Documente Academic
Documente Profesional
Documente Cultură
Abstract: As urban traffic congestion grows, traffic engineers must find ways to maximize the efficiency of traffic signal control. Different
control strategies, including actuated, coordinated, and adaptive, have their own strengths and weaknesses; therefore, it is necessary to
comprehensively evaluate these control modes to understand which strategy is most appropriate for users. This research carries out a case
study to evaluate the adaptive performance of Adaptive Control Software Lite (ACS-Lite) versus conventional coordinated-actuated and fully
actuated, noncoordinated control. The test was done along two congested arterials around Disneyland in Anaheim, California. The results
indicated that adaptive control did not perform as well as the well-calibrated and finely-tuned time-of-day coordination. These results also
indicated that for this type of congested network, the adaptive signal control is best suited to improving the efficiency when traffic demand is
unpredictable, variable, and in low volume. During peak hours, when traffic demand was high and predictable, conventional coordinated
time-of-day plans performed better. Future research could aim to improve adaptive control by adopting more coordinated features and
utilizing high-resolution data to improve the overall efficiency. DOI: 10.1061/JTEPBS.0000068. © 2017 American Society of Civil
Engineers.
Introduction introduced in 1928 (Klein et al. 2006). After almost a century, most
new traffic signal installations are equipped with vehicle-actuated
As the population is ever-increasing worldwide, so is the increase signal control. In fully actuated operation, a vehicle-actuated sig-
in urbanization and traffic congestion. With severe congestion nal uses sensors to detect and user-programmed settings to control
comes excess delays, vehicle operating costs, and emissions. It de- the timing of signals in response to traffic demand. This type of
creases the productivity of businesses, poses a safety threat to the operation is most effective at isolated locations or when traffic de-
public, and deteriorates the quality of life. The demand for traffic mand is predictably low. The user-programmed settings contribute
mobility is an issue that may no longer be overlooked. One of the to system operational flexibility and complexity. Jiang et al. (2011)
most cost-effective solutions is maximizing the efficiency of the evaluated the performance by using different controller parameters.
current transportation system through traffic signal control. Their results found that an actuated signal operates best when the
The conventional traffic signal control system uses historical volume-to-capacity (v/c) ratio is within 0.4–0.6 and the minimum
statistic data to create preset plans with various timings (El- green time for the main street is long enough to meet the demand.
Tantawy et al. 2013). However, over time, these timing plans In addition, they recommended relatively small vehicle extension
need to be updated and optimized for the current traffic demand. times and maximum green times for the side streets. Yarger (1993)
Outdated and poorly timed traffic signal timings account for about conducted a comparison on semiactuated versus fully actuated sig-
5–10% (about 300 million vehicle-hours) of all traffic delay and nals. While fixed-timed systems can produce good results, they
congestion on major routes (Chin et al. 2004). have to be updated regularly to reflect current traffic demands.
To improve the efficiency of signal control, several control strat- Yarger (1993) concluded that the benefit to fully actuated systems
egies have been developed, including actuated, coordinated, and was in their insensitivity to certain timing parameters such as cycle
adaptive control. The concept of actuated signal control was first length, split, and so forth.
The second signal control strategy, which has been most
1
Research Assistant, Dept. of Civil Engineering, California State widely applied by traffic engineers, is actuated-coordinated con-
Polytechnic Univ., 3801 West Temple Ave., Pomona, CA 91768. E-mail: trol. Actuated-coordinated control provides continuous progression
chiairene@yahoo.com
2 along an arterial with minimum stops, resulting in reduced travel
Assistant Professor, Dept. of Civil Engineering, California State
Polytechnic Univ., 3801 West Temple Ave., Pomona, CA 91768 (corre- delay on arterial streets, but could result in longer delays for mi-
sponding author). E-mail: xinkaiwu@cpp.edu nor movements such as left turns and traffic on side streets. When
3
Research Assistant, Dept. of Civil Engineering, California State intersections are closely spaced and volume on the coordinated
Polytechnic Univ., 3801 West Temple Ave., Pomona, CA 91768. E-mail: arterials is predictable, the actuated-coordinated signal system is
sdhaliwal@cpp.edu preferred compared to the isolated signal system (Park et al. 2014).
