Transportation Research Part C 55 (2015) 191202

Contents lists available at ScienceDirect

Transportation Research Part C

journal homepage: www.elsevier.com/locate/trc

Optimality versus run time for isolated signalized intersections

David K. Hale a,, Byungkyu Brian Park b, Aleksandar Stevanovic c, Peng Su d, Jiaqi Ma d
a

University of Florida, PO Box 116585, Gainesville, FL 32611-6585, United States

University of Virginia, 351 McCormick Road, Charlottesville, VA 22904, United States
Florida Atlantic University, 777 Glades Road, Building 36, Room 225, Boca Raton, FL 33431, United States
d
Leidos, Inc., 11251 Roger Bacon Drive, Reston, VA 20190, United States
b
c

a r t i c l e

i n f o

Article history:
Received 13 October 2014
Received in revised form 2 January 2015
Accepted 17 February 2015
Available online 7 March 2015
Keywords:
Trafc signal timing
Simulation-based optimization
Genetic algorithm
Simulated annealing
Heuristic method
Tabu search

a b s t r a c t
Simulation-based optimization of trafc signal timing has become pervasive and popular,
in the eld of trafc engineering. When the underlying simulation model is well-trusted
and/or well-calibrated, it is only natural that typical engineers would want their signal
timing optimized using the judgment of that same model. As such, it becomes important
that the heuristic search methods typically used by these optimizations are capable of
locating global optimum solutions, for a wide range of signal systems. However off-line
and real-time solutions alike offer just a subset of the available search methods. The result
is that many optimizations are likely converging prematurely on mediocre solutions. In
response, this paper compares several search methods from the literature, in terms of both
optimality (i.e., solution quality) and computer run times. Simulated annealing and genetic
algorithm methods were equally effective in achieving near-global optimum solutions.
Two selection methods (roulette wheel and tournament), commonly used within genetic
algorithms, exhibited similar effectiveness. Tabu searching did not provide signicant benets. Trajectories of optimality versus run time (OVERT) were similar for each method,
except some methods aborted early along the same trajectory. Hill-climbing searches
always aborted early, even with a large number of step-sizes. Other methods only aborted
early when applied with ineffective parameter settings (e.g. mutation rate, annealing
schedule). These ndings imply (1) todays products encourage a sub-optimal one size ts
all approach, (2) heuristic search methods and parameters should be carefully selected
based on the system being optimized, (3) weaker searches abort early along the OVERT
curve, and (4) improper choice of methods and/or parameters can reduce optimization
benets by 2233%.
2015 Elsevier Ltd. All rights reserved.

1. Introduction
Simulation-based optimization of trafc signal timing has become pervasive and popular, in the eld of trafc engineering. This is mainly due to the robustness and reliability of prominent trafc analysis tools, in terms of their ability to evaluate
existing conditions. The trafc analysis tools used for signal optimization usually employ detailed analytical and/or macroscopic simulation models. Despite their complexity it is possible to perform hundreds of model runs within seconds, given
the speed of todays computers. This is quite helpful for signal optimization, because each of these model runs can be used to
Corresponding author at: Leidos, Inc., 11251 Roger Bacon Drive, Reston, VA 20190, United States.
E-mail addresses: [email protected], [email protected] (D.K. Hale), [email protected] (B.B. Park), [email protected] (A. Stevanovic), peng.
[email protected] (P. Su), [email protected] (J. Ma).
http://dx.doi.org/10.1016/j.trc.2015.02.015
0968-090X/ 2015 Elsevier Ltd. All rights reserved.

