International Journal of Engineering and Advanced Technology (IJEAT)

ISSN: 2249 – 8958, Volume-8, Issue-6, August 2019

IoT Based Real Time Traffic Monitoring System

Using M/G/1 Queuing

Charushila Raskar, Shikha Nema 

ABSTRACT--- Vehicular Traffic crowding is paramount The traditional remedies to increase the road capacity
worry in urban cities. The use of technologies like Intelligent such as Ramp metering, congestion pricing, VIP lanes are
Transportation systems and Internet of Things can solve the not able to provide satisfactory results due to financial
problem of traffic congestion to some extent. The paper analyses
investments required are huge and rise in vehicle number is
the traffic conditions on a particular urban highway using
queuing theory approach. It researches on performance exponential. Based on the everyday variation in travel needs
framework such as time for waiting and queue length. The advanced analysis methods should be used. Advanced traffic
results can provide significant analysis to predict traffic modelling tools and network simulators can be used to carry
congestion during peak hours. A congestion controlling action out the research operations.
can be generated to utilize the road capacity fully during peak
hours by using these results.

Kewords—Vehicular Traffic congestion, queuing theory,

traffic congestion order, Intelligent Transportation systems (ITS),
traffic density, M/M/1, M/G/1, G/G/1, G/G/k model, traffic flow.

I. INTRODUCTION
In countries like India with high level of population
vehicular traffic congestion on roads and its subsequent
effects such as drastic increase in harmful gas emissions,
increase in fuel consumption, tremendous increase in driver
fatigue have become prime concerns for the government and Fig. 1: Traffic flow vs Density relationship.
town planners. Though the efforts have been put up to
control and avoid congestion, the results are not satisfactory
A. Traffic flow Theory
as the number of commutators and vehicles are exceeding
road capacity. The use of technologies such as Intelligent The traditional traffic flow theory [1] [8] [9] [10] explains
Transportation systems (ITS) are in the infantry stage or not the relationship between the traffic flow and density.
matured enough to fulfil the needs. In comparison with the Initially when the vehicle enters the lane and traffic density
foreign countries like USA the use of information and is less the vehicle starts accelerating. The vehicle flow
communication technology has not grown enough to increases till the traffic density reaches to a critical value. At
completely solve the traffic congestion problem in all this it is a free flow phase. It totally depends on dimensions
aspects. Government has taken initiative in the development of the road. After critical density value vehicular speed
of smart cities with the use of digital world. There is a huge starts decreasing and it is the start of congestion phase. Here
scope for the Internet of things to explore various it starts queue formation. This continues to a jam phase
applications in the context of smart city in India. Internet of where maximum density is reached and vehicles halts on the
Things along with ITS can solve the traffic congestion road, queue length is at max.
problem to large extent. As per Gartner’s hype cycle billions By observing the figure 1 it can be conclude that there are
and more than that of world population is living with smart two different phases of traffic. In first phase, if the density
devices. The use of these smart things can definitely help in of the vehicles is lesser then the speed of the vehicle is as
addressing the traffic congestion problem. IoT can provide a fast as they can. It is also called as free-flow traffic which is
way to smart transportation. It can offer a smart traffic seen by thick line in fig 1. This type of phase remains in this
monitoring, Analysis of the traffic data and dynamic state till density reaches to its maximum flow; it is also
congestion control techniques in the traffic management. called as critical value. During the first phase some of the
vehicle needs to influence their momentum to maintain the
distance from the other vehicle on the road. This is also
called as forced flow. It is seen by dashed lines. The traffic
Revised Manuscript Received on 14 February, 2019.
Charushila Raskar, PhD student at Usha Mittal Institute of Technology reaches to its second phase due to some of the reasons like
(UMIT), SNDT Women’s University, Mumbai-400049, India (e-mail: angle of road, slanting of road at upward direction i.e. this
[email protected]).
Dr. Shikha Nema, Professor with the Electronics and
Telecommunication Engineering Department, Usha Mittal Institute of
Technology (UMIT), SNDT Women’s University Mumbai-400049, India
(e-mail: [email protected]).

