See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/329663129

Traffic Lights Control using Wireless Sensors Networks

Conference Paper · October 2018

DOI: 10.1145/3286606.3286791

CITATIONS READS
9 1,856

2 authors:

Nouha Rida Abderrahim Hasbi

Mohammadia School of Engineers Mohammed V University of Rabat
14 PUBLICATIONS 156 CITATIONS 49 PUBLICATIONS 374 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Nouha Rida on 14 April 2019.

The user has requested enhancement of the downloaded file.

 Traffic Lights Control using Wireless Sensors Networks

Nouha Rida Aberrahim Hasbi

mohammed 5 university mohammed 5 university
P.O. Box 11000 P.O. Box 6221
Morocco Morocco
[email protected] [email protected]

ABSTRACT Smart transportation regroups several elements: traffic

management system to optimize traffic flow, signal priority or
Optimization of signal control at intersection by using real time emergency vehicles, adaptive traffic signalization control,
traffic collected by wireless sensor network presents an important rerouting of vehicles and passengers, system wide detection
focus of research into intelligent transportation system. Several and reporting, travel information system and smart parking
studies have proposed adaptive algorithm road traffic, which management system.
concentrates on determining green light length and sequence of In this paper we study one of the most important elements
the phases for each cycle in accordance with the real time traffic in smart transportation which is traffic management system in
detected, and in order to minimize the waiting time at the isolated intersection, its principle is to know how to react in
intersection, these methods have chosen to give priority to the
real time to the traffic change at the intersection lanes. In order
largest queue. In this paper we propose adaptive traffic light
control method for an isolated intersection that considers a to improve the control efficiency of the traffic light, to
number of traffic factors such as traffic volume, waiting time, etc. reducing the waiting time for the green light and avoiding
In this method we give priority to the shortest queue and we show traffic congestion at the intersection.
that this method gives an optimal waiting time. Knowing the traffic is an essential element in real time for
the operation of the road network and in particular to feed
KEYWORDS traffic management systems at the intersection. To do this, we
Smart traffic, intelligent transportation system, traffic light use a network of wireless sensors where the role of each sensor
control, traffic congestion is to collect a set of data in its environment and transmit it to
the base station, which is responsible for traffic management.
ACM Reference format:
The document organized as follows in section 2; we present
N. Rida, A. Hasbi,. 2018. SCA Proceedings Paper in word Format. In some traffic management algorithms. In section 3, we propose
Proceedings of ACM Woodstock conference, Tetouan MOROCCO,
an adaptive algorithm to determine the sequence of phases and
October 2018( SCA 2018), 4 pages. https://doi.org/10.1145/123 4
the time of green light. In section 4, we evaluate our method
via the SUMO simulator and we demonstrate its effectiveness
1 INTRODUCTION by comparing the results to other methods.
A smart city uses information and communications
technology to enhance its livability, workability, and 2 RELATED WORKS
sustainability. There are eight key aspects that define a Smart Optimization of control signalization in isolated
City: smart governance, smart energy, smart building, smart intersection is an important axis of research in intelligent
mobility, smart infrastructure, smart technology, smart transportation system. This contribution focuses on two
healthcare and smart citizen. essential criteria: to maximize the flow crossing the
Smart mobility is a part of this integrated system because intersection during the green light and to minimize vehicles
transportation depends on power and communications and waiting time, by exploiting the data collected by the wireless
emergency services need real time travel condition information sensors. In the literature we find several theoretical models:
and optimization of signals. fuzzy logic [1], fluid mechanics [2], neural networks [3] and
Through smart mobility, smart city benefits in: queues [4].
• Reducing travel time by reducing traffic and congestion In [1], a method based on fuzzy logic employed, in the
• Reducing pollution from transport method the authors present a table, which define the green
• Empowering people with choice and control light time according to the numbers of vehicles waiting. For
• Improving transportation safety example, for an intermediate flow with a number of vehicles
arrives between 5/min and 10/min, we allow duration of 20s
for green light.
SCA 2018, October 2018, Tetouan Morocco N. Rida and Hasbi.