192

D.K. Hale et al. / Transportation Research Part C 55 (2015) 191202

test a different timing plan alternative. Although the microscopic simulation models are also trusted, their computer run
time requirements often preclude their direct involvement in the optimization process.
Optimization tools such as PASSER (TTI), TRANSYT (TRL), TRANSYT-7F (McTrans), Synchro (Trafcware), HCS-Streets
(McTrans), Vistro (PTV), and ArteryLite (AGA) employ variations of simulation-based optimization. One crucial advantage
of simulation-based optimization is the ability to simulate existing conditions, and perform model calibration based on those
existing conditions, prior to any design or optimization. When the underlying simulation model is well-trusted and/or
well-calibrated, it is only natural that typical engineers would want their signal timing optimized using the judgment of that
same model. Only simulation-based optimization can provide this.
1.1. Heuristic search methods used for trafc signal timing
As such, it becomes important that the heuristic search methods typically used by these optimizations are mathematically capable of locating global optimum solutions, for a wide range of signal systems. However, what we see in practice is that
the popular software solutions only offer a limited subset of the available search methods. Moreover, the software solutions
only offer a limited subset of the available parameters, which can often improve the efciency of those methods. The result is
that many signal optimization efforts are likely experiencing premature convergence on mediocre solutions, because they
involved ineffective methods and/or parameters. Some trafc signal timing studies (Oda et al., 1996) have shown genetic
algorithms to be more effective than hill-climbing methods, and one signal timing study (Agbolosu-Amison et al., 2009)
has shown genetic algorithms to be more effective than Harmony search. However no known trafc signal timing studies
have examined many other available search methods and parameters.
For example, the effectiveness or viability of heuristic search methods such as simulated annealing and Tabu search in the
signal optimization process appears to be unknown. No known trafc signal timing optimization studies have compared the
effectiveness of tournament selection and roulette wheel selection, within the genetic algorithm; or the number of steps and
step-sizes, within the hill-climbing method. There is a need to perform apples-to-apples comparisons of all these search
methods; by applying them to the same underlying evaluation model, across a wide variety of intersection geometries
and signal phasing. Finally, no known signal timing studies have focused on the ramications of widely-varying computer
run times required by these various methods.
1.2. Importance of computer run times
The issue of computer run times appears to be particularly underrated. Although many off-line optimizations complete
within seconds, these efforts are likely producing sub-optimal solutions for small networks, analyzed across a small number
of time periods. Engineers would care more about computer run times if they knew that (a) other available heuristic search
methods would often provide better designs for such networks and (b) future optimizations may require larger trafc
networks, numerous time periods, and micro-simulation of all candidate timing plans. Regarding real-time optimizations,
run times are important because timing plans are changing so frequently. This highlights an incentive to maximize the ratio
of optimality (i.e., solution quality) over run time.
The trade-off between optimality and run time becomes clear when working with these heuristic search methods. For
example, it is common to observe signicant improvement in the rst fraction of run time, followed by modest improvements requiring a much longer period of run time. It is common to observe the fastest-converging search methods producing
reasonable and effective timing plans, whereas longer-running (without converging) search methods arrive much closer to
the global optimum solution. It would be ideal if a new search method could obtain global optimum solutions in the shortest
run times, but such a method has been elusive.
Given the trade-off between run times and optimality, it stands to reason that different optimization projects would
benet from different search methods. Faster methods (e.g. hill-climbing search, greedy algorithms, equalizing degree of
saturation) are needed for real-time adaptive control, whereas slower methods (e.g. genetic algorithms, simulated annealing,
particle swarm optimization) are more acceptable for off-line applications. Faster methods are needed to process larger
networks across numerous time periods, whereas slower methods are more acceptable for small datasets. Different engineers have different standards regarding how much time they are willing and able to wait, during an off-line (non-adaptive)
optimization run. Faster methods are needed to directly optimize computationally expensive simulation models, whereas
slower methods are quite effective for simple capacity analysis (analytical) models.
1.3. Scope of this paper
To shed more light on the trade-offs between solution quality and time needed, this paper compares optimality and
computer run times for several top optimization search methods from the literature. Given that no optimization method outperforms all others under all conditions (Ciuffo and Punzo, April 2014), this paper will not try to determine the best
method for all conditions. Instead the paper will present evidence that some optimization methods, when used with the
right parameters, have a strong tendency to avoid premature convergence. To produce this evidence, over a thousand test
optimizations of isolated intersections were performed using one objective function, and one underlying trafc model.