Published By:
Retrieval Number: F9397088619/19©BEIESP Blue Eyes Intelligence Engineering
DOI: 10.35940/ijeat.F9397.088619 177 & Sciences Publication
 IOT BASED REAL TIME TRAFFIC MONITORING SYSTEM USING M/G/1 QUEUING

happens due to some slower and big transportation Priority - vehicles are assisting in order of their
vehicles and accident occurrence etc. The traffic reaches its importance such as emergency vehicles, ambulance on the
jam phase where vehicle stops or moves slowly. basis of their service requirements.
Paper gives the study of different queuing model for
B. Queuing approach
traffic monitoring. One of the models among all from study
will be the best suitable to conclude the paper. As the aim of From the flow density diagram it can be seen that queuing
the work is to monitor the traffic and provide some action by theory approach is very much related to traffic management
using the IoT devices depending on the type of traffic in real and can become helpful in reducing the congestion. The
time, we use the suitable model for monitoring the traffic on authors in [2] have provided strong foundation to encourage
highway. We also observe the traffic flow during the research in application of queuing theory in traffic
different time slot. In order to see the traffic pattern we have management and used basic queueing models to analyze the
taken one month traffic samples. We also monitor the traffic traffic data. There are various queuing models to analyses
on real time so that we can have some action to normalize the traffic data. It can be found that queuing theory has been
the traffic. exclusively used for traffic analysis at signalized
Queuing theory expresses mathematical distribution of intersections. Using queuing theory approach the various
probability and expected number of vehicles on the road. It characteristics such as waiting time, queue length can be
described the effect of different queuing parameters such as found to avoid the traffic congestion. . Depending on the
number of vehicles of the road and traffic congestion. nature of vehicle arrival rate and service rate we have
M/M/1, M/G/1, G/G/1, G/G/k and few other queuing
II. RELATED WORKS models. Queuing symbols are used to specify the nature of
queuing model.
A traffic monitoring system consists of one or more
hardware devices such as Matlab enabled Raspberry pi
TABLE I: Queuing Models
device, camera to monitor the traffic which provide some
traffic data in the form of count vs time to arriving vehicles.
We find various real life examples of queuing systems Arrival Service
such as manufacturing system, ATM service and so on. Model Procedure Procedure
Vehicles arriving on the road find the lanes busy and
generally join a waiting line which is a queuing system. M/M/1 Poisson Poisson
A. Queuing System Elements:
A queuing system is defined with three things. First, M/G/1 Poisson General
Arrival rate second is Departure rate and last is scheduling
policies.
G/G/1 General General
1) Arrival Rate:
Arrivals of vehicles may originate from one or multiple G/G/1 State General General
rode lanes referred to as the calling population. Rate of Dependent
arrival of vehicles in the monitoring region is calculated. An
example of arrival rate may be of an arrival of vehicles
during the specific time in the monitoring region by the total
number of vehicle passes from the day starts till that 1) M/M/1 Model:
movement of time. If the λ is the arrival rate then it is The M/M/1 forms the basic queue knowledge where
calculated by number of vehicle detected on that particular both arrival rate and departure rate follow Poisson
frame At time t by the total number of vehicles counted distribution and it suitable for normal traffic conditions.
from each frames from starts of the day i.e. addition of Authors in [3] optimized service rate by analyzing the traffic
vehicle count detected from starting time to the current time flow parameters using M/M/1 model. Using multiple servers
of the day. that is M/M/n model has been used by authors in [4] to
analyses traffic data at the urban intersection.
2) Departure Mechanism:
Departure rate is defines by the number of lanes (R) on 2) M/G/1 Model:
the road. If vehicles has Ti as a service time at the ith time With M/G/1 queuing model arrival rate follow
instance at the average service rate of a vehicles is µ = exponential distribution and service time is generally
1/(T(R)). The service rate is totally depends on queuing distributed. During traffic congestion the arrival rate is the
discipline that the server choses to process the vehicles. exponential but the departure or service rate is not
exponential but follows general distribution which results in
3) Scheduling Policy: queue formation. M/G/1 queueing model can be used in
Queuing discipline defines the service mechanism of lane analysing the delay in data packets in telecommunication
or in how much time the vehicle is processed on the road.
There are various scheduling policies depending upon
applications. The most among used policies are:
LIFO - vehicles are assisting in a last-in first-out manner,
FIFO - vehicles are assisting on a first-in first-out basis.

