles because of wider bands within each subsystem and smooth band connections between each subsystem to minimize negative impacts caused by speed fluctuations.

Figure 11.

Comparison of eastbound vehicle trajectories traveling the entire arterial: (a) arterial band-based timing plan and (b) OD-based partition timing plan.

Figure 12.

Comparison of westbound vehicle trajectories traveling the entire arterial: (a) arterial band-based timing plan and (b) OD-based partition timing plan.

To comprehensively measure the quality of designed signal coordination, total travel time—the product of average travel time and the number of vehicles—of main-street left-turn flows starting from the through movement at the first intersection (both EB and WB), was considered at intersections with major left-turn flows (Table 3). For example, the EB average travel time and the number of vehicles starting from the through movement at Brass Lantern Drive and turning left to South Beach Boulevard were 2.12 min/veh (minutes per vehicle) and 116 veh (vehicles) under the arterial-band-based timing plan, respectively. The total travel time was calculated to be 245.92 (min) = 2.12 (min/veh) * 116 (veh). Following the same process, the sums of the total travel times of listed intersections for both directions are displayed at the bottom of the Table 3 sections—for example, 2192.70 min for EB and 4233.73 min for WB under the arterial band-based timing plan. It should be noted that the average travel time and number of vehicles shown at North Berry Street (EB) and Brass Lantern Drive (WB) represent the vehicles traveling the entire arterial.

Table 3.

Measures of Effectiveness Comparison of Major Left-Turn Flows on the Main Street

Destination intersection	Arterial band-based timing plan			OD-based partition timing plan
Destination intersection	Average travel time (min/veh)	Number of vehicles (veh)	Total travel time (min)	Average travel time (min/veh)	Number of vehicles (veh)	Total travel time (min)
Eastbound
S Beach Blvd	2.12	116	245.92	2.34	116	271.44
Hills Dr	3.30	88	290.40	2.56	90	230.40
Market Place	4.34	30	130.20	3.26	30	97.80
S Idaho St	3.60	54	194.40	4.52	52	235.04
S Harbor Blvd	6.83	31	211.73	6.74	31	208.94
N Berry St	8.55	131	1120.05	8.33	133	1107.89
Total	na	na	2192.70	na	na	2151.51
Westbound
Arovista Ave	1.69	141	238.29	1.69	143	241.67
S Puente St	2.38	78	185.64	2.35	79	185.65
Palm St	2.95	11	32.45	2.72	10	27.20
Mervyn Dr	4.28	17	72.76	4.23	17	71.91
S Harbor Blvd	4.60	211	970.60	3.60	219	788.40
Walmart	4.59	25	114.75	3.84	25	96.00
Village Dr	4.50	11	49.50	4.09	11	44.99
Lakeview Ave	5.07	15	76.05	5.35	15	80.25
S Euclid St	6.02	46	276.92	4.96	46	228.16
S Idaho St	7.28	40	291.20	6.16	42	258.72
Market Place	7.37	13	95.81	6.07	14	84.98
Hills Dr	7.32	34	248.88	6.77	35	236.95
S Beach Blvd	7.48	50	374.00	7.01	49	343.49
Brass Lantern Dr	7.94	152	1206.88	7.06	152	1073.12
Total	na	na	4233.73	na	na	3761.49

Note: OD = origin–destination; veh = vehicle; S = South; N = North; Blvd = Boulevard; Dr = Drive; St = Street; Ave = Avenue; na = not applicable.

According to results, although the EB average travel time from Brass Lantern Drive to South Beach Boulevard and South Idaho Street increased under the OD-based partition timing plan, the accumulated total travel time for the listed intersections decreased by 2% overall compared with the arterial-band-based timing plan. For the WB, except at Arovista Avenue, and Lakeview Avenue, the average travel time declined at the remaining intersections under the OD-based partition timing plan, which shortened the accumulated total travel time by 11%.

A comparison was also made between the two timing strategies for the side-street left-turns traveling to the end intersection (Table 4), either Brass Lantern Drive (NB left turn) or North Berry Street (SB left turn). Among all the NB left-turn flows to Brass Lantern Drive, the OD-based partition timing plan shortened the accumulated total travel time by 7.8%. For the SB left-turns, on the other hand, the total travel times produced by the two timing plans were pretty close.