D.K. Hale et al. / Transportation Research Part C 55 (2015) 191202

193

Follow-up studies will hopefully perform similar data collection with more optimization methods, more objective functions,
more trafc models, and more complex signal systems (i.e., coordinated signalized arterials and grid systems).
2. Literature review
A comprehensive review of the available signal optimization methods was provided by Yun and Park (2012). Early signal
optimization strategies involved minimizing intersection delay as predicted by the original Webster (1958) formula, and by
subsequent evolutionary versions (Akcelik, 1998; Husch, 1996; Carson and Maria, 1997) of this formula. This paper will not
evaluate these early optimization strategies, because modern methods for determining delay have much better accuracy.
Modern computational engines contain thousands of lines of programming code, and cannot easily be summarized into a
smaller set of equations. To achieve 100% agreement with the chosen underlying model during the optimization process,
it is preferable to execute the full underlying model, to evaluate candidate timing plans.
Hill-climbing optimization (Robertson, 1969) is one example of a simulation-based optimization method, and has been
employed by multiple signal timing tools (Hale, 2009; Chang and Messer, 1991; Wallace and Courage, 1991) since the
late 20th century. Additional simulation-based optimization methods are provided in the literature review by Yun and
Park (2012); who state that heuristic methods are ideal for complex engineering models, having numerous and complex
sub-optimal solutions (Perentonos et al., 2002). Examples of heuristic optimization methods include hill-climb search,
genetic algorithms, simulated annealing, Tabu search (Carson and Maria, 1997), and Harmony search (Geem et al.,
2001). Efciency of the genetic algorithm, which has become popular (Yun and Park, 2012) for use in signal optimization,
is also known to be affected by its underlying selection methods (e.g., roulette wheel and tournament) (Chevalier, 1994).
These simulation-based optimization methods from the literature are believed to be good candidates for signal timing
optimization in general.
2.1. Hill-climbing search
Hill-climbing search optimization is apparently applied in many different elds, and its use has been well-documented
in the context of signal optimization (Robertson, 1969; Hale, 2009). Hill-climbing optimization is advertised by off-line
products including Vistro, PASSER, and TRANSYT. Synchro provides a quasi-exhaustive method of offset optimization
(Yun and Park, 2012), which is very similar to hill-climbing. On-line adaptive signal products tend to have proprietary
classied algorithms; but during workshop presentations, some have been said to use hill-climbing methods. Other adaptive products are said to use the greedy algorithm and/or Equisat (Akcelik et al., 1997) which, like hill-climbing, appear
to be some of the fastest available methods (i.e. computationally inexpensive, in terms of computer run times) for timing
signals. Thus hill-climbing and related methods have become somewhat dominant in U.S. signal timing practices, over
the past few decades.
When the hill-climbing search method is in effect, miscellaneous jumps are made in search of the lowest objective
function value. When a single jump nds a better solution, another jump is made in the same direction, and using the same
step-size. Additional jumps are made until nally a worse solution is encountered, at which time the process backtracks to
the previous best solution. The hill-climb method then switches to a different step-size, and sometimes reverses jump
directions.
By jumping across the solution space at various step-sizes, the optimization process saves time by not evaluating each
point along the curve. By using a variety of step-sizes, the optimization process tries to avoid getting stuck in a local optimum solution. Unfortunately the number of steps and step-sizes is somewhat arbitrary, and no set of steps or step-sizes is
perfect for all solution spaces.
2.2. Genetic algorithm
The genetic algorithm (GA) is an evolution-inspired (Whitley, 1994) algorithm; capable of simultaneously searching
numerous areas of promising solution space, across numerous generations of individuals. This ability to simultaneously
search numerous areas of solution space gives GA an advantage over the hill-climbing method, in which one search must
solve the entire solution space. For trafc signal optimization, each individual within the population must be translated
(Park, 1998) into a long array of bits (1s and 0s), which represents the DNA chromosome. These chromosomes are then
manipulated by optimization processes such as competition, crossover, and mutation (Michalewicz, 1994).
Regarding the determination of which individuals are performing well enough to participate in crossover, GA uses tournament selection or roulette wheel selection. When the tournament method is in effect, pairs of individuals are randomly
selected from the population to compete with each other, and competition losers are eliminated from the gene pool.
When roulette wheel selection is in effect, weaker individuals tend to have higher probabilities of being eliminated from
the gene pool. No known signal timing studies have compared these selection methods.
Following crossover there is also a process of mutation, to randomly change a small number of characteristics within each
individual. Thus the crossover process allows simultaneous searching across the search space, while the mutation process
allows for consideration of fresh solutions. Some research (Chevalier, 1994) indicates that relatively low mutation rates

194

D.K. Hale et al. / Transportation Research Part C 55 (2015) 191202

are more efcient when applied in conjunction with tournament selection, whereas relatively high mutation rates are more
efcient with roulette wheel selection. No known signal timing studies have examined combinations of selection method
and mutation rate, but this paper will examine various combinations.
2.3. Simulated annealing
Simulated annealing (SA) is an optimization method (Kirkpatrick et al., 1983) inspired by metallurgy; capable of evaluating numerous sub-optimal solutions over time, but gradually reducing the number of sub-optimal solutions considered.
Similar to the mutation process within GA, the consideration of sub-optimal solutions prevents SA from quickly getting stuck
on a local optimum solution. However, the metallurgy-inspired slow cooling process ensures optimization, gradually
reducing the probability of accepting sub-optimal solutions over time.
The Boltzmann factor (Kittel and Kroemer, 1980) is used within SA to compute the probability of accepting new solutions,
as shown below in Eq. (1). Delta S represents the distance between tness of the current solution, and tness of any new
candidate solution. Alpha T represents the annealing schedule, which will affect the speed of optimization. The existence of
Delta S causes better solutions to have higher probabilities of acceptance. Similar to the mutation rate within GA, values of
Alpha T that are too low or too high could slow down the process of optimization, or prevent optimization altogether.