The research work [5] proposes a light plan based on

movement combinations can make simultaneously without any
conflict. The authors use the same model of intersection
presented in figure 1 with a topology of two sensors in such
movement, which separated, by a variable distance that
depends on the maximum green time. In this model of
intersection, there are 4 lanes and 8 distinct combinations of
non-conflict movements. This algorithm then selects the
sequence of phases composing a cycle, according to several
criteria: the presence of priority vehicles, the duration of the
periods when no detection of new arrival, the cases of famine,
the total waiting time and the length of the queues wait.
The model proposed in [5], however, is based on Figure 1: Intersection model.
unrealistic assumptions, requiring vehicles to be of the same Our system (Fig 1) consists mainly of four directions (N, S,
type and run at the same speed. E, W) each direction contains two lanes (go straight and turn
The authors, in [4], use the model of intersection presented left). Vehicles always allowed turning right without restricting
in figure1, each lane equipped by two sensors that separated by traffic. Traditionally, each lane was equipped with two
a distance of 10 m. In addition, they presented each lane as a
magnetic sensors: one is located near the traffic light to count
M/M /1 queue model and they use the Little law that gives
the number of departures of vehicles and the other is installed
and equation 2 to calculate the queue length in each lane for a
at a variable distance, which depends on the maximum time
cycle.
allowed for the green light, from the first sensor to detect the
arrival of vehicles. Each controller of the traffic lights defines
Q L j = Q L(j-1) + G G - µG G + G R (1) a time called cycle which is a sequence of the phases and each
Where of the phases presents a time of green light necessary for two
W is the waiting time, Q L is the queue length, λ is the arrival movements (for example ES, EW, SN ...) which occur
simultaneously.
rate of the vehicles, j represents the traffic cycle number, Q L j is All sensors communicate and transfer traffic information to
the expected queue length in this cycle, QL is the queue base station which calculates the queues’ lengths for each
( j 1) direction and its average waiting time in order to control the
length from the previous cycle. traffic flow.
GG is The arrival rate in the green phase , G R the arrival
rate in the red phases and is considered equal to µG R within the TABLE 1: The conflict directions matrix
same cycle, µ is the departure rate. The G the green period of one WN WE SN SW EW ES NS NE
phase in seconds, and R represent the red light period in seconds
and it is equal to difference between the T (Traffic cycle duration) WN 1 0 0 0 1 0 0
and the green time. WE 1 0 0 1 0 0 0
The algorithm in [4] selects at each phase the combination of non-
conflicting movements whose have the largest number of SN 0 0 1 0 0 1 0
vehicles, to minimize the average length of the intersection. SW 0 0 1 0 0 0 1
The green light time is determined in proportion to the length of EW 0 1 0 0 1 0 0
the queues in the intersection.
ES 1 0 0 0 1 0 0

3 PROPOSED SOLUTION NS 0 0 1 0 0 0 1
NE 0 0 0 1 0 0 1
3.1 Traffic modeling

The table 1 presents the conflict directions matrix. Each

column and lane in the table demonstrates a direction in the
intersection and its status if it allowed or not. That means that
all movements with the number 1 can be made at the same
time without creating a blocking at the intersection, and the
other number 0 shows that if we choose that movement
direction and column direction in the same time we will have a

2
Smart City Applications SCA 2018, October 2018, Tetouan Morocco

conflict inside the intersection which increases the waiting Then we give the green light to a phase of two
time at the intersection. lines among the four lines remaining.