4
Principal Traffic Engineer, Traffic Management Center, 201 S. The Manual on Uniform Traffic Control Devices (MUTCD) recom-
Anaheim Blvd., #502, Anaheim, CA 92805. E-mail: jthai@anaheim.net mends that traffic signals within 0.5 miles of each other be coor-
5
Dept. of Civil Engineering, California State Polytechnic Univ., dinated under a common cycle length (FHWA 2009). Stevanovic
3801 West Temple Ave., Pomona, CA 91768. E-mail: xjia@cpp.edu
et al. (2011) evaluated the robustness of signal timing for condi-
Note. This manuscript was submitted on August 11, 2016; approved on
March 17, 2017; published online on June 30, 2017. Discussion period tions of varying traffic flows. They found that basing signal timing
open until November 30, 2017; separate discussions must be submitted plans from average traffic flow performs most optimally when ex-
for individual papers. This paper is part of the Journal of Transportation posed to day-to-day traffic flow variability. Day et al. (2008) evalu-
Engineering, Part A: Systems, © ASCE, ISSN 2473-2907. ated coordinated signal control and found that using fully actuated
disruption of progression (Park et al. 2014). Nathan Ficklin (1973) of case studies that were recently conducted, one of which was in
first observed the early return to green problem. Over the years, Gresham, Oregon, where SCATS was deployed at 11 intersections.
some studies have tried to solve this problem; for example, Abbas In 1998, new timing plans were installed, with travel times aver-
et al. (2001) proposed an algorithm that continually adjusted end- aging 6 min 8 s. However, by 2004, the travel times averaged 6 min
of-green offsets at an intersection with the objective of providing 33 s. This indicates that the TOD coordinated plans had degraded
smooth progression of a platoon through an intersection using the over time as traffic volumes changed. This same effect was evident
volume and occupancy profile of advance detectors. However, it between 2004 and 2007. After the implementation of the SCATS
still presents a problem when the split utilization is highly variable adaptive system, travel times have been reduced to the lowest re-
from cycle to cycle, as it often can be. corded levels, without the need to continually optimize the travel
The third highly promising signal control strategy is adaptive times to reflect current traffic conditions.
signal control. Since its introduction in the 1980s, Adaptive Traffic Shelby et al. (2008) conducted a performance evaluation on
Signal Control (ATSC) systems have shown great potential in alle- ACS-Lite in 2008. The system had been integrated and tested
viating the urban traffic congestion problem. It works by adjusting through Corridor Simulation (CORSIM). The results indicated that
signal timings in real time in response to traffic fluctuations. For ACS-Lite demonstrated significant benefits in the context of sub-
example, the Split, Cycle, and Offset Optimization Technique optimal settings, and no harm done in the context of signal timing
(SCOOT) adaptive system responds to change in traffic flow by that had been optimized with perfect knowledge of the traffic con-
estimating queue length through the use of exit detectors that pro- ditions. With the suboptimal timing settings, they found that signal
vide arrival flow information (Hunt et al. 1981). On the other hand, timing adjustments by ACS-Lite provided substantial reductions in
the Sydney Coordinated Adaptive Traffic System (SCATS) instead vehicle delay, travel time, vehicle stops, and fuel consumption. The
uses stop bar detection to diagnose the level of congestion by es- corresponding economic benefits of improved traffic flow were es-
timating the degree of saturation (Sims and Finlay 1984). Recently, timated to surpass system deployment costs within the first year.
similar to SCATS, the Adaptive Control Software Lite (ACS-Lite) Actuated, coordinated, and adaptive control strategies have
application determines the degree of saturation for each traffic shown great potential for improving the operational efficiency of
phase by calculating phase utilization (Luyanda et al. 2003). current traffic control systems as shown by many previous evalu-
Occupancy data from the stop bar detectors combined with phase ation studies. But these strategies also have drawbacks; thus it
timing information is used to calculate used green time. That is critical to find out under which circumstances which control
information plus data from the upstream advanced detectors are strategy will outperform others. Searching for the answer may
used to measure when the approaching vehicle will arrive at the be challenging because of the costly and timely extensive data col-
intersection. lection that is needed. This case study aims to shed light on
Several adaptive signal control systems have been evaluated. these signal control strategies by bringing significantly valuable
Stevanovic and Zlatkovic (2013) evaluated Rhythm Engineering’s suggestions/recommendations to agencies that will benefit from
InSync in 2013 by means of a microsimulation environment. The such information.