Pacceptance e/T
Ds

Research has shown that the success of SA depends strongly on the choice of a suitable annealing schedule (Christoph and
Hoffmann, 1993). Other research has shown that SA is proven to locate global optimum solutions; but only as the annealing
schedule approaches innity (TTG Incorporated, xxxx), implying a relationship between optimality and run times. No known
signal timing studies have examined the potential effectiveness of SA, but this paper will compare effectiveness of SA against
other methods.
2.4. Tabu search
Tabu search (TS) is an optimization method (Glover, 1986) that stores its most effective solutions, or at least components
of its most effective solutions, into short-term or long-term memory. During the optimization process, solution components
stored within this memory are marked as taboo and cannot be used by new candidate solutions. Solution components are
stored within memory in a rst-in-rst-out process; such that the oldest taboo components become fair game again, as new
components are added to the list. Similar to the mutation process within GA, the forced consideration of sub-optimal solutions prevents TS from quickly getting stuck on a local optimum solution. However if the Tabu memory is too long this could
slow down the process of optimization, or prevent optimization altogether. No known signal timing studies have examined
the potential effectiveness of TS, but this paper will perform such tests.
2.5. Harmony search
Harmony search (HS) is an optimization method (Geem et al., 2001) inspired by musical orchestras; capable of evaluating
numerous disharmonious melodies over time, but gradually converging on one harmonious melody. The original objective of
HS was to develop a new heuristic algorithm with better performance (better solutions, fewer iterations) than the prior art
algorithms (GA, SA, TS, etc.). Similar to the crossover process within GA, new harmonies are improvised by using elements
of the prior music. Similar to the competition process within GA, weaker harmonies are eliminated from consideration.
Similar to the mutation process within GA, harmony memory and pitch adjustment processes force random harmonies into
the solution space, at various intervals. The HS optimization method encountered criticism (Weyland, 2010; Padberg, 2012)
for not being substantially different from GA, and prior signal timing research (Oda et al., 1996) concluded that HS was less
effective than GA. For these reasons, HS was not included within the detailed comparison of optimization methods performed for this paper.
3. Methodology
This research focused on signal timing optimization for maximum green times at isolated, actuated signals. Chapter 18 of
the 2010 Highway Capacity Manual (HCM) (Transportation Research Board, 2010) contains computational procedures
designed to evaluate the operational efciency of signalized intersections. No known studies have ever focused on
Chapter 18, HCM-based optimization. Because of this, and due to the interest level in HCM Chapter 18, the choice was made
to use these analytical procedures as the underlying evaluation model. There was no need to explicitly consider cycle length
optimization, because cycle lengths are an automatic output of the Chapter 18 evaluation procedures. There was also no
need to consider phasing sequence optimization, which primarily benets coordinated intersections. Offset optimization
is not applicable to isolated intersections.
The choice of optimization objective function (Hajbabaie and Benekohal, 2013) is an important consideration, in the context of signal timing. However, the scope of this research was limited to comparing the effectiveness of various optimization

D.K. Hale et al. / Transportation Research Part C 55 (2015) 191202

195

search processes from the literature. The objective function used by this research was the intersection-wide average delay
per vehicle; which is currently the primary measure of effectiveness for determining level-of-service, in Chapter 18.
3.1. Implementation of optimization methods
To facilitate the comparison, several heuristic optimization methods were computer-programmed into a research version
of the HCS 2010 signalized analysis module. This analysis module is intended to be a 100% faithful implementation of the
HCM Chapter 18 procedures, and provides an apples-to-apples comparison platform for the optimization methods. Before
the beginning of this research, the signalized analysis module already contained functionality and logic for genetic algorithm
(GA) optimization, using the tournament selection method. Therefore, the implementation of new methods was performed
in this sequence:





GA optimization with roulette wheel selection.

Hill-climbing optimization.
Tabu search (implemented within a GA framework).
Simulated annealing (implemented within a hill-climbing framework).

Although the hill-climbing method seems likely to benet from adjustments to avoid premature convergence, implementing Tabu search (TS) within a hill-climbing framework was too difcult, so TS was implemented within a GA framework
instead. Because GA already contains elements of randomness, and was thus unlikely to benet, simulated annealing was not
implemented within a GA framework. The result of these modications was a group of optimization methods to be
compared, in terms of their optimality and run times; including:







GA with tournament selection,

GA with roulette wheel selection,
TS with tournament selection,
TS with roulette wheel selection,
standard hill-climbing search, and
simulated annealing.

3.2. Intersection conditions

It was desirable to test a wide variety of intersection geometries and signal phasing. Table 1 illustrates eight major
categories of test cases: protected-permitted left-turn phasing, protected-only left-turn phasing, permitted-only green ball
phasing, side-street split phasing, protected-permitted left-turn phasing, protected-permitted phasing from a shared lane,
T-shaped intersections, I-shaped intersections, and advanced conditions. The eight major categories were further divided
into 23 subcategories, based on various forms of left-turn treatment and intersection geometry. Finally, the 23 subcategories
were expanded into 72 test cases by considering dual turn lanes (dual, left, right), dual thru lanes (thru), leading lefts
(lead), lagging lefts (lag), and combinations (e.g., lagdual). Table 1 contains four sets of three columns showing the
major category (left), subcategory (middle), and case name (right). Input demand volumes for these test cases were adjusted
prior to the comparison experiment; until intersection-wide delays fell between 80 and 85 s per vehicle (i.e., near-capacity
conditions), to allow signicant potential space for optimization improvement.
The 72 test cases were not subjected to any amount of calibration or validation. Insight into the effectiveness of simulation-based (or black box-based) optimization methods should not require accurate modeling of specic, real-world locations. Instead, the most insight on optimization effectiveness should be gained through comprehensive testing of a wide
variety of typical real-world locations, which is how the experiment was designed.
3.3. Choice of optimization parameters
It is difcult to design a perfect optimization experiment, due to the large number of optimization parameters, and the
amount of time required for optimization. Hill-climb methods can have any number of steps and step-sizes. Simulated
annealing can have various annealing schedules. Tabu searching can be applied with short-term, medium-term, or long-term
memories. GA can be applied with different mutation rates, or different numbers of generations. Different intersection
conditions might possibly benet from different types of optimization. Despite these challenges, experience and judgment
were used to design a proper experiment, to shed light on the most effective optimization methods. This experimental design
process is described below.
3.3.1. Hill-climbing parameters
With hill-climbing optimization, it was necessary to determine an effective number of steps, and distribution of
step-sizes. A trial-and-error process was employed, to determine when additional steps would not yield signicant

