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

net/publication/301935781

Intelligent Traffic Alert System for Smart Cities

Conference Paper · December 2015

DOI: 10.1109/SmartCity.2015.65

CITATIONS READS
7 458

3 authors, including:

Assil Ksiksi Rashad Ramzan

Georgia Institute of Technology National University of Computer and Emerging Sciences
1 PUBLICATION   7 CITATIONS    65 PUBLICATIONS   496 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Moisture Sensors View project

Testability-oriented Design Techniques for M-S/ RF Integrated Circuits View project

All content following this page was uploaded by Assil Ksiksi on 19 September 2016.

The user has requested enhancement of the downloaded file.

 2015 IEEE International Conference on Smart City/SocialCom/SustainCom together with DataCom 2015 and SC2 2015

Intelligent Traffic Alert System for Smart Cities

Assil Ksiksi, Saeed Al Shehhi, Rashad Ramzan
Department of Electrical Engineering
UAE University
Al Ain, United Arab Emirates

Abstract—Traffic accidents are one of the main causes of sudden The major downside to VANETs is that powerful hardware
death worldwide. Inadequate driver response toz changing road needs to be installed in each vehicle in the network. This results
and traffic conditions leads to a higher probability of road in an expensive system, thus inhibiting widespread deployment.
accidents. With the rise of the concept of Smart Cities, safety and Furthermore, automotive companies need to approve of any new
security are two important issues that need to be addressed. In this system before adopting it in their vehicles, which means that
paper, we propose an Intelligent Traffic Alert System (iTAS) that further standardization and security audits must be performed
warns drivers of potential dangers on the road using audio and first.
visual alerts. The iTAS consists of transmitter units installed on In this paper, we present the Intelligent Traffic Alert System
the side of the road that broadcast vital information such as the (iTAS). The system consists of FM transmitter units deployed
speed limit, road conditions, and unexpected traffic situations to
citywide that broadcast relevant information to nearby vehicles
drivers in real-time over the commercial FM radio frequency
over the commercial FM band. The transmitters broadcast a
band. On the receiver side, the system uses the radio already
present in the vehicle to deliver audio alerts that help keep drivers
message consisting of both audio and digital versions of road
informed on road and traffic conditions. The installation of an and traffic information. Drivers simply tune the car’s radio to the
optional receiver unit inside the vehicle makes it possible to display desired frequency to get the latest alerts and updates in the form
the alerts on a LCD screen in the form of visual warning symbols. of short audio messages. By installing a receiver unit inside the
A lab prototype of the proposed system was built using off-the- vehicle, drivers can view alerts and further information on a
shelf commercial components. The results acquired from testing LCD display. The information stored in each road side
the prototype were good enough to validate our approach to the transmitter can be updated in real-time over GSM by a TIC,
design. making the system scalable to large areas.
The use of commercial FM and the existing FM radio in the
Keywords—smart cities; traffic alerts; traffic alert system; safe car as a receiver makes the system low cost and free of any
driving assistance; intelligent road system; road safety needed standardization and legal approval. Unlike the systems
reported in [4] and [5], there is no signal transmitted from within
I. INTRODUCTION the car. As a result, the identity, privacy, and security of the
Road safety is a worldwide issue in general and in the United vehicle are preserved.
Arab Emirates (UAE) in particular. The UAE for example has a The paper’s structure is as follows. In Section II, a high-level
fatality rate of 12.7 per 100,000 people [1], a number that is outline of the system design is discussed. Section III provides
much higher than that of developed countries such as the UK, details on the construction of the lab prototype. Section IV
Sweden and Japan. There are many different approaches that can discusses the software written for use in the hardware prototype.
help reduce the number of accidents in any given country. A few Section V lists future directions for the research, and Section VI
popular ones are more aggressive road safety campaigns, concludes the paper.
education of future drivers in topics related to safety on the road,
and implementation of stricter traffic laws. II. SYSTEM DESIGN
The use of technology to improve quality of life and safety The primary goal of the iTAS is to keep drivers informed of
is one major goal of future smart cities. Quite a few systems have changing road and traffic conditions. The system consists of a
been proposed in the literature that aim to improve safety on the transmitter unit (TU) and a receiver unit (RU). As seen in Fig. 1,
road. In [2] and [3], a system is discussed that involves sending TUs will be installed at fixed intervals along roads in the city,
alerts to vehicles by placing passive RFID tags on the side of the while RUs will be located inside vehicles passing by. Each TU
road before points of interest, such as sharp turns, construction, will continuously broadcast information to RUs within range
pedestrian crossings etc. Physical access is required to update using FM as a simplex communication channel.
the data on a particular RFID tag. Therefore, this is not a scalable The commercial FM frequency (88-108 MHz) band has a
system which can meet the demand of a large city. In [4], the few distinct advantages. The FM band can be received by all
use of Vehicular Ad-Hoc Networks (VANETs) and a centralized vehicles with no additional modification, has a relatively good
Traffic Information Center (TIC) is proposed. In [5], a system is transmission range compared to GHz frequencies, and is not
outlined that involves the use of a completely independent adversely affected by changes in weather and air channel
VANET where the database is stored separately in each vehicle conditions.
and all computation is performed locally to remove the
dependency on an external TIC.

