An Efficient Simulation Methodology For The Design of Traffic Lights at Intersections in Urban Areas

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

An Efficient Simulation Methodology

for the Design of Traffic Lights at


Intersections in Urban Areas
Balqies Sadoun
College of Engineering
Al-Balqa’ Applied University
Al-Salt, 19117, Jordan
balqiessadoun@yahoo.com

City traffic light control is a complex, challenging problem, making it difficult to perform analyses with
traditional analytical techniques. An essential goal of traffic management, planning, and design is
setting traffic signals at intersections to minimize the queue length and vehicle delay time. The au-
thor presents a simulation methodology that can be used as an efficient design aid for the analysis,
management, and design of efficient traffic signaling in intersections. A four-leg signalized intersec-
tion is used as a design example. The performance of the three-phase and four-phase plans using
simulation was compared, and simulation results showed that the three-phase plan has better perfor-
mance characteristics than the four-phase plan. The main performance metrics considered are the
mean waiting time, mean queue length, and mean utilization of green-light time. The results obtained
demonstrate the efficiency of the methodology.
Keywords: Traffic light signaling, transportation engineering and planning, simulation, optimization,
traffic control

1. Introduction tors are placed on minor approaches to the intersection.


The signal is green for the major street at all times until
Traffic congestion is considered an important issue for the street detectors are activated. In the fully actuated case,
many parties worldwide, including city planners and engi- every intersection approach has a detector or detectors, and
neers, local governments, urban policy makers, legislators, both cycle length and the green-light time for every phase
economists, and law enforcement agencies, among others. of the intersection can be varied.
The time spent idling in traffic wastes fuel, increases emis- The frequently used measures of the effectiveness of a
sions, deteriorates the environment, and affects the overall signalized intersection are the mean delay, length of queue,
standard of living. City and regional planners and engi- and number of stops. The delay is the most often used
neers have been working hard to improve the efficiency of measure of the effectiveness of a signalized intersection.
transportation systems and optimize their operation. It is directly related to the driver’s experience, in that it
Signalized intersections are considered the most com- describes the amount of time consumed in traversing the
plex tasks in the design of a traffic system. This means that intersection. Delay may include stopped time delay, the
a complete analysis of any signalized intersection requires approach delay (time lost when a vehicle decelerates or
thorough analysis, modeling, and simulation to obtain an accelerates from or to its ambient speed), the travel time
efficient system. Typically, traffic signals operate in one of delay (the difference between the actual and desired time
three different control models: (1) pretime operation, (2) to traverse the intersection), and the queuing delay. The
semiactuated operation, and (3) fully actuated operation. length of the queue at any given time is critical in deter-
In the pretime operation case, the cycle length, phases, mining when a given intersection will begin to impede the
and all intervals are preset. Each cycle of the signal is ex- discharge from an adjacent upstream intersection. For ex-
actly like another, and all intervals and phase times are the ample, if we assume a uniform arrival rate, then when the
same. There can be different pretimed settings for different red phase begins, vehicles begin to queue as none are be-
times of the day, and this is called the pretimed multipro- ing discharged. When the signal turns green again, queued
gram control [1-3]. In the semiactuated operation, detec- vehicles begin to depart at the saturation flow rate. This de-
parture continues until the departure curve intersects with
SIMULATION, Vol. 79, Issue 4, April 2003 243-251 the arrival curve, signifying the dissipation of the queue.
© 2003 The Society for Modeling and Simulation International
| The arrival and departure curves coincide until the next
|
DOI: 10.1177/0037549703038878
|
|