results concluded that the ATSC outperformed time-of-day (TOD) Another case study reported by Day et al. (2012) evaluated
signal timings in terms of traffic efficiency in all areas but side- traffic responsive (TR) control and the ACS-Lite system on a
street delay, where InSync was neither best nor worst. However, 19-intersection network in Morgantown (West Virginia) for a spe-
since the study was based on a microsimulation, the results may cial scenario with heavy volumes generated by a university football
be skewed in that the model cannot reliably replicate all intricacies game. Five alternative strategies were compared in the test network
of real-world traffic conditions. Hutton et al. (2010) also conducted using Vissim software: (1) fully actuated (or free) operations,
a research on InSync’s operational capabilities in 2010. Their evalu- (2) time-of-day (TOD) control under a special-event plan,
ation was along a 12-signalized intersection corridor and the (3) TOD with ACS-Lite, (4) TR, and (5) TR with ACS-Lite. Free
results indicated that the ATSC was effective in reducing travel operation, TR, and TOD had delay results that illustrated sliding
time, delay, and number of stops. The minor-street delay, however, trade-offs between local capacity assignment and corridor progres-
increased for most approaches where mainline delay showed sub- sion objectives, with free operation having the lowest noncoordi-
stantial decline. In another ATSC evaluation, Brilon and Wietholt nated phase delay and the highest coordinated phase delay, and vice
(2013) analyzed their experiences with the system MOTION in versa for TOD operation. ACS-Lite was shown to yield slight im-
Germany in 2013. Their results concluded an increase in traffic provements under both TOD and TR control, but the magnitude of
flow performance by around 30%. They came across a problem, improvement was small compared with the replacement of TOD
however, with MOTION switching the framework programs as with TR because of the ability of TR to select cycle lengths in re-
it is in transition every 15 min. This leads to periods of temporary lation to traffic volume.
standstill at some intersections and results in periods of unsatisfac- This case study will run three different control strategies
tory coordination. including (1) fully actuated (noncoordinated vehicle-actuated) con-
In 2010, Fehon and Peters (2010) compared systems that have trol, (2) coordinated-actuated (TOD vehicle-actuated) control, and
been successfully installed in the United States and they described (3) ACS-Lite adaptive control through Econolite’s Centracs Adap-
the evaluation techniques that were used to provide statistically tive model for a duration of one week each. A comprehensive
presented in the section, “Data Analysis and Results.” Lastly, from the advanced detectors, or exit detectors from the previous
“Concluding Remarks” concludes this paper with a number of intersection, are used for estimating vehicle movement to cal-
propositions for future research. culate offsets. The splits and offsets are updated once every three
cycles, as the software needs a minimum of three cycles’ worth
of data to optimally adjust the signal timing.
Background: Fully Actuated, Actuated-Coordinated,
and Adaptive Signal Control
Data Preparation
This research aims to carry out a case study to evaluate actuated,
coordinated, and adaptive signal control strategies. Some brief
Test Site
background information about these three traffic signal control
modes: The test site is located in the city of Anaheim in Southern
1. Fully actuated (non-coordinated vehicle-actuated) control: California. The city is home to one of the most traffic-generating
A vehicle-actuated signal uses sensors to detect and user- attractions, Disneyland, and is a host to more than 20 million
programmed settings to determine the phases served and the visitors each year. The Centracs Adaptive model has been imple-
amount of green time signals provided in response to traffic de- mented in Anaheim, along seven intersections and two corridors
mand. This control is governed by the phasing structure and the bordering Disneyland, as shown in Fig. 1. These arterials see a sig-
minimum and maximum green times allocated per phase. Based nificant increase in traffic flow due to the theme parks, especially
on whether or not there is a vehicle actuation from the detectors, during the summer, when this study was taking place. For this
the green time will run between the minimum and maximum reason, the study area is a great location to test all three control
times. If there is an actuation from the advanced detectors strategies’ capability under high, unstable traffic flow.
for the current green phase, that phase will acquire a prepro- Both Harbor Boulevard and Katella Avenue corridors have six
grammed green-time extension, which varies based on posted/ through lanes. The Harbor Blvd. corridor, with a speed limit of
average approach speed, until it gaps-out or maxes-out. If there 56.33 kph (35 mph) and a distance of 1.45 km (0.9 miles), has an
is an actuation on a conflicting phase, there will be a call to that Interstate 5 on- and off-ramp immediately north of the study area;
phase. This type of signal control allows for variable phase se- the Katella Ave. corridor, with a speed limit of 64.37 kph (40 mph)
quences, variable green times, and variable cycle lengths. Such and a distance of 0.8 km (0.5 miles), has an Interstate 5 on- and
variability allows the signal to efficiently allocate green time off-ramp immediately east of the study area. All seven intersections
based on current demands and operations, but lacks the ability
to synchronize due to varying green splits without a cycle length
and an offset.