978-1-5090-1893-2/15 $31.00 © 2015 IEEE 165

DOI 10.1109/SmartCity.2015.65
 and digital. The first option is taken care of entirely by the car’s
built-in FM radio, making it easy for any driver to simply tune
in. A block diagram of the unit is shown in Fig. 3.
Receiving further information in digital form requires
forwarding the radio’s output to custom hardware, which
consists of a Arduino microcontroller board interfaced with a
LCD display. The audio is sampled by the firmware running on
the Arduino and, if any embedded data is detected, the firmware
will use Goertzel’s algorithm to recover the data embedded in
the received audio message. Finally, the retrieved data is
displayed on the LCD screen. The hardware should ideally be
housed in a clean and portable enclosure. The unit should also
be easy to place within the vehicle for convenience. As far as
Fig 1. The system overview of iTAS. The TU broadcasts alerts at some power goes, both the Arduino and LCD display can be easily
frequency (92 MHz in the prototype), and the car radio and optional RU deliver powered by 2 or 3 AA batteries.
the alerts to the driver.
III. PROTOTYPE OVERVIEW
A. Transmitter Unit (TU) We built a prototype of the iTAS to test the core idea and
The TU’s task is to broadcast a signal containing both audio operate the system in a realistic setting. As seen in Fig. 4, the
and digital versions of pertinent traffic information over prototype consists of a basic transmitter unit attached to the rear
commercial FM. It will need a power source, preferably one that of a speed limit sign replica, and a receiver unit. The system was
is renewable and cheap. Additionally, the information stored on built using off-the-shelf components to reduce implementation
any single TU must be capable of being updated remotely via time and overall cost.
GSM. Based on these requirements, our proposed transmitter The power module consists of three 6V solar panels
unit is shown in Fig. 2. The TU consists of several modules that connected in series and two 6V lead-acid batteries. The panels
work in concert to broadcast information to vehicles in range. and batteries are connected to a battery charger unit to provide
The heart of the transmitter unit is the microcontroller backup in case of poor sunlight.
module. It has two main responsibilities. Firstly, it retrieves The prototype lacks some of the features outlined in the
stored traffic and road information from the on-board non- previous section. Instead of a microcontroller that generates the
volatile flash memory. Secondly, the firmware loaded onto it audio independently, we used a off-the-shelf car FM transmitter
builds audio and digital representations of the information, with a built-in SD card slot. Every time we needed to update the
merges the two formats, and finally passes them as an audio information for testing, we wrote a new MP3 file to the SD card.
signal to the FM chip. This process is repeated every t seconds The MP3 file was generated using custom MATLAB code. We
to form a continuous broadcast. In addition, any incoming did not include a GSM module in the prototype.
updates to the information (traffic jam, accidents, etc.) will be The receiver unit is similar in structure to the one proposed
written to memory for the next iteration of the broadcast. in the previous section. It consists of a portable FM radio
A GSM module will allow the transmitter to receive updated connected to a Arduino microcontroller that is used for
traffic and road information for alerts over-the-air. The module extraction of the embedded data. A LCD display interfaced with
will then forward the newly received traffic data to the the Arduino allowed us to display small icons depending on the
microcontroller, which will first extract relevant information, information retrieved from the digital data. We wrote the
then write it to the flash memory. firmware for the microcontroller in standard AVR C.
The radio module consists of a FM radio chip connected to In the next section, we will provide an overview of how the
a dipole antenna. The chip is interfaced with the microcontroller. MP3 was generated using MATLAB, as well as an outline of
When the FM chip receives a signal, it will modulate it using how the Arduino firmware we wrote works.
FM and forward it to the antenna for transmission. To keep
things simple, the system will transmit using mono FM. IV. SOFTWARE IMPLEMENTATION
Finally, to power the transmitter, we propose the use of a There are two parts to the software side of the prototype:
solar panel and voltage regulator interfaced with a Li-ion
battery. The power consumption of the TU depends mainly on 1. Generation of the MP3 file containing both audio and
the GSM module and FM radio chip. A typical FM transmitter embedded visual alerts.
IC, such as the NTE7200, consumes around 4.5 mW [6], 2. Retrieval of the embedded data frame from the incoming
compared to at least 30 mW for a RF transceiver IC. As a result, FM audio broadcast.
if the design is optimized, the capacity of the battery and the
solar panel rating can be quite low. The first part was implemented in MATLAB, while the
B. Receiver Unit (RU) second part was implemented in AVR C. The source code for
both parts is available on Github under the MIT License [6].
The receiver unit provides two options for the user to
consume road and traffic information: audio only, or both audio

166
 Fig 2. Transmitter unit of iTAS. In the proposed design, the SD card is replaced by flash memory, the MP3 player is not used, and the battery voltage is 5V. The
prototype on the other hand lacks a GSM module, and is supplied by a 12V battery, which is why the voltage regulator is included.