Downloaded from sim.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016


Sadoun

red phase. The number of stops is considered an important OPAC from the United States. The last three systems de-
metric in air quality models [1-7]. cide whether to switch the traffic lights at each step of time
The development of signal timing plans and design in- [16].
volves a number of important components. These include Congestion and traffic jams are major socioeconomic
(1) development of the phase plan and sequence, (2) de- issues that are supposed to be addressed by city planners
termination of the cycle length, and (3) allocation of ef- and engineers. Building a new road or highway is not al-
fective green-light time to the various phases. Develop- ways a feasible or cost-effective approach; therefore, one
ment of an appropriate phase plan for a given situation is good approach to tackle this problem is to use adaptive
one of the most important aspects of signal design. There (dynamic) traffic control. This approach has become of
are no simple schemes to develop a phase plan. Any fur- interest to some Departments of Transportation (DOTs)
ther calculation or analytic analysis for cycle length and worldwide [8-12].
green-light time determination requires the phase plan to be The patterns of traffic change during the day depend
specified. Thus, the development of a phase plan involves on external influences such as weather conditions, inci-
more professional judgment than determining the timing dents, holidays, and so on. To obtain optimal traffic con-
plan. The main aspect of any phase plan is determining trol policies, the model should adapt to these changes as
the left-turn movement requirement for protection [4-12] well changes in drivers’ reactions to such conditions. Also,
(see also http://www.cerf.org/hitec/eval/ongoing/atc.htm traffic load may need to be continuously measured by loop
and http://lcewww.et.tudelft.nl/∼crweb/archive/). detectors connected to a microcontroller-based system that
In traditional signal control systems, prestored timing also performs all intersection control functions [11]. In this
plans that have been developed offline using historical data study, we consider the traditional approach of traffic signal
are used. These systems are not responsive to dynamic traf- control systems since the adaptive traffic control scheme
fic demands. Adaptive traffic light signal control systems cannot be generalized and only applies to that specific case,
are capable of adapting to actual traffic conditions; they meaning that it is case dependant.
aim for continuous optimal system performance. These
adaptive traffic signal control systems require extensive de- 2. Phase Plans and Modeling
ployment of traffic detectors and surveillance equipment,
which can eliminate the need for signal retiming and ide- The two-phase plan is the simplest plan (see Fig. 1). Each
ally reduce maintenance and operating costs. street receives one phase during which all movement from
Among the industry standard of adaptive control sys- that street is made. All left and right turns are made on a
tems are the split-cycle offset optimization technique permitted basis. This form of signaling is appropriate for
(SCOOT) and the Sydney Coordinated Adaptive Traffic cases in which the mix of left turns and opposing through-
System (SCATS) [13-16] (see also http://signalsystems. flows is such that no unreasonable delays or unsafe condi-
tamu.edu/documents/TRBWorkshop2000/SCOOT_TRB tions are created by drivers.
2000Part 2.pdf). SCATS, which was developed by the The issue of whether left-turn protection on any or all
Roads and Traffic Authority (RTA) of New South Wales intersection approaches should be provided is difficult to
in Australia, is reputed to be the world’s most advanced determine but can be simplified by some guidelines:
dynamic traffic control system. It is basically a computer-
based area traffic signal control system. It consists of a 1. Left-turn protection should be used for left-turn vol-
complete system of hardware, software, and control phi- umes of more than 250 vehicles per hour (vph).
losophy and operates in real time, adjusting signal timings
throughout the system in response to variations in traffic 2. For volumes less than 250 vph, the opposing vol-
demand and system capacity. SCATS is meant to control umes and number of lanes should be considered.
traffic on an areawide basis rather than an uncoordinated in- 3. For small left-turn volumes less than 100 vph, left
tersection basis. SCATS can respond dynamically to traffic protection is rarely used.
pattern changes by continuously adjusting signal timings,
cycle by cycle, so that optimum traffic flow is achieved The three-phase plan is used where permitted. In addi-
[13, 14]. tion, protected phasing is allowed. This approach is more
SCOOT is a second-by-second command-and-monitoring difficult for drivers to understand, but it can provide ben-
adaptive control system. It uses the power of a local con- efits by reducing delays (see Fig. 2). The four-phase plan
troller, which controls the clearance, minimums, local ac- can be used to provide left- turn protection for all left-turn
tuation, and detector preprocessing. It is widely installed movements. This approach may cause more delays in some
in the United Kingdom, the United States, and Canada. cases. Figure 3 illustrates the four-phase plan.
The system has a DEC Alpha server running OpenVMS In general, many phase plans can be introduced to
with needed peripherals, local-area networks (LANs), de- intersections, and the traffic engineer and planner have
tectors, and transmission lines [13-16]. much flexibility in designing an appropriate phase plan.
Other adaptive systems include LHOVRA from Swe- The main objectives are efficiency of operation and safety
den, UTOPIA from Italy, PRODYN from France, and to the motorist. On the other hand, complicated phasing

244 SIMULATION Volume 79, Number 4

Downloaded from sim.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016


SIMULATION METHODOLOGY FOR DESIGNING TRAFFIC LIGHTS

Phase1 Phase2

Figure 1. Phase diagram of two-phase plan. Dashed line indicates the opposed movement in which the movement is not protected.