Table 4.

Measures of Effectiveness Comparison of Major Left-Turn Flows on the Side Streets

Origin intersection	Arterial band-based timing plan			OD-based partition timing plan
Origin intersection	Average travel time (min/veh)	Number of vehicles (veh)	Total travel time (min)	Average travel time (min/veh)	Number of vehicles (veh)	Total travel time (min)
Northbound
S Beach Blvd	1.86	251	466.86	1.80	254	457.20
Hills Dr	3.35	29	97.15	2.52	30	75.60
Market Place	3.34	59	197.06	2.85	58	165.30
S Idaho St	3.95	72	284.40	3.69	71	261.99
S Euclid St	5.19	41	212.79	5.02	40	200.80
Walmart	5.88	14	82.32	5.01	14	70.14
S Harbor Blvd	6.33	50	316.50	6.07	49	297.43
Total	na	na	1657.08	na	na	1528.46
Southbound
Brass Lantern Dr	8.54	4	34.16	8.20	4	32.80
S Beach Blvd	8.00	33	264.00	8.56	34	297.56
Hills Dr	6.97	11	76.67	7.78	12	93.36
Market Place	6.65	24	159.60	6.61	24	158.64
S Idaho St	6.99	42	293.58	6.92	43	297.56
S Euclid St	5.93	32	189.76	5.44	33	179.52
S Cypress St	5.70	25	142.50	5.77	25	144.25
Lakeview Ave	4.77	25	119.25	5.67	25	141.75
S Harbor Blvd	4.49	129	579.21	4.06	131	531.86
Palm St	3.19	126	401.94	3.04	128	389.12
S Puente St	2.83	128	362.24	2.79	129	359.91
Total	na	na	2622.91	na	na	2619.81

Note: OD = origin–destination; veh = vehicle; S = South; N = North; Blvd = Boulevard; Dr = Drive; St = Street; Ave = Avenue; na = not applicable.

From the above analyses, the OD-based partition timing plan outperformed the arterial-band-based timing plan by improving arterial average travel speeds and shortening arterial average travel times at the arterial level. Moreover, total travel times both for the main-street and side-street major traffic flows were shortened under the OD-based partition timing plan. Although no arterial bands existed for both directions in the OD-based partition timing plan, invaluable OD information extracted from connected vehicle data shed light on arterial traffic flow patterns for developing more effective signal coordination. Additionally, enabling minimum recall on the main-street leading left phases played a significant role in serving upcoming heavy left-turn flows to avoid phase gapping out, which could result in unnecessary stops and delay.

Conclusions and Discussions

This study proposed an innovative and easy-to-implement method, the OD-based partition technique, to improve arterial signal coordination integrated with connected vehicle data. The procedure includes the following five steps: arterial locating, OD extractions from connected vehicle trajectory data, arterial partition, subsystem signal coordination design, and subsystem side-street signal timing adjustment. First, locating boundaries of a signalized arterial lays the foundation for filtering vehicle trajectory data in light of geographical information. Second, traffic OD data needs to be extracted from the connected vehicle trajectory database: main-street left-turn traffic volumes starting from the through movement at the first intersection to each intersection ( $N_{MLT i}$ and $N_{MLT i}^{'}$ ) for both directions during a specific period of the day. Based on fluctuations of left-turn traffic flow patterns on the main street and adjacent intersection distance, partition points are determined separately for both directions and adjustments made to finalize arterial partition. The fourth step is to utilize traffic signal timing software to design signal coordination by maximizing bandwidths within each subsystem and building band connections between them to achieve an ideal arterial signal coordination. Finally, side-street left-turn volumes for both directions at each intersection ( $N_{SLT i}$ and $N_{SLT i}^{'}$ ) can be obtained to reveal side-street left-turn demands. Adjustments of side-street phasing sequence and intersection offset fine-tuning may improve the progression of major side-street left-turn movements by creating left-turn bands onto the main street.