196

D.K. Hale et al. / Transportation Research Part C 55 (2015) 191202

Table 1
Test cases with a wide variety of intersection geometry and signal phasing.
Subcat.
Protected-permitted
Leading lefts

Case
1-1
1-1dual
1-1thru

Lead-lag

1-2
1-2dual
1-2thru

Lag-lead

1-3
1-3dual
1-3thru

Lagging lefts

1-4
1-4dual
1-4thru

Protected only
No right turns

2
2dual
2thru

Right turns

3
3left
3right
3thru

Shared rights

4
4dual
4thru

Permitted only
All exclusive lanes

5
5left
5right
5thru

Shared rights

6
6dual
6thru

Shared lefts

7
7dual
7thru

One lane
Split phasing
All exclusive lanes

Shared rights

10
10dual
10thru

Shared lefts

11
11dual
11thru

One lane
Prot-perm-shared
Exclusive thru lanes

12

One lane

14
14lag

T-shape
Exclusive lanes on side street

9
9left
9right
9thru

13
13dual
13thru
13lag
13lagdual
13lagthru

15
15left
15right
15thru

197

D.K. Hale et al. / Transportation Research Part C 55 (2015) 191202

Table 1 (continued)
Subcat.

Case

Single lane on side street

16
16left
16right
16thru

I-shape
Exclusive lefts

Shared left

Advanced
Shielded right turns

Extensive satow adjustments

17
17dual
17thru
18
18thru

19
19left
19right
19thru
20
20dual
20thru

Fig. 1. Implementations of hill-climbing search (HC) and simulated annealing (SA).

improvements. This evaluation process showed that making 46 passes through a series of 8 step-sizes would capture
most of the optimization improvement, in most cases. The hill-climbing process is illustrated in Fig. 1.

198

D.K. Hale et al. / Transportation Research Part C 55 (2015) 191202

3.3.2. Simulated annealing parameters

Because simulated annealing (SA) was implemented within the framework of hill-climb optimization, it was necessary to
determine (1) an appropriate number of steps, (2) an effective distribution of step-sizes, and (3) an effective annealing schedule. Numerous combinations of steps, step-sizes, and schedules were tested, to estimate the most effective parameters. The
process showed that 10 annealing passes through 4 step-sizes would capture most of the optimization improvement, in most
cases. Although 10 annealing passes was more effective than 5 passes (as expected), 4 step-sizes was more effective than 8
(not expected). The effective SA parameters are shown in Fig. 1. It became clear that SA would sometimes terminate on solutions weaker than those evaluated earlier in the process, due to forced consideration of sub-optimal solutions. Because of
this, it was important to store the best solution observed so far during the process, so it could be retrieved after the full
annealing schedule expired.

3.3.3. Genetic algorithm parameters

With GA optimization, it was necessary to determine an appropriate mutation rate and number of generations. Mutation
rates of 1% and 4% were used, because these values have been effective for isolated intersections in the past (Hale, 2009).
Regarding the number of generations, it can frequently be observed that most improvement occurs in the early generations.
The evaluation process showed that 250 generations would capture most of the optimization improvement, in most cases.
For example, Fig. 2 is an image generated by the HCS software. This gure illustrates percentage improvement (Y-axis) over
time (X-axis). The improvement percentage was calculated as shown in Eq. (2); and this percentage was used extensively to
compare optimization methods, as shown later. The gure illustrates that, although maximum improvement occurs after
1500 generations, over 90% of this improvement occurred within the rst 250 generations. This improvement rate was typical of most of the intersections tested. Therefore, 250 generations was used for most GA-based optimization runs. Note that
although generations are somewhat synonymous with time, it took approximately 89 min to run 1500 generations. Given
the population size of 10 individuals, 1500 generations represents exactly 15,000 executions of the HCM Chapter 18
procedure.