 At the moment t 3  t 0  GT1  GT2  GT3 :

3.2 Shortest phase first The GT3 is the green light time for the third phase.

In order to manage real time traffic at road intersections, the

existing methods in the literature give priority to the road that has GWT (t  t 3 )  WTi  6GT1  4GT2  2GT3
the greatest number of vehicles. In the opposite of these methods, (5)
we propose a solution based on the choice of the smallest phase
first. in this method we give the green light to phase with Finally, we give the green light to the last two lanes
minimum of vehicle waiting.
Below we calculate the total waiting time at each change of traffic
light and for one cycle at isolated intersection. And after we show  Table 3 summarizes the previous steps.
that the waiting time will be reduced with the smallest phase first
controller then the longest phase first controller, which choses the TABLE 2. THE VARIATION OF WAITING TIME FOR ONE CYCLE

phases with longest queue length. Time the GWT The Green
remain next time
 At the moment t 0 : ing phase of the
lanes next
The waiting time in the intersection GWT is the sum of all phase
the waiting times WT in the lines of the intersection: GWT (t  t 0 )  WTi GT1
t0 8 Phase
8
1
GWT (t  t 0 )  WT j (t  t 0 ) (2)
j 1 t1  t0 6 GWT (t  t1 )  WTi Phase GT2
We put the initial waiting time GWT (t  t 0 )  WTi  GT1  6GT1 2
and we choose a phase among the phases to give it a green
light with a green time . t 2  t0  4 GWT (t  t 2 )  WTi Phase GT3
GT1  GT2  6GT1  4GT2 3
 At t1  t 0  GT1 :
The is the green light time for the second phase.
t3  t 0  2 GWT (t  t 3 )  WTi Phase GT3
The equation 3 gives the global waiting time at t1 which is
GT1  GT2  6GT1  4GT2 4
the waiting time at t 0 added to six of GT1 , we add six of
 GT3  2GT3
GT1 because there are 6 lanes of 8 that are waiting to get the
green light.
Equation 5 presents the waiting time at the end of one
GWT (t  t1 )  WTi  6GT1 (3)
cycle, which shows that the waiting time for a cycle is
After we calculate the GWT, we choice one phase among proportional to the time of the green lights and the latter is
the remaining lanes to be the next phase. proportional to the number of vehicles in such lane. We
present below the methods to calculate the number of vehicle,
 At t 2  t 0  GT1  GT2 the waiting time and the time of the green light in such lane.
The first step is to collect the information received by
The GT2 is the green light time for the second phase.
the wireless sensors and analyze it in order to determine the
To calculate GWT we add to the GWT in t1 to lengths of the queues and their waiting times. The queues’
length for an instant t and for the lane j is given by equation
four times the green light time of the second phase
(6).
GT2 because there are four lanes that did not
have the green light yet for this cycle, Q j t   AR j  AG j  Q j (t  1)  DPj (6)

GWT (t  t 2 )  WTi  6GT1  4GT2 (4)

3
SCA 2018, October 2018, Tetouan Morocco N. Rida and Hasbi.

Where: in order to fix a coherent, complex Tmax and the possible

diversity of an intersection.
Q j t  : The queue length for lane j and at time t
AR j : Vehicles arriving during the red light and for the lane j. After calculating the waiting times on each change in the
signalization, the queues’ lengths and the green time, we will
AG j : Vehicles arriving during green light and for the lane j. exploit these criteria in order to determine the waiting time for
Q j (t  1) : Vehicles are remaining from the last green light. one cycle and determine which of the methods minimizes the
waiting time the largest phase first or smallest phase first. The
DPj : Number of departure during current green light.
table 1 presents the waiting time for different instants and
AR, AG and DP are calculated from the information received by during one cycle, at the end of the cycle the maximum waiting
the two sensors. time for all vehicles in this cycle is given in equation 5.
To calculate the waiting time W there are two cases:
 If there are no vehicles waiting from the last green light, GWT (t  t )  WT  6GT  4GT  2GT (5)
the waiting time is the time of the first vehicle arrived 3 i 1 2 3
during this red time.
 If there are still vehicles from the last green light, the According to the equation 7, the green time is proportional to
waiting time is the current red light time. the number of vehicles in each lane, so the equation 5
becomes:
Equation (4) gives the waiting time value for each lane as a
function of FA, time of the first vehicle arrived, RT the time of
GWT  WT  6(T  T * Q )  4(T  T * Q )  2(T  T * Q )
the current red light and K which is an index equal to 1 if there i s h 1 s h 2 s h 3
are vehicles waiting from the last selection of the green light  WT  12 * T  T * (6 * Q  4 * Q  2 * Q ) (9)
and 0 otherwise.
i s h 1 2 3
 Where Q is the maximum queue of the two lanes for the