2. Actuated-coordinated (time-of-day vehicle-actuated) control:
The signal timings are optimized and pretimed based on sepa-
rate timing plans for major day-to-day traffic patterns. The tim-
ings are based on historical traffic data such as volume counts,
turning movement counts, travel speed, and travel time. They
are governed by three main factors: cycle length, phase splits,
and offset. In this study, the intersections are running with ac-
tuated coordination, in which the controllers were operating in
coordinated mode and that all intersections in the study area are
fully actuated, i.e., having detection for every phase/movement.
This coordinated operation is based on a group of signals oper-
ating on a nonvarying cycle length with a predetermined start or
end of the coordinated phases where each vehicular movement
is allocated a green split based on historical data. Since the study
area consists of a low-speed network where the volumes are
always near capacity, a fixed force-off was not used because
the seconds gained, if any, would not have a made a material
difference. Existing timing plans with a lead-lag left turn phase
order to optimize coordination patterns were used.
3. ACS-Lite Adaptive control: This case study will be based on the
Fig. 1. Map of Harbor Blvd. and Katella Ave. corridors (Google Map
Centracs Adaptive model that utilizes the benefits of the original
data, ©2015 Google)
ACS-Lite in its Advanced Transportation Management System
6/25/15)
Week 3: Adaptive control (Tue. 6/30/15–Thu. 7/2/15). *Note:
Harbor Blvd./East Shuttle Area and Harbor Blvd./Manchester Traffic Flow
Ave. were not on adaptive control due to time and budget Traffic volume data along the two corridors were collected and
constraints. analyzed to see if the volume counts were consistent during the
During each of these weeks, data were collected both from three-week study period. The traffic volume data were collected
the field and from the Centracs model. From the Centracs model, from every detector. Table 1 displays the mainline southbound (SB)
high-resolution second-by-second detector flow and occupancy and northbound (NB) through movement of 2-h volume counts for
and phase signal timing data were archived for five of the inter- the Harbor Blvd. corridor, along with the number of travel runs
sections: Harbor Blvd./Disney Way, Harbor Blvd./Katella Ave., made. Table 2 displays the volume counts and number of travel
Harbor Blvd./Convention Way, Katella Ave./Clementine St., and runs for the mainline eastbound (EB) and westbound (WB) through
Katella Ave./Anaheim Blvd. The other two intersections, Harbor movements for the Katella Ave. corridor.
Blvd./Manchester Ave. and Harbor Blvd./East Shuttle Area, were To better visualize how the traffic flow varies during the three-
turned live on Centracs days before the project started, and they week study period, Fig. 2 displays the average volume counts for
were not connected to the Centracs Adaptive system. As a result, the 2-h time periods for the mainline through movements by week.
they did not have the same high-resolution data collection capa- The figure indicates that the intersections followed a similar pattern
bility as the other intersections. For those two intersections, the that increased during certain times of day and decreased during
volume and signal timing data were event-based, and data were other times of day.
collected through the ASC/3 controller trip log. For the Harbor Blvd. NB direction, Week 2 had a consistent
Travel-time data were collected on the field through the use increase in volume throughout the whole corridor. The increase
of mobile phone applications that tracked the drivers’ speed, GPS may represent an impact on the intersection performance results.