PctImprov ement



dorig  dfinal
 100
dorig

where dorig and dnal are intersection-wide delays before and after optimization

3.3.4. Tabu search parameters

For Tabu search, implemented in the framework of GA, it was necessary to determine an effective Tabu memory. Low
storage could cause Tabu and GA to produce identical results. Overly-high storage could prevent solutions from improving.
A trial-and-error process indicated that storing 5 chromosome bits would allow efcient optimization, in most cases. Storing
10 bits made optimization less efcient in several cases. Thus whenever a new elite individual was born, as proven by comparison to all others in the population, up to 5 bit positions (differentiating this individual from the prior elite individual)
were stored into Tabu memory. These bit positions would then become unavailable for crossover and mutation by all future
candidate individuals, at least until replaced by future additions to the Tabu memory.

Fig. 2. Plot of GA optimization improvement percentage over time.

D.K. Hale et al. / Transportation Research Part C 55 (2015) 191202

199

4. Comparison of methods
4.1. Primary results
Fig. 3 illustrates four scatterplots, evaluating heuristic search methods in terms of optimality versus run time. Optimality
(Y-axis) is expressed as an improvement percentage between the initial and nal solutions. Run time (X-axis) is listed in
units of seconds. Each plotted point represents an average improvement percentage and run time based on 72 optimization
runs. In other words, the 72 basic test cases from Table 1 were optimized using a given heuristic method with given parameter settings, and these 72 results were averaged to produce one scatter point. Fig. 4 further compares the methods by
combining four plots into one.
Regarding the given parameter settings, traditional hill-climbing search was tested with three different passes (2, 4, and
6) through each step size. Simulated annealing (SA) was tested with four different annealing schedules (5, 10, 15, and 20).
The genetic algorithm (GA) with tournament selection was tested with three combinations of mutation rate and number of
generations (1% @ 250, 4% @ 250, 4% @ 375). GA with roulette wheel selection was tested with slightly different combinations (1% @ 250, 4% @ 250, 1% @ 375). Thus the four plots contain a total of 13 scatter points, based on 13 72 = 936 optimization runs.
GA-based roulette wheel and tournament selection were tested with different mutation rates at 375 generations, because
the initial data collection implied they would operate more efciently under different mutation rates. Specically, after 250
generations, the 4% mutation rate produced a much better improvement percentage (31.4% versus 20.7%) for tournament
selection. However, also after 250 generations, the 4% mutation rate produced a worse improvement percentage (28.8% versus 31.5%) for roulette selection. Therefore, when seeded with their favorite mutation rates for 375 generations, tournament and roulette selection both performed quite well with 32.0% and 32.1% improvements, respectively.

4.2. Tabu search and other results

During the overall experiment there were actually 16 72 = 1152 optimization runs performed, requiring a total of 14.6 h
of computer run time. Three points (based on 216 optimization runs) that could have been plotted were instead not included
on the graphs.
Two of these points involved Tabu searching (TS) at the 4% mutation rate, within the genetic algorithm. When TS was
combined with tournament selection, the average improvement changed from 31.4% to 31.3%. When TS was combined with
roulette wheel selection, the average improvement changed from 28.8% to 29.2%. Because the impact of TS was fairly negligible, these points were omitted from the graphs, in order to make the graphs more understandable.
A third discarded point involved simulated annealing (SA). In the original implementation of SA, Delta S represented the
distance between tness of the current and candidate solutions. In an alternative implementation that produced the third
discarded point, Delta S was re-programmed to represent distance between tness of the best solution encountered so
far, and candidate solutions. When using an annealing schedule of 10, this alternative implementation achieved a 26.8% average improvement, compared to 28.5% for the original implementation. However, the alternative SA required 17.4 s of average
run time, compared to 58.6 s for the original SA. Because the alternative SA (with annealing schedule 10) and original SA
(with annealing schedule 5) produced nearly identical ratios of optimality to run time, there was no need to consider or plot
the alternative SA.

4.3. Interpretation of results

These results provided some evidence that changing from regular searching to TS had mixed effects, which were possibly
self-cancelling effects, on optimization efciency. TS essentially species that, if a new signal setting (e.g., maximum green of
29 s for phase #3) is found that works well, the optimization process must temporarily try to nd optimized plans without
using that setting. This strategy leads to searching a wider range of signal setting values, possibly facilitating discovery of the
best values. The disadvantage is that TS spends less time examining interactions between that max green (e.g., 29 @ p3) and
other max greens, possibly preventing discovery of the best overall combination. Thus TS increases the number of values
tested, but spend less time testing combinations of values known to be effective. This could explain why TS had a somewhat
negligible effect on optimization efciency.
The recurring pattern within Figs. 24 is intriguing, and was not expected. The heuristic optimization methods have signicantly different mathematics and algorithms. The hill-climbing methods (traditional and SA) have only one optimum
solution during each iteration, whereas the genetic algorithm has a full population of current solutions. Traditional hillclimbing (HC) contains no randomness, whereas other methods (SA, GA, TS) randomly accept weaker solutions. Methods that
do employ randomness do so in very different ways. Even GA employs randomness in very different ways, depending on
which selection method (roulette wheel or tournament) and mutation rate are in effect. Given these fundamental differences, a consistent relationship of optimality to run time was not expected across all heuristic search methods, but was
observed nonetheless. This relationship observed in the prior gures is generalized now within Fig. 5.