West Imperial Highway, containing 17 signalized intersections, in City of La Habra, California was selected as a case study site to validate the effectiveness of the proposed OD-based partition technique. The study compared two signal coordination strategies: an arterial-band-based timing plan (no connected vehicle data support) and the OD-based partition timing plan integrated with connected vehicle data. The results demonstrated improved average travel speeds and travel times at the arterial level and reduced total travel times for both main-street and side-street major flows under the OD-based partition timing plan.

This study is the first of its kind to utilize connected vehicle data and apply an OD-based partition technique on a signalized long arterial (17 intersections covering 3.72 mi). Further research can dive into comparing and evaluating the effectiveness of the two signal coordination strategies, arterial-band-based and OD-based partition, on a short-distance (less than 10 intersections) or a medium-distance (10–15 intersections) arterial to explore the necessity of using connected vehicle data for benefiting signal coordination. Further study can focus on refining the OD-based partition technique as well if no obvious fluctuations exist on the main-street left-turn OD flows with more even adjacent intersection distance.

Footnotes

Acknowledgements

Connected vehicle data used in this research was provided by Ticon International, and time-of-day signal timing plans of West Imperial Highway were provided by City of La Habra, California.

Author Contributions

The authors confirm contribution to the paper as follows: study conception and design: Z. Tian, J. Xu; data collection: J. Xu; analysis and interpretation of results: J. Xu; draft manuscript preparation: J. Xu, Z. Tian. All authors reviewed the results and approved the final version of the manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Jianyuan Xu

Zong Tian

References

Morgan

J. T.

Little

J. D. C.

Synchronizing Traffic Signals for Maximal Bandwidth. Operations Research, Vol. 12, 1964, pp. 896–912.

Little

J. D. C.

The Synchronization of Traffic Signals by Mixed Integer Linear Programming. Operations Research, Vol. 14, 1966, pp. 568–594.

Messer

C. J.

Whitson

R. H.

Dudek

C. L.

Romano

E. J.

A Variable-Sequence Multiphase Progression Optimization Program. Highway Research Record: Journal of the Highway Research Board, 1973. 445: 24–33.

Little

J. D. C.

Kelson

M. D.

Gartner

N. H.

MAXBAND: A Versatile Program for Setting Signals on Arteries and Triangular Networks. Massachusetts Institute of Technology, Cambridge, 1981.

Chang

E. C.-P.

Cohen

S. L.

Liu

Chaudhary

N. A.

Messer

MAXBAND-86: Program for Optimizing Left-Turn Phase Sequence in Mutiarterial Closed Networks. Transportation Research Record: Journal of the Transportation Research Board, 1988. 1181: 61–67.

Gartner

N. H.

Assmann

S. F.

Lasaga

Hous

D. L.

MULTIBAND—A Program for Variable-Bandwidth Progress Optimization of Multiarterial Traffic Networks. Transportation Research Record: Journal of the Transportation Research Board, 1990. 1287: 212–222.

Stamatiadis

Gartner

N. H.

MULTIBAND—96: A Variable-Bandwidth Arterial Progression Scheme. Transportation Research Record: Journal of the Transportation Research Board, 1996. 1554: 9–17.

Tian

Urbanik

System Partition Technique to Improve Signal Coordination and Traffic Progression. Journal of Transportation Engineering, Vol. 133, 2007, pp. 119–128.

Fan

Tian

Arterial Signal Timing and Coordination: Sensitivity Analyses and Partition Techniques. Proc., International Conference on Traffic and Transportation Studies (ICTTS), Kunming, China, 2010, pp. 338–350.

10.

Lee

Chang

Carrion

Zhang

An Approach to Grouping Traffic Signals for Coordination Using Clustering Methods. Proc., IEEE International Conference on Models and Technologies for Intelligent Transportation Systems (MT-ITS), Naples, Italy, IEEE, New York, 2017, pp. 792–797.

11.

Zhang

System Partition Method to Improve Arterial Signal Coordination. Presented at 93rd Annual Meeting of the Transportation Research Board, Washington, DC, 2014.

12.

Zang

Meng

An Area Traffic Control Algorithm Based on Multi-Group Partition. Proc., International Symposium on Information Science and Engineering (ISISE), Shanghai, China, 2008, pp. 228–231.

13.