 FA if K 0
i
W  phase i.
 The equation 8 shows that the waiting time is propositional to the

 RT if K 1
(7) queues of the first three phases then the GWT will be minimum if
Q1 < Q2 < Q3 < Q4 , and maximum if Q1 > Q2 > Q3 > Q4 .
The green light time TG required to serve the vehicles
present on the lane of one phase. It is therefore possible to obtain a minimum waiting time if
we choose the shortest phase first. Consequently, the
TG  TS  Th * Ql scheduling of the phases by the method of the shortest phase
(8) first is optimal that allows giving a minimum of waiting time.

With Ql  max( Qu , Qv ) , and TS is the average

starting time of a queue, Th is the mean time of light traversal 4 PERFORMANCE EVALUATION
In this section, we evaluate the performance of the method
for a vehicle, and Ql is the vehicle size of the largest queue proposed in this manuscript and compare it with the fixed time
for the chosen phase. controller and with the algorithm proposed by Youssef et al
The time of the green light must not exceed a maximum [5], which prioritizes the largest queues. For the simulation
limit Tmax . environment, we use the model of the intersection presented in
figure 1.
The maximum green light time has been the subject of
several studies. Kell and Fullterton [7] observed that Tmax must
be between 30 and 60 seconds. Lin [8] studied the relationship
between Tmax and the mean waiting time, so that Tmax is

consistent even at peak times. Orcutt [9] suggested that Tmax

should be long enough to let 1.3 times the average length of
the queue. Courage [10] indicated that a high Tmax had little
impact on an adaptive system if the traffic was too low.
Finally, [11] and [12] proposed methods that are more modern
4
Smart City Applications SCA 2018, October 2018, Tetouan Morocco

100 700
90 600

Queue's length
80
70 500
Waiting time(S)

60 400
50 300
40
30 200
20 100
10
0 0

2688
134
296
636
824
1136
1354
1546
1870
2102
2282
2478

2876
0
362
194

768
986
1204
1546
2030
2220
2414
2620
2790
0

Time(s)
SPF Time
Smallest Phase First
Longest pPhase First Youssef & al
Fixed time Fixed time

Figure 2: Average waiting time comparison between fixed- Fig. 4. Queue length comparison between fixed-time control,
time control, Smallest Phase First algorithm, and Youssef and Smallest Phase First (SPF) algorithm, and Youssef and al
al algorithm (Longest queue first). algorithm (Longest queue first).

2000 Figures 2 and 3 compare the results of simulations for the

Average Waiting time(s)

1700 waiting time between three methods and shows that the
1400 algorithm (SPF :Smallest Phase First) give a reduced waiting
1100 time compared to that obtained by the algorithm proposed by
800 Youssef et al, and by the fixed-time controller. However,
Figure 4, which shows the variation of queue lengths during
500
the simulation, shows that the algorithm of Youssef and al
200
(Longest queue first) gives queues smaller than those obtained
-100
by the other methods.
134
246
362
694
824
986
1136
1280
1546
1790
2102
2220
2344
2478
0

The results of the simulations show that the algorithm

SPF Time(s) proposed in this research work provides a significant
Yousse et
& al improvement in the waiting time but it increases the queue in
Fixed time comparison with Youssef and Al algorithm, which reduces the
waiting time and increases congestion at the intersection, so
we can use our method SPF in intersections that do not have
Fig. 3. Accumulated Average waiting time comparison congestion constraints.
between fixed-time control, Smallest Phase First(SPF)
algorithm and Youssef and al algorithm,(Longest queue
first).