Phase1 Phase2 Phase3

Figure 2. Phase diagram of three-phase signal

Phase1 Phase2 Phase3 Phase4

Figure 3. Phase diagram of four-phase plan

must be measured not only from the efficiency point of the reference vehicle crosses the curb line. The first head-
view but also from the drivers’ point of view. A very com- way is relatively long because reaction time and the time
plicated phase plan may confuse drivers in most cases. to accelerate for the first are longer. The second headway
Although there are general criteria available to help in is shorter; after four or five vehicles, the headway tends to
the design process, the traffic engineer/planner must apply level out to some level, called the saturation headway (h).
knowledge and understanding of various phasing options Saturation flow rate (vehicles/hour) can be calculated using
and how they affect other critical aspects of signalization, S = 3600/ h, where h is the saturation headway. The sat-
such as capacity and delay. Some mechanisms and con- uration headway was in the range of 2.11 to 5.66 veh/sec.
cepts must be understood before any description is made The critical lane concept involves finding how much time
of cycle length of signalized intersections: (1) discharge is allocated for signals. During any given signal phase,
headway at the signalized intersection, (2) the critical lane, several lanes of traffic on one or more approaches are per-
and (3) peak-hour factor (PHF) and the required volume- mitted to move. One of them will have the most intense
to-capacity ratio (v/c). traffic. This is the lane for which time must be allocated
The discharge headway is the time between two suc- during the subject phase. It requires more time than the
cessive vehicles crossing the curb line as the rear wheel of other lanes.

Volume 79, Number 4 SIMULATION 245

Downloaded from sim.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016


Sadoun

Thus, if sufficient time is allocated for the critical lane, for the same intersection, and it will be compared to the
then all other lanes are accommodated as well. To calculate three-phase plan.
the cycle length for a phase plan, we should calculate the The second step is to convert all volumes to the
critical lane volumes and their summation. To account for “through-car unit” (tcu). The through-car equivalent for
peaking within the hour, PHF is used as a metric. We also nonprotected left-turning vehicles depends on the oppos-
consider the required v/c. ing vehicle flow. From tabulated data, we can get the
It should be stated that the utilization of effective green- through-car equivalent, as shown in Table 1. The through-
light time will never be 100%, and it is not desired to have car equivalent for protected left turns is equal to 1.05,
the v/c equal to 1 because this may result in high delay. while the through-car equivalent for right-turning vehi-
Then, cycle length can be computed using the following cles with the moderate conflicting condition of pedestrians
formula: (200 peds/hour) is equal to 1.32.
The third step is to identify the critical lane volumes for
C = [(N · tl )]/[1 − V c/[PHF · v/c · (3600/ h)], (1) each phase. The larger volume for each phase is considered
the critical volume for that phase. From Figure 5, we can
where N is the number of signal phases, tl is the lost time calculate the sum of critical lane volumes for this signal
per phase, V c is the sum of critical lane volumes, PHF is plan as follows:
the peak-hour factor, v/c is the required volume-to-capacity
ratio, and h is the saturation headway (h = 2.23 for a 12- V c = 263 + 516 + 375 = 1154 tcu.
foot lane).
The fourth step is to calculate the cycle length for this
3. Simulation Modeling model, assuming that the pretimed signal is in use. For
cycle length, the following is used:
In our study, we considered the intersection shown in Fig-
ure 4. To design an effective signal-timing plan for this C = [(N · tl )]/{1 − V c/(PHF · v/c · 3600/ h)]}, (2)
intersection, the following steps were considered: (1) de-
sign of an appropriate phase plan, (2) appropriate data con- where N is the number of signal phases, tl is the lost time
version, (3) calculation of critical lane volumes, (4) cycle per phase, V c is the sum of critical lane volumes, PHF is
length determination based on the above steps, (5) alloca- the expected peak-hour factor, v/c is the required volume-
tion of time between different phases, and (6) simulation to-capacity ratio, and h is the saturation headway (h =
and evaluation of the system by comparing it to alternative 2.23 for a 12-foot lane). For our model, we assume the
designs. following: N = 3 phases, tl = 3 sec, PHF = 0.92, v/c = 0.90,
The other timing plan that we worked on was a four- and V c = 1154 tcu. Therefore, the desired cycle length (C)
phase plan, which is popular in some countries. We studied is about 70 sec.
this plan to evaluate the original design and to show the The fifth step is to allocate the effective green-light time
differences between those two design approaches. for each phase. The lost time per phase is 3 sec. For the
The first step in our work is to design the appropriate three phases, the total lost time is 3 · 3 = 9 sec. The ef-
phase plan that will optimize the flow through the inter- fective green-light time is 70 − 9 = 61 sec. The allocation
section. There is some consideration regarding left turns, of effective green-light time for each phase will be propor-
which involves deciding whether to use protected left-turn tional to the critical lane volumes. The effective green-light
phases. This can be checked by the cross-products of left- time for each phase is given in Table 2.
turn volumes and opposing through-volumes. If they are The minimum cycle length can be obtained by keeping
less than 50,000 and the left-turn volume is less than 200 PHF = 1.0 and v/c = 1.0 in equation (2) and can be written
vph, then the protected left-turn phase is not required. The as follows:
calculation of the left-turn protected requirement for our
model is as follows: C = [(N · tl )]/{1 − V c/(3600/ h)]}.