Tian

Yuan

Bandwidth Optimization of Coordinated Arterials Based on Group Partition Method. Proc., International Conference on Traffic and Transportation Studies (ICTTS), Changsha, China, 2012, pp. 232–244.

14.

Wang

Zhang

Correlation Degree Analysis of Arterial Adjacent Intersections for Coordinated Control Subunit Partition. Advances in Mechanical Engineering, Vol. 10, 2018, pp. 1–12.

15.

Hao

Lin

Cheng

Yang

Signal Progression Model for Long Arterial: Intersection Grouping and Coordination. IEEE Access, Vol. 6, 2018, pp. 30128–30136.

16.

Kun

Gartner

N. H.

Wang

A Partition-Enabled Multi-Mode Band Approach to Arterial Traffic Signal Optimization. IEEE Transactions on Intelligent Transportation Systems, Vol. 20, 2018, pp. 313–322.

17.

Yang

Cheng

Chang

A Multi-Path Progression Model for Synchronization of Arterial Traffic Signals. Transportation Research Part C: Emerging Technologies, Vol. 53, 2015, pp. 93–111.

18.

Arsava

Xie

Gartner

N. H.

Mwakalonge

Arterial Traffic Signal Coordination Utilizing Vehicular Traffic Origin-Destination Information. Proc., International IEEE Conference on Intelligent Transportation Systems (ITSC), Qingdao, China, IEEE, New York, 2014, pp. 2132–2137.

19.

Arsava

Xie

Gartner

N. H.

Arterial Progression Optimization Using OD-BAND: Case Study and Extensions. Transportation Research Record: Journal of the Transportation Research Board, 2016. 2558: 1–10.

20.

Arsava

Xie

Gartner

N. H.

OD-NETBAND: An Approach for Origin-Destination Based Network Progression Band Optimization. Transportation Research Record: Journal of the Transportation Research Board, 2018. 2672: 58–70.

21.

Hill

C. J.

Garrett

J. K.

2011 AASHTO Connected Vehicle Deployment Analysis. Publication FHWA-JPO-11-090. FHWA, US Department of Transportation, Washington, DC, 2011.

22.

Zheng

Liu

Estimating Traffic Volumes for Signalized Intersections Using Connected Vehicle Data. Transportation Research Part C: Emerging Technologies, Vol. 79, 2017, pp. 347–362.

23.

Christofa

Argote

Skabardonis

Arterial Queue Spillback Detection and Signal Control Based on Connected Vehicle Technology. Transportation Research Record: Journal of the Transportation Research Board, 2013. 2366: 61–70.

24.

Day

C. M.

Richardson

L. M.

Howard

Platte

Sturdevant

J. R.

Bullock

D. M.

Detector-Free Optimization of Traffic Signal Offsets with Connected Vehicle Data. Transportation Research Record: Journal of the Transportation Research Board, 2017. 2620: 54–68.

25.

Feng

Zamanipour

Head

K. L.

Khoshmagham

Connected Vehicle-Based Adaptive Signal Control and Applications. Transportation Research Record: Journal of the Transportation Research Board, 2016. 2558: 11–19.

26.

Beak

Head

K. L.

Feng

Adaptive Coordination Based on Connected Vehicle Technology. Transportation Research Record: Journal of the Transportation Research Board, 2017. 2619: 1–12.

27.

Yan

Lin

Wang

Network-Level Multiband Signal Coordination Scheme Based on Vehicle Trajectory Data. Transportation Research Part C: Emerging Technologies, Vol. 107, 2019, pp. 266–286.

28.

Teng

Liu

Rai

Transit Priority Signal Control Scheme Considering the Coordinated Phase for Single-Ring Sequential Phasing under Connected Vehicle Environment. IEEE Access, Vol. 7, 2019, pp. 61057–61069.

29.

Argote-Cabanero

Christofa

Skabardonis

Connected Vehicle Penetration Rate for Estimation of Arterial Measures of Effectiveness. Transportation Research Part C: Emerging Technologies, Vol. 60, 2015, pp. 298–312.

30.

Yue

Yang

Lin

Wang

Tian

Traffic Signal Retiming to Improve Corridor Performance. Journal of Transportation Engineering, Part A: Systems, Vol. 147, 2021, p. 05020009.