Table 1. Mainline through 2-h Volume Counts for Harbor Blvd. Corridor
(CW) (CW) (DW) (DW) (KA) (KA) (MA) (MA) (ESA) (ESA)
Time period Week Day NB TH SB TH NB TH SB TH NB TH SB TH NB TH SB TH NB TH SB TH
AM peak Week 1 Fully actuated (18 runs) Tue. 928 686 634 607 749 1,016 505 702 886 537
(7:00–9:00 a.m.) Wed. 1,088 818 804 1,053 735 1,011 513 772 947 713
Thu. 1,103 862 783 1,101 791 986 558 793 932 716
Week 2 Actuated-coordinated Tue. 1,231 1,099 969 1,018 810 1,486 709 823 N/A N/A
(16 runs) Wed. 1,306 1,146 1,008 1,036 835 1,399 635 830 N/A N/A
Thu. 1,269 1,030 945 1,015 755 1,471 680 745 N/A N/A
Week 3 Adaptive (9 runs) Tue. 1,267 660 653 1,156 790 825 374 857 N/A N/A
Wed. 1,178 633 590 1,035 731 889 375 868 N/A N/A
Thu. 738 736 709 1,045 762 774 482 737 N/A N/A
Midday Week 1 Fully actuated (18 runs) Tue. 785 740 781 753 801 737 625 623 1,057 984
(12:00–2:00 p.m.) Wed. 882 732 879 1,104 684 794 636 653 1,107 838
Thu. 908 713 769 1,067 755 815 619 617 N/A N/A
Week 2 Actuated-coordinated Tue. 1,241 936 878 942 641 1,043 859 465 N/A N/A
(15 runs) Wed. 1,151 855 895 913 708 1,086 898 443 N/A N/A
Thu. 1,140 942 789 978 578 1,038 909 491 N/A N/A
Week 3 Adaptive (7 runs) Tue. 1,234 796 519 912 718 733 263 412 N/A N/A
Wed. 1,133 661 627 1,037 722 744 506 715 N/A N/A
Thu. 746 870 932 1,190 732 767 603 720 N/A N/A
PM Peak Week 1 Fully actuated (18 runs) Tue. 1,207 848 890 988 831 970 459 847 1,113 1,070
(4:00–6:00 p.m.) Wed. 1,237 916 1,015 1,368 808 865 454 933 1,273 948
Thu. 1,153 935 867 1,330 868 950 514 967 1,036 919
Week 2 Actuated-coordinated Tue. 1,072 1,200 1,155 1,273 819 1,461 733 1,005 N/A N/A
(17 runs) Wed. 1,258 1,019 1,082 1,328 984 1,497 673 936 N/A N/A
Thu. 1,342 1,018 1,086 1,445 919 1,466 662 1,038 N/A N/A
Week 3 Adaptive (12 runs) Tue. 1,251 865 742 1,293 765 817 542 797 N/A N/A
Wed. 1,176 1,041 603 1,150 832 896 497 762 N/A N/A
Thu. 828 1,021 842 1,406 809 894 336 1,014 N/A N/A
Note: CW ¼ convention way; DW ¼ Disney Way; ESA ¼ east shuttle area; KA ¼ Katella Ave:; MA ¼ Manchester Ave:; NB ¼ Northbound; SB ¼
Southbound; TH ¼ through.
(a)
(b)
(c)
(d)
Fig. 2. Average 2-h volumes by intersection, where H-C ¼ Harbor Blvd:=Convention Way, H-K ¼ Harbor Blvd:=Katella Ave:, H-D ¼ Harbor
Blvd:=Disney Way, H-E ¼ Harbor Blvd:=East Shuttle, H-M ¼ Harbor Blvd:=Manchester Ave, K-H ¼ Katella Ave:=Harbor Blvd:, K-C ¼
Katella Ave:=Clementine St:, K-A ¼ Katella Ave:=Anaheim Blvd: (a) Harbor Blvd. NB direction; (b) Harbor Blvd. SB direction; (c) Katella
Ave. EB direction; (d) Katella Ave. WB direction
(a)
(b)
Fig. 3. Travel-time box plots: (a) Harbor Blvd. Corridor; (b) Katella Ave. corridors
Fig. 4. Time-distance and speed-distance diagrams for Harbor Blvd. NB direction during AM peak (different solid lines in figure represent
trajectories of different probe vehicles) (a) week 1: fully actuated; (b) week 2: actuated-coordinated; (c) week 3: adaptive
Coordination adaptive control performed the worst for the NB direction. It is clear
from the space-distance diagram that the runs during Week 3 are
The time-distance and speed-distance diagrams for the Harbor
Blvd. NB direction during the AM peak period are presented in more sporadic and less consistent.