Published By:
Retrieval Number: F9397088619/19©BEIESP Blue Eyes Intelligence Engineering
DOI: 10.35940/ijeat.F9397.088619 178 & Sciences Publication
 International Journal of Engineering and Advanced Technology (IJEAT)
ISSN: 2249 – 8958, Volume-8, Issue-6, August 2019

channel [5]. In this paper we consider M/G/1 model to

analyse the traffic data as it suits for traffic congestion 5. Vehicle’s overall Waiting time in the system
scenario where vehicles enter exponentially but departure is
not exponential and can follow general distribution if
remedy is provided for congestion control. Following table
gives the list of parameters used and their notations.
6. Total Number of vehicles in the system
TABLE II: M/G/1 model parameters

Parameter Description
ℵ Arrival rate 7. Proportion of time road is idle
µ Assistance rate
ρ Server utilization
a Maximum service rate III. RESULTS AND FINDINGS
b Minimum service rate A. Traffic data acquisition
2
σ s Variance in the service time Gathering the traffic management data plays an important
role in development of ITS [6].The traffic data can be
Lq Number in the queue
collected through traffic light camera or vehicular sensing
Wq Waiting time in queue [7].In order to acquire the traffic data a smart traffic
monitoring system consisting of CCTV camera and
L Number in the system processor was designed. Data gathered on a cloud platform
c Number of servers to get the real time acquisition. Figure 2 shows the IoT
based camera which uses the Raspberrypi-3 mounted on
W Waiting time in system highway to collect traffic data. Pi-3 collects the real-time
data separately from left and right side of the road from the
For single channel queuing system if arrival rate is less camera which is mounted on it. After collecting the data
than the departure rate that is σ < 1 system is stable but if σ using camera pi sends this data to the web by using internet
> 1 queue formation happens. For an M/G/1 system has connection.
following performance parameters: B. Performance Analysis
The system monitors the number of vehicles passing on
1. Traffic Intensity the road section of highway of length 500 meters for a
The median number of vehicles being processed is the period of 24 hours on both left and right side of the road.
proportion of arrival and service rate The arrival rate λ is the number of vehicles per unit time and
depends upon the vehicles passing the road. The system
does not classify the vehicles such as car or heavy load
vehicles. The average maximum speed that can be achieved
For a steady type of system the assistance rate always by the vehicle 60 km/hr. and the average minimum speed
exceed the arrival rate and therefore value should be each during congestion can be 10 km/hr. Following figure 3
time less than 1(one). Because of this, it is popularly known shows the sample data collected on cloud platform which is
as component factor of the server (in our case it is utilization the vehicle count on left and right side of the road.
of road by vehicles).
2. Variance in the service time

3. Number of vehicles in the Queue

(a) Traffic on the road(Left)

4. Vehicles waiting time in the Queue

Published By:
Retrieval Number: F9397088619/19©BEIESP Blue Eyes Intelligence Engineering
DOI: 10.35940/ijeat.F9397.088619 179 & Sciences Publication
 IOT BASED REAL TIME TRAFFIC MONITORING SYSTEM USING M/G/1 QUEUING

The following figure 6 a and b shows the waiting time

versus arrival rate. It is observed that waiting time in queue
increases as the arrival rate increases leading to a dead lock
condition.

(b) Traffic on the road(Right)

Fig. 3: Vehicle count vsTime a) Traffic on the road
(Left) b) Traffic on the road (Right).

Figure 3 (a) shows lest side traffic is lesser compared to the (a) Traffic on the road (Left).
right side as shown in figure 3(b)

Fig. 4: Traffic Density (b) Traffic on the road (Right).

Fig. 6: Traffic Intensity a) Left side of the road b) Right
We analyzed the traffic density for week as well. During side of the road.
peak hours when commutators travel for offices high traffic
peaks are observed during morning and evening in figure 4.It
can be observed that right side of the road is heavily loaded
with the vehicles compared to the left side and can lead to a
congestion during peak hours. Whereas left side is not
utilized fully.
The arrival rate can be found based on vehicle count
collected throughout the day which is number of vehicles per
minute. Following figures 5 (a) and (b) show rate of arrival
for left and right side of the road for 24 hours.
The traffic intensity which is a measure of road utilization
is found in figure 6(a) and (b).It is observed that traffic (a) Traffic on the road (Left).
intensity on right side is more compared to the left side of the
road.