5 CONCLUSION

We have presented some models of road traffic

management in isolated intersection, those methods allow
collect traffic flow information and transmit data in real time
by using wireless sensor networks and it can choose the next
green light according to various criteria based on the data
collected by the sensors. One of these criteria gives the priority
to the lane that has the largest queue in order to minimize the
waiting time in the intersection. In this document we also
proposed another method of phases scheduling which is based
on the choice of the smallest queue first that prioritizes the
phase with minimum vehicles. We have shown that our
method reduces the waiting time compared to the method

5
 SCA 2018, October 2018, Tetouan Morocco N. Rida and Hasbi.

based in the choice of largest phase first. According to the

simulations, we found that largest phase first controller is more
adequate at the intersection that suffers from congestion
problems because the number of vehicles waiting is smaller
than that in the intersection with a controller used in our
method, so we can use our method in intersections that do not
have congestion constraints.

REFERENCES
[1] . ZOU, Fuqiang, YANG, Bo, et CAO, Yitao. Traffic light control for a single
intersection based on wireless sensor network. In : Electronic Measurement &
Instruments, 2009. ICEMI'09. 9th International Conference on. IEEE, 2009. p.
1-1040-1-1044.
[2] LIU, Binyang et LIU, Weifang. Evaluation of traffic control methods at traffic
circles. In : Control and Decision Conference (CCDC), 2011 Chinese. IEEE,
2011. p. 3371-3377.
[3] WEI, Wu et ZHANG, Yi. FL-FN based traffic signal control. In: Fuzzy
Systems, 2002. FUZZ-IEEE'02. Proceedings of the 2002 IEEE International
Conference on. IEEE, 2002. p. 296-300.
[4] YOUSEF, Khalil M., AL-KARAKI, Mamal N., et SHATNAWI, Ali M.
Intelligent traffic light flow control system using wireless sensors networks. J.
Inf. Sci. Eng., 2010, vol. 26, no 3, p. 753-768.
[5] ZHOU, Binbin, CAO, Jiannong, ZENG, Xiaoqin, et al. Adaptive traffic light
control in wireless sensor network-based intelligent transportation system. In :
Vehicular technology conference fall (VTC 2010-Fall), 2010 IEEE 72nd. IEEE,
2010. p. 1-5.
[6] FAYE, Sébastien, CHAUDET, Claude, et DEMEURE, Isabelle. Un algorithme
distribué de contrôle des feux de circulation sur plusieurs intersections par un
réseau de capteurs sans fil. In : Nouvelles Technologies de la
Répartition/Colloque francophone sur l'ingénierie des protocoles
(NOTERE/CFIP). 2012. p. 1.
[7] KELL, James H. et FULLERTON, Iris J. Manual of traffic signal design. 1991.
[8] LIN, Feng-Bor. Optimal timing settings and detector lengths of presence mode
full-actuated control. 1985.
[9] YANG, Qi et KOUTSOPOULOS, Haris N. A microscopic traffic simulator for
evaluation of dynamic traffic management systems. Transportation Research-
Part C Emerging Technologies, 1996, vol. 4, no 3, p. 113-130.
[10] FAYE, Sébastien, CHAUDET, Claude, et DEMEURE, Isabelle. Contrôle du
trafic routier urbain par un réseau fixe de capteurs sans fil. 2012.

[11] KIM, Jin-Tae et COURAGE, Kenneth. Evaluation and design of maximum

green time settings for traffic-actuated control. Transportation Research
Record: Journal of the Transportation Research Board, 2003, no 1852, p. 246-
255.
[12]
. ZHANG, Guohui et WANG, Yinhai. Optimizing minimum and maximum
green time settings for traffic actuated control at isolated intersections. IEEE
Transactions on Intelligent Transportation Systems, 2011, vol. 12, no 1, p. 164-
173.
[13] PRABHU, Boselin, ANTONY, A., et BALAKUMAR, N. A research on smart
transportation using sensors and embedded systems. 2017.
[14] RIZWAN, Patan, SURESH, K., et BABU, M. Rajasekhara. Real-time smart
traffic management system for smart cities by using Internet of Things and big
data. In : Emerging Technological Trends (ICETT), International Conference
on. IEEE, 2016. p. 1-7.
[15] CHEN, Bi Yu et LAM, William HK. Special issue: Smart transportation:
Theory and practice. Journal of Advanced Transportation, 2016, vol. 50, no 2,
p. 141-144.

View publication stats