• East: 75 < 200 In our model, the minimum cycle length is


75 · 610/2 = 22, 875 < 50, 000—no protected phase C = 3 · 3/1 − (1154/1615) = 32 sec,
• West: 35 < 200
35 · 500/2 = 8750 < 50, 000—no protected phase which is much smaller than the desired cycle length. How-
• North: 250 > 200—protected left-turn phase is required ever, this is not a practical case because it does not take into
• South: 220 > 200—protected left-turn phase is required account the peak-hour factor. It also produces an absolute
minimum cycle length in which every second of effective
Therefore, the recommended signal phasing is the three- green-light time would be used, for which the volume-to-
phase plan, which is shown below. capacity ratio of 1.0 is considered.
The other possible phasing is the four-phase plan, with Accordingly, we will select the cycle length, which
all left turns protected. This phasing will be simulated later takes PHF and v/c factors into account. In our model, the

246 SIMULATION Volume 79, Number 4

Downloaded from sim.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016


SIMULATION METHODOLOGY FOR DESIGNING TRAFFIC LIGHTS

1 250

800 NORTH

50

500

35
WEST
EAST
75
610

70

SOUTH 700

220 185

Figure 4. The intersection to be studied and the related vehicles’ volumes

Table 1. The through-car unit (tcu) equivalent


Shared
Lane Per Lane
Volume Volume Volume Volume
Approach Movement (vph) Equivalent (tcu) (Tcu) Tcu/lane

East Left 75 5 375 375


Through 500 1 500 500 + 66 566/2
Right 50 1.32 66 = 566 = 283
West Left 35 4 140 140
Through 610 1 610 610 + 92 702/2
Right 70 1.32 92 = 702 = 351
North Left 250 1.05 263 263
Through 800 1 800 800 + 23 1031/2
Right 175 1.32 231 1 =516
= 831
South Left 220 1.05 231 231
Through 700 1 700 700 + 24 944/2
Right 185 1.32 244 4 =472
= 944