200

D.K. Hale et al. / Transportation Research Part C 55 (2015) 191202

Fig. 3. Percentage improvement (Y-axis) versus seconds of run time (X-axis).

35.0

Improvement (%)

30.0
25.0
20.0

SA

15.0

HC
GA (T)

10.0

GA (R)

5.0
0.0

0.0

20.0

40.0

60.0

80.0

100.0

120.0

Run Time (sec)

Fig. 4. Combined percentage improvement versus seconds of run time.

Fig. 5. The OVERT curve.

Region A is the area of signicant sub-optimality. In this study, HC optimization runs were rarely able to escape Region A,
whose endpoint is marked by Point X. For optimizations with very little time available, or extremely slow computers, or
extremely slow underlying simulation models, or extremely high numbers of intersections and time periods, Region A might
represent the only feasible and practical solution space.
Region B is the area of transition, whose endpoint is marked by Point Y. For optimizations with medium time available, or
medium-speed computers, or medium-speed underlying simulation models, or medium numbers of intersections and time

D.K. Hale et al. / Transportation Research Part C 55 (2015) 191202

201

periods, Point Y might represent the most desirable and efcient outcome. Beyond Point Y, there would presumably be
diminishing returns on signicant additional run times required. However, it might be a challenge to make advance predictions on the amount of time needed to reach Point Y.
Region C is the area of ne-tuning, whose endpoint is marked by the global optimum solution (Point Z). For optimizations
with signicant time available, or extremely fast computers, or extremely fast underlying simulation models, or extremely
small numbers of intersections and time periods, Point Z might be frequently attainable. In this study both SA and GA appeared
fully capable of reaching Point Z; given a sufcient annealing schedule, or number of generations. However GA became trapped
in Region A when seeded with the wrong mutation rate, and SA might do the same if seeded with ineffective step-sizes.
Although the OVERT curve illustrates results averaged over approximately a thousand optimization runs, some optimization results deviated signicantly from this curve. For example in test case 11dual, both SA and GA performed poorly, with
approximately 23.0% improvement after 120 s. This represents a severe downward bending of the curve. In this same case, HC
achieved 23.6% improvement within 9 s. Then in test case 19thru, HC was able to achieve 55.2% improvement after 10 s,
which represents a severe upward bending of the curve. In case 19thru SA and GA achieved 52% and 55% improvement respectively, but required 109 and 94 s respectively. These results imply that the OVERT curve is not accurate in 100% of all cases; and
that the outlier cases might share common characteristics, which could possibly be identied by future research.
5. Conclusions
Despite the popularity of simulation-based optimization for trafc signal timing, the software tools may unintentionally
be encouraging a one size ts all philosophy. Popular tools in the industry offer only a limited subset of the available
heuristic search methods. The result is that many signal optimization efforts are likely experiencing premature convergence
on mediocre solutions, and optimization strategies not consistent with the trafc signal system in question. No known signal
timing studies have compared several top search methods from the literature in terms of optimality (i.e., solution quality)
versus computer run times, under the widest possible combination of intersection geometries and signal phasing.
Experimental results from this paper may be summarized as follows:
 The OVERT curve relationship between optimality and run time appears to be consistent for several heuristic search
methods.
 Traditional hill-climbing search, which still exists within popular off-line and adaptive optimization products, consistently fell far short of reaching global optimum solutions.
 When seeded with proper parameters, simulated annealing (SA) and genetic algorithm (GA) methods were equally effective at locating solutions close to the global optimum.
 Despite SA and GA improving performance by 3032% when applied with proper parameter settings, traditional hillclimbing search produced only 2223% improvement.
 Compared to when SA or GA were applied with proper parameters, other methods and parameters reduced optimization
benets by 2233%.
 When seeded with proper parameters, roulette wheel and tournament selection were equally effective within GA.
 The best mutation rate for use in conjunction with roulette wheel selection was distinctly different from the best mutation rate for use in conjunction with tournament selection.
 Tabu search did not signicantly improve or degrade optimization efciency.
Although this research tested all heuristic search methods considered to be top candidates at the time, there is now motivation to examine additional methods. The simultaneous perturbation stochastic approximation method could be capable of
bending the OVERT curve, producing global optimum solutions within minimal times, and might be fast enough for implementation within adaptive control products. In addition, because the greedy algorithm is used by some adaptive control
products, it would help to test the efciency of this method. However on the surface, the greedy algorithm appears to be
a quick-and-dirty method, similar to traditional hill-climbing.
Finally, upcoming research should investigate whether the OVERT curve relationship also holds for arterials and grids.
Such studies would likely conrm that the best parameter settings for mutation rate, annealing schedule, Tabu memory,
hill-climbing stepsizes, number of generations, etc. differ signicantly for problems of different complexity (e.g. different
numbers of intersections, or different numbers of trafc signal settings being optimized). The ongoing hypothesis is that
optimization methods and parameter settings should be considered and chosen more carefully. More search methods should
be developed inside the optimization tools. More methods and parameter settings should be made available to the engineer.
Proper guidelines should be developed on how to choose and apply these methods and settings. Expert systems could even
be designed to automatically apply different methods and settings, based on problem complexity.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.trc.
2015.02.015.