(a) Traffic on the road (Left)

(b)Traffic on the road (Right).

Fig. 7: Arrival Rate vs waiting time a) Traffic on the
road (Left) b) Traffic on the road (Right).

(b) Traffic on the road (Right).

Fig. 5: Date/Time vs Arrival Rate a) Traffic on the road

(Left) b) Traffic on the road (Right).

Published By:
Retrieval Number: F9397088619/19©BEIESP Blue Eyes Intelligence Engineering
DOI: 10.35940/ijeat.F9397.088619 180 & Sciences Publication
 International Journal of Engineering and Advanced Technology (IJEAT)
ISSN: 2249 – 8958, Volume-8, Issue-6, August 2019

Figure 7 (a) and (b) shows the graph, it explains Arrival Intersection”, in International Scholarly and Scientific
Rate vs Waiting time of vehicles, as the number of vehicle Research and Innovation 8(6) 2014.
increases on the road with respect to time, waiting time is 10.Nico Vandaele, Tom Van Woensel, “Modelling Traffic
Flows With Queueing Models: A Review”,
increases linearly. in Asia-Pacific Journal of Operational Research at
World Scientific Publishing Company and Operational
IV. CONCLUSION Research Society of Singapore.
In this paper we monitored the real time traffic data using
IoT device on highway which is also called as a smart traffic
monitoring system by collecting the data on data aggregator.
We found the various performance parameters using M/G/1
queuing model which suits to the congestion scenario
perfectly. It is clear that during peak hours when traffic flow
on one side of the road is higher compared to the other side.
One side of the road is congested compared to the other side
which leads to more travel time and waiting time.
The results suggest that underutilized side of the road
having lesser traffic can be used to divert the traffic flow of
the heavily loaded side which can be implemented with lane
management technique. The advanced technology such as
Internet of Things and Intelligent transportation system
(ITS) can definitely provide methods to implement these
remedies. This will help in reducing congestion.

V. FUNDING
This work is an outcome of the R and D work undertaken
project under the Visvesvaraya PhD Scheme, Ministry of
Electronics and Information Technology, Government of
India, being implemented by Digital India Corporation.

VI. REFERENCES
1. Yousefi, S., Altman, E., & El-Azouzi, R., “Study of
connectivity in vehicular ad hoc networks”, 5th
International Symposium on Modelling and Optimization
in Mobile, Ad Hoc and Wireless Networks and
Workshops, 2007.
2. Vandaele, N., Woensel, T. V., & Verbruggen, A., “A
queueing based traffic flow model”, Transportation
Research Part D: Transport and Environment, 5, 2000.
3. M. K. Hasan, A. H. Abdallah and N. S. Nafi, "Traffic
flow model for vehicular network”, International
Conference on Computer and Communication
Engineering (ICCCE), Kuala Lumpur,pp 738-743, 2012.
4. Xiaoyan Yang, Shuguo Yang, “The Application of the
Queuing Theory in the Traffic Flow of Intersection”,
International Journal of Mathematical and
Computational Sciences, Vol:8, No:6, 2014
5. Giambene G., “M/G/1 Queuing Theory and
Applications”, In: Queuing Theory and
Telecommunications, Springer, Boston, MA.
6. Asha Anand, Lelitha Vanajakshi, Gitakrishnan
Ramadurai, “Intelligent Transportation Systems synthesis
report on ITS including issues and challenges in India” ,
Center of Excellence in Urban Transport, December
2010
7. Hafid, S. Cherkaoui and J. Sahoo, " A novel vehicular
sensing framework for smart cities", 39th Annual IEEE
Conference on Local Computer Networks, Edmonton,
AB, pp. 490-493, 2014.
8. Mingzoo Wu and Sheng-Tzong Cheng, “Performance
Evaluation of Ad-Hoc WLAN by M/G/1 Queueing
Model,” in International Conference on Information
Technology: Coding and Computing (ITCC’05).
9. Xiaoyan Yang, Shuguo Yang, “The Application of the
Queuing Theory in the Traffic Flow of

Published By:
Retrieval Number: F9397088619/19©BEIESP Blue Eyes Intelligence Engineering
DOI: 10.35940/ijeat.F9397.088619 181 & Sciences Publication