Table 2. Allocation of the effective green-light time


Phase 1 Phase 2 Phase 3

GPh1 = (263/1154) · 61 GPh2 = (516/1154) · 61 GPh3 = (375/1154) · 61


= 13.9 sec = 27.3 sec = 19.8 sec

Volume 79, Number 4 SIMULATION 247

Downloaded from sim.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016


Sadoun

Phase1 Phase2 Phase3


283 tcu
231 tcu 472 tcu

375 tcu

140 tcu
351 tcu

263 tcu
516 tcu

Critical volume=263 Critical volume=516 Critical volume=375

Figure 5. The critical lane volumes

Figure 6. Four-phase plan

cycle length of 70 sec or greater is considered. The whole From equation (3), we can find that C equals 125 sec, which
design was simulated for the cycle length of C = 70 sec. is the desired cycle length.
We assumed that the time between arrivals is exponentially The total lost time for all phases is 3 · 4 = 12 sec.
distributed in all cases. The effective green-light time is 125 − 12 = 113 sec. The
The other possible design for our intersection is to use allocation of effective green-light time for each phase will
four-phase signal, with all protected left turns, as shown in be proportional to the critical lane volumes. We assumed
Figure 3. The same methodology used for the three-phase that the time between arrivals is exponentially distributed
plan is used for the four-phase plan (see Table 3). The for each approach.
through-car equivalent for protected left turns is equal to
1.05 for east and west approaches. From Figure 6, we can 4. Simulation Results
calculate the sum of critical lane volumes for this signal
plan: The design of infrastructures for automobile traffic poses
many challenges. Among these is the issue of automobile
V c = 263 + 516 + 79 + 351 = 1209 tcu. drivers’ behaviors; these drivers are not as easily manage-
The next step is to calculate the cycle length for this model, able and predictable as are drivers for mass-transportation
assuming that the pretimed signal is in use. For cycle systems, such as public buses and trains. Automobile
length, the following is used: drivers act as individual agents, managed only through
traffic rules, signs, and traffic lights, in particular. Clearly,
C = [N · tl ]/{1 − [V c/(PHF · v/c · 3600/ h)]}. (3) this makes predictions of the overall traffic dynamics

248 SIMULATION Volume 79, Number 4

Downloaded from sim.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016


SIMULATION METHODOLOGY FOR DESIGNING TRAFFIC LIGHTS

O verall mean waiting time average que length

40 36.09 4 3.705

No. of vehicles
30 3 2.397
Time (sec.)

19.06 3-phase 3-phase


20 2
4-phase 4-phase
10 1
0 0
3-phase 4-phase 3-phase 4-phase
No. of phases No. of phases

Figure 7. Mean waiting time results Figure 8. Average queue length results

challenging. Due to this complexity, most traffic control


as well as its short- and long-term planning today relies on ave ra ge utiliza tion of g re e n tim e
models of global dynamics derived from statistical proper- 0.40
ties of observed traffic. Therefore, the stochastic simulation
29.0 0%
approach is considered in this study. 0.30
We compared the performance of the three-phase and 22.5 5%
3-phas e
four-phase plans. The performance metrics (measures) 0.20
4-phas e
considered in our study are the average waiting time, av-
erage queue length, and average utilization of green-light 0.10
time. Other performance metrics can be derived easily from
0.00
these measures. The results are summarized in Table 4. The
simulation results were taken with 95% confidence. Each 3-pha se 4-p has e
experiment was repeated five times, and the average result
was taken. Figure 9. Average utilization of green-light time results
Waiting time is defined as the time a vehicle spends from
the moment it arrives at the intersection until it leaves it.
Figure 7 compares the mean waiting time in seconds for the use models to predict short-term traffic flow and optimiza-
three-phase and four-phase plans. It is clear from the figure tion routines to determine signal timings that minimize de-
that the mean waiting time for the three-phase plan is less lays and stops. Adaptive traffic control systems constantly
than that for the four- phase plan. The second metric is the monitor traffic on main streets and adjust the cycles of sig-
average queue length, which gives the degree of congestion nal lights accordingly. Such systems use sensors to capture
in the intersection. Figure 8 depicts the average number traffic flow every few seconds and analyze it instantly. Then
of vehicles waiting in queue in both the three-phase and they can change the length of the traffic light cycles accord-
four-phase plans. Clearly, the average queue length for the ingly, not just at one intersection but throughout the corri-
four-phase plan is higher than that for the three-phase plan. dor. Clearly, such a mechanism would improve the flow of
The utilization of green- light time is an also an important traffic. It was reported that in Toronto, Canada, such adap-
performance metric (measure) as it tells how much of a tive traffic control systems have reduced delays by 30%.
given green-light time was used. The higher the value of We plan to study this in order to evaluate the effectiveness
green-light time utilization, the better the design will be. of adaptive traffic light control on streets that experience
Figure 9 shows the average utilization of green-light time in large, sudden, unpredictable changes in traffic flow. Such
percentages for both the three-phase and four-phase plans. a study requires extensive real-time measurements and ex-
It is observed from the figure that the three-phase plan has perimentation. We would like to add that neural network
better utilization characteristics than the four-phase plan. systems, genetic algorithms, and fuzzy logic–based con-
As a future work, we plan to consider adaptive traffic trollers can be used to design such adaptive systems.
conditions. Such systems are designed to overcome the
limitations of pretimed control and respond to changes in 5. Conclusions
traffic flow by adjusting signal timings in accordance with
short-term fluctuations in traffic demand. Based on infor- In conclusion, optimal traffic signal settings at a signal-
mation from traffic detectors and sensors, these systems ized intersection can minimize the vehicle delay time,