Fig. 4. Note that some inconsistencies between the laminar flow
in time-space diagram and zero speed in speed-space diagram Signal Timing and Cycle-by-Cycle Vehicle Throughput
[particularly see Fig. 4(b)] are mainly due to the detection errors of
cell phone applications, which rely on an accelerometer in the cell It would be interesting to see if there are any correlations between
phone to detect motion and thus derive instantaneous speed. green time and cycle-by-cycle throughout under different con-
The time-distance diagram demonstrates that the intersections trol. Fig. 5 depicts cycle-by-cycle vehicle throughput and green
along Harbor Blvd. are more coordinated during Week 2. This time plots for the Harbor Blvd./Katella Ave., SB direction. The
may be because during the AM peak hour, the traffic flow is more duration of green time per cycle is displayed in green (curved
predictable and constant, and thus the coordinated time-of-day solid line). As can be seen in the figure, the green times for
control outperforms the fully actuated control and adaptive control. Week 1 are highly varied. This is because during the fully actuated
During all the runs, the vehicles all seem to come to a stop at control, the cycle lengths vary and are solely based on vehicle
Katella Ave. and from there drive freely through the corridor with actuations.
very little delay. From the speed-distance diagram, only a few of the During Week 2, the traffic signal control was running on
runs’ speeds dropped below 24.14 kph (15 mph) after Katella Ave. actuated-coordinated plans. For this particular intersection and
During Week 1’s fully actuated control, a third of the runs were phase, the AM peak split time was 39 s, Midday off-peak was 43 s,
able to clear through the corridor after Katella Ave., while the re- and the PM peak split time was 39 s. From the figure, it can be
mainder of the runs come to a stop at every light. This is because observed that the green times followed extremely closely to the
fully actuated control does not coordinate traffic between intersec- split times allocated for that TOD. What may account for the vari-
tions; its sole purpose is to move traffic based on vehicle actuations ance is the green extension time that may have occurred as a result
at that intersection only. of a vehicle actuation at the advanced detectors.
Week 3’s adaptive control seems to be less coordinated going The green times for Week 3 are governed initially by the TOD
NB than Week 1. The travel time and delay results showed that split times. However, with adaptive running behind the scenes, the
(b)
(c)
Fig. 5. Cycle-by-cycle vehicle throughput and green time plots for Harbor Blvd./Katella Ave. SB direction: (a) AM peak; (b) midday;
(c) PM peak (different solid lines in the figure represent the trajectories of different probe vehicles)
green times are reallocated based on traffic demand on all ap- Concluding Remarks
proaches. It can be seen in the figure that the green times are varied,
however not as varied as Week 1, which is attempting to stay close As the population increases and there are more drivers on the road,
to the original split times. traffic engineers must turn their focus to maximizing the efficiency
The blue bars in Fig. 5 display the cycle-by-cycle vehicle of the current transportation system through traffic signal control.
throughput: The number of vehicles per second that were able A case study was conducted to evaluate ACS-Lite’s adaptive
to go through the intersection during the green time. The red performance versus conventional actuated-coordinated signal con-
(dotted) line is the average vehicle throughput. During the AM trol, and fully actuated, vehicle-actuated control. The test was
and PM peak periods, Week 2’s coordinated control was able to done along heavy flow arterials around Disneyland in Anaheim,
California. The results indicated that adaptive control did not
move more vehicles per green time second than the other two
perform as well as the well-calibrated and finely tuned actuated-
weeks. During the Midday time period however, all three traffic
coordinated control. These results also indicated that for this type
signal controls performed similarly in terms of moving vehicles
of congested network, the adaptive traffic signal control is best
through the intersection.
suited to improving the efficiency of arterial intersections when
From Fig. 5, it can be seen that Week 1’s fully actuated control
traffic demand is unpredictable, variable, and in low volume.
performed slightly better than Week 3’s adaptive control. This may During the AM and PM peak hours, the traffic demand was high
be because the fully actuated control is only governed by actua- and predictable, thus in those situations, conventional actuated-
tion. Thus if there is a vehicle call on the current phase, it may coordinated plans performed better. Particularly, the TOD timing
extend the green to accommodate that phase, and in doing so, push patterns are based on traffic engineering principles and soundly de-
more vehicles through. The adaptive control takes into account veloped theory, so it is not a surprise that it may perform better
vehicle calls from all approaches. So instead of maximizing the than an ATSC (Gettman et al. 2015). This result was similar to the
vehicle throughput for a particular phase, it follows an access- conclusion that Shelby et al. (2008) came to, that ACS-Lite does
equity objective and favors all phases equally. Whereas the fully not have substantial performance improvements where the signal
actuated control does not have a limited cycle length, adaptive timings had already been previously optimized. The timing plans
control does. Thus it must take into account the set cycle length, for this test area were well calibrated and based on historically
and allocating the green times to all phases. For this reason, the recorded traffic data.
adaptive control may not move as many vehicles per green time The minor and subtle gains in improvement of adaptive signal
second as the other signal controls. control might be caused by its access-equity objective, which aims