Fig 3. iTAS receiver unit. Audio stream from the radio is forwarded to the microcontroller for further extraction.

The transmitted data frame structure is shown in Fig. 5. It is signal is repeated to fill up the entire duration of audio to ensure
32 bits long, including the start and stop sequences to simplify faster delivery to the receiver. The audio and data signals are
recovery at the receiver - 1110 and 0101, respectively. We kept added together to provide the final alert signal, as shown in Fig.
plenty of empty values for future modification. 7. Finally, the raw audio is encoded as an MP3 using FFmpeg
[7], and then copied onto the SD card for use by the FM
A. MP3 File Generation transmitter.
First, a sample voice recording of a speed limit alert (“The
speed limit is 60 km/h”) is created with some fixed duration at a B. Data Frame Retrieval
specified sampling frequency – we chose a duration of 2 seconds The receiver code is more complex. Since the transmitter is
and a sampling frequency of 44 kHz. This was selected to ensure always broadcasting, the receiver does not know when the
that the embedded data signal is inaudible when played on the transmission begins. In addition, it does not know at what time
car radio. the frame actually starts in the audio signal.
Next, the data frame is modulated using FSK at just under To solve the first issue, the microcontroller’s analog-to-
half the sampling frequency. More specifically, we chose a mark digital converter (ADC) is continuously sampled until a voltage
frequency of 21 kHz to represent binary 1, and a space frequency greater than some threshold determined by the radio is sensed
of 20 kHz to represent binary 0, as shown in Fig. 6. The data (in the lab, this was around 0.2V).

167
 92 MHz
FM Antenna

Three 18V, 3W Solar

Panels

Battery Charger
& Regulator

12V
Batteries

92 MHz
MP3 Player &
FM Antenna
FM Transmitter

Fig 4. iTAS hardware prototype built in the lab for testing.

Fig 6. Frequency spectrum of the modulated signal. The audio can be seen at
lower frequencies, while the embedded data is in the 20-21 kHz range.
Truck Max Min Warning Traffic CRC
Sign Speed Speed Sign Once the algorithm terminates, the start and stop sequences
[1] [4] [3] [6] [2] [8] are used to find the data frame. If a frame is found, the CRC code
Fig 5. Frame structure with field sizes in bits. Start and stop sequences are not is checked. In case of a mismatch, the entire process is repeated.
shown. In our tests, one run of the algorithm takes approximately t ms,
which for our case can be found using the following relation:
Once this happens, the firmware enters the main loop. The

 
ADC is further sampled for two entire frame lengths to  = 2  × 32 ×5 = 320 
guarantee at least one frame is captured. This addresses the 

second issue mentioned above. Given the approximate range of transmission of the
Unfortunately, the ADC interrupt timings provided by the prototype’s TU (around 500 m), the value of t computed above,
ATmega328P proved to be inaccurate for our choice of sampling and the maximum expected vehicle speed, we computed n, the
frequency. We tried modifying the sampling rate of the maximum number of attempts the receiver has to recover the
generated MP3 file to accommodate for this, but it could not be frame from the broadcast:
decreased due to limitations in the MP3 specification. Instead,
the ADC pin was manually sampled using tight loops. Each loop 0.5 3600
= × = 35.156 ≈ 35
was optimized based on the clock cycles required for each 160 0.32
instruction executed within it (jumps, FP multiplications, The receiver therefore has approximately 35 attempts to
addition, etc.). The timing for the instructions was acquired from retrieve the frame in the worst case. At lower speeds, the number
running the code in Atmel’s simulator for the ATmega328P. of attempts available will increase.
The actual recovery of the embedded frame bits is done using An additional piece of code was written to decode the
Goertzel’s algorithm, a modified form of the DFT with contents of the frame (speed limit, traffic conditions, etc.) and
extremely low memory and computation requirements. The idea display the information on the LCD screen. A sample of a
is to compute the power of the mark frequency in a time slice of generated speed limit symbol can be seen in Fig. 8.
the signal equivalent to 1 bit. The power computation is
performed using the current sample along with the computed V. RESULTS AND FUTURE WORK
values of the last two samples. The prototype and the software detailed above allowed us
At the end of the time slice, if the total power is above a pre- gain further insight into whether or not the system is feasible in
computed threshold, the bit is read as a 1, and a 0 otherwise. The real working conditions. We identified two major issues that
bit is finally stored in a bit array. The storage required for the require improvement before the system can be commercialized.
algorithm depends on the two samples in floating-point format Firstly, the microcontroller used, the ATmega328P, was not
and the array of bits of length twice the frame size (64 bits in our powerful enough to receive transmissions at a high sampling
case). rate. Decreasing the sampling rate proved ineffective, and led to
significant audible noise caused by the embedded data frame.
The obvious solution is to use a more capable processing unit
such as a 32-bit ARM microprocessor. Secondly, installing
transmitter units across a large city and maintaining them over a