Volume 79, Number 4 SIMULATION 249

Downloaded from sim.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016


Sadoun

Table 3. The through-car unit (tcu) equivalent for the four-phase case
Shared
Lane Per Lane
Volume Volume Volume Volume
Approach Movement (vph) Equivalent (tcu) (Tcu) Tcu/lane

East Left 75 1.05 78.75 79


Through 500 1 500 500 + 66 566/2
Right 50 1.32 66 = 566 = 283
West Left 35 1.05 36.75 37
Through 610 1 610 610 + 92 702/2
Right 70 1.32 92 = 702 = 351
North Left 250 1.05 263 263
Through 800 1 800 800 + 23 1031/2
Right 175 1.32 231 1 = 516
= 831
South Left 220 1.05 231 231
Through 700 1 700 700 + 24 944/2
Right 185 1.32 244 4 = 472
= 944

Table 4. Results of simulation


Equivalent
Experiment
Per Lane Time
Volume Average
Approach Movement (Tcu/lane) (sec)

East Left 375 9.6


Through 566/2
Right = 283 12.72
West Left 140 25.70
Through 702/2
Right = 351 10.25
North Left 263 13.68
Through 1031/2
Right = 516 6.97
South Left 231 15.6
Through 944/2
Right = 472 7.60

stops, and queue length at a stop line. They can also save 6. References
fuel, decrease overall emissions, and improve the overall
standards of living, especially in urban areas. The devel- [1] Roess, R. P., and W. R. McShane. 1998. Traffic engineering. 2d ed.
opment of an appropriate phase plan for a given intersec- Engelwood Cliffs, NJ: Prentice Hall.
[2] Clement, S., and J. Koshi. 1997. Traffic signal determination: The
tion is an important aspect of any traffic signal system cabal model. 1997 Genetic Algorithms in Engineering Systems:
design. Simulation analysis of such systems can provide Innovations and Applications Conference, pp. 63-68.
cost-effective and optimum designs. In this study, we pre- [3] McShane, W., and R. Roess. 1990. Traffic engineering. Engelwood
sented a detailed simulation study of three-phase and four- Cliffs, NJ: Prentice Hall.
phase plans to optimize traffic signal timings. Our simu- [4] Kuwahara, M., and M. Koshi. 1990. Decision of timings of signal
program switching in pretime multi-program control. In Trans-
lation study has shown that the three-phase plan is better portation and traffic theory, edited by M. Koshi, 531-50. New
than the four-phase plan for the given intersection under York: Elsevier.
study. The three-phase plan provided an improvement of [5] Nakats, T., S. Seki, S. Shibliya, and I. Tkaku. 1994. Artificial in-
29%, 35%, and 47% in the mean utilization, queue length, telligence approach for optimizing traffic timings on urban road
networks. Proceedings of the 1994 Vehicle Navigation and Infor-
and waiting time, respectively, over the four-phase plan. mation Systems Conference, pp. 199-202.
However, other performance metrics should also be con- [6] Wilshire, R., R. Black, R. Grochoske, and J. Higinbotham. 1985.
sidered when designing such systems, including safety, in- Traffic control systems handbook. FHW A-IP-85-12. Washing-
tersection capacity, and location of intersection. The traffic ton, DC: Federal Highway Administration, U.S. Department of
engineer/planner should apply knowledge and understand- Transportation.
[7] Sano, Y., M. Shiozawa, K. Nakamura, H. Inoue, T. Yokota, and
ing of various phasing options and how they affect critical T. Inoue. 1995. Traffic signal control system based on genetic
aspects and goals of the required signaling.