202

D.K. Hale et al. / Transportation Research Part C 55 (2015) 191202

References
Agbolosu-Amison, S., Park, B., Yun, I., 2009. Comparative evaluation of heuristic optimization methods in urban arterial network optimization. In: IEEE ITSC
2009, St. Louis, MO, October 37.
Akcelik, R., 1998. The highway capacity manual delay formula for signalized intersections. ITE J. 2327.
Akcelik, R., Chung, E., Besley, M., 1997. Recent research on actuated signal timing and performance evaluation and its application in SIDRA 5. In:
Compendium of Technical Papers (CD), 67th Annual Meeting of the Institution of Transportation Engineers. Boston, USA.
Carson, Y., Maria, A., 1997. Simulation optimization: methods and applications. In: Andradttir, S., Healy, K.J., Withers, D.H., Nelson, B.L. (Eds.), Proceedings
of the 1997 Winter Simulation Conference, pp. 118126.
Chang, E.C., Messer, C.J., 1991. PASSER II-90 Program Users Guide, Texas Transportation Institute. Texas A&M University, College Station, TX.
Chevalier, R.F., 1994. Balancing the effects of parameter settings on a genetic algorithm for multiple fault diagnosis. In: Proceedings of the Tenth Conference
on IEEE Articial Intelligence for Applications.
Christoph, M., Hoffmann, K.H., 1993. J. Phys. A: Math. Gen. 26, 32673277.
Ciuffo, B., Punzo, V., April 2014. No free lunch theorems applied to the calibration of trafc simulation models. IEEE Trans. Intell. Transport. Syst. 15 (2), 553
562.
Geem, Z.W., Kim, J.H., Loganathan, G.V., 2001. A new heuristic optimization algorithm: harmony search. Simulation.
Glover, F., 1986. Future paths for integer programming and links to articial intelligence. Comp. Operat. Res. 5, 533549.
Hajbabaie, A., Benekohal, R., 2013. Trafc signal timing optimization choosing the objective function. Transport. Res. Rec.: J. Transport. Res. Board, No.
2355 (Transportation Research Board of the National Academies, Washington, DC, 1019).
Hale, D.K., 2009. TRANSYT-7F Users Guide, McTrans Center. University of Florida, Gainesville, FL.
Husch, D., 1996. The Percentile Delay Method. Trafcware Internet Technical Support Articles, Albany, CA.
Kirkpatrick, S., Gelatt, G.C., Vecchi, M.P., 1983. Optimization by simulated annealing. Science 220 (4598).
Kittel, C., Kroemer, H., 1980. Thermal Physics, second ed. Freeman & Co., New York.
Michalewicz, Z., 1994. Genetic Algorithms + Data Structures = Evolution Programs. Springer-Verlag, Berlin.
Oda, T., Otokita, T., Tsugui T., Kohno, M., Mashiyama, Y., 1996. Optimization of signal control parameters using a genetic algorithm. In: Paper Presented at
the 3rd World Congress on ITS. Orlando, FL.
Padberg, M., 2012. Harmony search algorithms for binary optimization problems. Operat. Res. Proc. 343348, 2012.
Park, B., 1998. Development of Genetic Algorithm-Based Signal Optimization Program for Oversaturated Intersections, Ph.D. Dissertation. Texas A&M
University.
Perentonos, K.P., Arbantis, K.G., Sigrimis, N., 2002. Heuristic optimization methods for motion planning of autonomous agricultural vehicles. J. Global
Optimiz. 155170.
Robertson, D.I., 1969. TRANSYT: A Trafc Network Study Tool, Road Research Laboratory Report, LR 253. Crowthorne.
Transportation Research Board, Highway Capacity Manual, 2010. Transportation Research Board, Washington, DC, 2010.
<http://www.alignstar.com/support/clients/webhelpv5/Simulated_Annealing.htm>, TTG Incorporated (accessed on 27.11.13).
Wallace, C.E., Courage, K.G., 1991. PASSER II-90 Users Guide, Volume 3 of a Series Prepared for FHWA by COURAGE AND WALLACE. Gainesville, FL.
Webster, F.V., 1958. Trafc Signal Settings, Road Research Paper Number 39. Scientic and Industrial Research, HMSO, London.
Weyland, D., 2010. A rigorous analysis of the harmony search algorithm: how the research community can be misled by a Novel methodology. Int. J. Appl.
Metaheur. Comput. 2, 5060.
Whitley, D.A., 1994. Genetic algorithm tutorial. Stat. Comput. 4, 6585.
Yun, I., Park, B., 2012. Stochastic optimization method for coordinated actuated signal systems. ASCE J. Transport. Eng. 138 (7), 819829.