168
 long period of time is a very difficult and costly job. In countries Conference, 2003. VTC 2003-Spring. The 57th IEEE Semiannual, 2003,
with hot climates, such as the UAE, the transmitter unit must be vol. 4, pp. 2442 – 2446.
built using industrial grade components to avoid overheating and [6] “iTAS on Github.” [Online]. Available:
https://github.com/Cyph0n/iTAS.
malfunctions.
[7] “FFmpeg.” [Online]. Available: https://www.ffmpeg.org/.
We are working on a new system that addresses these two
major issues, as well as other smaller issues we encountered [8] A. Botta, A. Pescapé, and G. Ventre, “Quality of service statistics over
heterogeneous networks: Analysis and applications,” Eur. J. Oper. Res.,
while building and testing the prototype. The new system 2008.
consists of a centralized server that houses traffic and road data
for the entire city. An app installed on the driver’s smartphone
periodically pings the server over 3G/LTE/WiMAX to receive
both visual and audio alerts on traffic conditions in near real-
time.
One of the main requirements for the updated system was to
design a protocol that minimizes data usage for the user, but at
the same time is easy to process on both ends. For this reason,
we opted to use JSON (JavaScript Object Notation) on top of
UDP. Our testing shows that this scheme uses 200 bytes per
exchange on average. From a QoS standpoint, UDP is a good
choice for reliable communications due its relatively predictable
throughput, even when operating over low-bandwidth protocols
such as GPRS [8].
VI. CONCLUSION
In this paper, we outlined the design of the Intelligent Traffic
Alert System (iTAS), a smart traffic alert system that aims to
help keep drivers updated on road and traffic conditions. The
system consists of two main components: a transmitter unit and
a receiver unit. Transmitters are installed on the side of the road
and broadcast information relevant to their location over
commercial FM. Drivers can tune in to the specified broadcast
frequency and listen to audio alerts without installing or
Fig 7. Modulated signal consisting of the linear sum of the audio message and
purchasing any additional hardware. The receiver unit is
FSK modulated data for embedding.
optional, but can provide more visual information and alerts
extracted from the broadcast if installed. The data stored by each
transmitter can be updated remotely over GSM.
We built a functional prototype of the system to test it and
determine if it’s feasible. In the future, we plan on adding GSM
functionality to it, and replacing the microcontroller used to one
that is significantly more powerful to avoid the issue
encountered at the receiver end. We are also working on a new
system that is much easier to deploy at scale. It consists of a
centralized traffic information server and an app installed on the
driver’s smartphone. Fig 8. Arduino programmed to show a speed limit of 60 on the LCD display.
REFERENCES
[1] A. S. Mohammed, “Leading cause of road traffic morbidity and mortality
in the United Arab Emirates (UAE) and the main adaptations to reduce
it,” in Innovation Arabia 2015, p. 90.
[2] H. Zoghi, M. Tolouei, K. Siamardi, and P. Araghi, “Usage of ITS in the
in-vehicle signing system with RFID tags and vehicle routing and road
traffic simulation,” in IEEE Conference on Intelligent Transportation
Systems, Proceedings, ITSC, 2008, pp. 408–413.
[3] A. Paul, R. Jagriti, N. Bharadwaj, S. Sameera, and A. S. Bhat, “An RFID
based in-vehicle alert system for road oddities,” in 2011 IEEE Recent
Advances in Intelligent Computational Systems, RAICS 2011, 2011, pp.
19–24.
[4] G. S. Khekare and A. V. Sakhare, “A smart city framework for intelligent
traffic system using VANET,” in 2013 International Mutli-Conference on
Automation, Computing, Communication, Control and Compressed
Sensing (iMac4s), 2013, pp. 302–305.
[5] L. Wischhof, A. Ebner, H. Rohling, M. Lott, and R. Halfmann, “Self-
organizing Traffic Information System,” in Vehicular Technology

169

View publication stats