250 SIMULATION Volume 79, Number 4

Downloaded from sim.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016


SIMULATION METHODOLOGY FOR DESIGNING TRAFFIC LIGHTS

algorithm. Proceedings of the Second World Congress on Intelli- nal timing: The Cabal model. M.S. thesis, University of South
gent Transport Systems, Japan, pp. 211-18. Australia.
[8] Boillot, F. 1994. Evaluation of the real time urban traffic control al-
gorithm CRONOS: First phase. Seventh IFAC/IFORS Symposium
on Transportation, Tianjin, China, pp. 585-90.
[9] Boillot, F., J. Blosseville, J. Lessort, V. Motyka, M. Papageorgiou, and Balqies Sadoun received her M.S. in civil engineering from the
S. Sellam. 1992. Optimal signal control of urban traffic networks. Ecole Nationale Des Travaux Publique de L’Etat, Lyon, France.
Sixth International Conference on Road Traffic Monitoring and
Control, IEE, London, pp. 75-79.
She received M.S. and Ph.D. degrees in city and regional plan-
[10] Tavladakis, K., K. Kostalias, A. Dollas, K. Kalaitzakis, and N. ning, College of Engineering, The Ohio State University, Colum-
Voulgaris. 1997. Development of a microcontroller-based, dis- bus. Currently, she is an associate professor and chairperson
tributed, adaptive traffic control system. Proceedings of the 8th in the Department of Surveying Engineering and Geomatics,
IFAC/IFIP/IFORS Symposium on Transportation Systems, Cha- Faculty of Engineering, Al-Balqa’ Applied University, Jordan.
nia, Crete, pp. 50-57.
[11] Hoar, R., J. Penner, and C. Jacob. 2002. Evolutionary swarm traf- Among her previous positions are chair of the Department of
fic: If ant roads had traffic lights. Proceedings of the 2002 IEEE Architectural Engineering at Applied Science University and an
Congress on Evolutionary Computation, Honolulu, HI. Available: assistant professor at the City University of New York. She has
http://www.dsg.cs.tcd.ie/∼biegelg/research/st authored or coauthored more than 30 refereed journal articles,
[12] Tavladakis, K., and N. C. Voulgaris. 1999. Development of an
autonomous adaptive traffic control system [Online]. Available:
as well as conference articles and book chapters. She served
http://www.erudit.de/erudit/events/esit99/12624_p.pdf on the program and organization committees of several interna-
[13] Taylor, W. C., A. S. Abdel-Rahim, and Z. R. Shaikh. 1998. Incident tional conferences. In 1999, she served as the program cochair
management under SCAT adaptive control system. No. 11. Final of the 1999 Summer Computer Simulation Conference, SCSC’99.
report on incident management under the SCATS Adaptive Con- She has served as a track chair and session chair in different
trol System. Technical report EECS-ITS LAB-FT98-084, Univer-
sity of Michigan, April. international conferences. Her main research interests include
[14] Lowrie, P. R. 1990. SCATS—Sydney Coordinated Adaptive Traffic modeling and simulation applications, applied neural networks,
System: A traffic responsive method of controlling urban traffic. transportation engineering and planning, GIS applications, envi-
Technical report, Roads and Traffic Authority, Sydney, Australia. ronmental planning and engineering, passive solar design, quan-
[15] Hunt, P. B., and T. R. Holland. 1985. The effects of an incident in
SCOOT system, in simulation and on-street. Traffic Engineering
titative analysis and methods, system recognition and identifica-
and Control 28:55-58. tion, and application of information technology to city planning,
[16] Clement, S. J. 1996. Genetic algorithms in dynamic traffic sig- surveying, and civil engineering.

Volume 79, Number 4 SIMULATION 251

Downloaded from sim.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016

You might also like