A Minor Project Report

On

Smart-Highways-Internet-of-Things
Submitted in partial fulfillment of requirements for the award of the
Degree of

Masters of Technology

In

Digital Communication

Under the guidance of

Prof. Rashmi Gupta

Submitted By

VIJAY RATNA PRABHAKAR

(00610100719)

DEPARTMENT OF ELECTRONICS AND COMMUNICATION

AMBEDKAR INSTITUTE OF ADVANCED COMMUNICATION
TECHNOLOGY & RESEARCH

GEETA COLONY, DELHI

JUNE, 2020
Objective
This is the implementation of a robust system for real time monitoring of highways.
Various sensors have been deployed on highway which report the status of the
same to a central control station.

Introduction

The highway monitoring system is divided into different modules:

 Smart fencing
 Sound signal based accident detection
 Weather detection
 toll tax collection
 Activity based road light
 Collision avoidance system

Smart Fencing for Landslide and Accident Detection

Highways built on hilly terrains have very frequent sharp turns which might
become hazardous for vehicles. Now and then we hear incidents of vehicles falling
off the cliff resulting in casualties as help could not be offered on time. The
probability of detecting such a mishap and sending help on time is very less as
there is no preventive system for that.

Another aim of this module is to detect landslides which are too very common in
hilly terrains. The same method will be used to detect the impact of debris on the
fence so that the alert can be broadcasted.

Woking Principle

 The idea proposed is that accelerometers will be mounted on the fence

at particular intervals on the fence running throughout the highway.
 The Arduino on the sensor side will receive the acceleration of the
fence in X, Y and Z direction at a particular sampling rate.
 This vector will be sent to control centre using wireless
communication. At the control centre magnitude of the acceleration is
calculated.
 If there is an impact on the fence, there will be a sudden rise in the
magnitude of the acceleration. If the rise is greater than the preset
threshold, the impact will be reported and will initiate the process to
send help at the location.
Hardware Requirement

• ADXL345 MEMS accelerometers

• ESP32/Arduino controller

• Communication through wireless protocol (LoRa WAN)

• Solar panel and battery

• MEMS Microphone

Smart Street Light

Streetlight consumes huge amount of generated power. High need to save unused
power supply on highway. Streetlights in areas with sparse activity are source of
major power drainage. Here is proposed an activity-based rood light. Streetlight
consumes huge amount of generated power.

Working Principle
• The module consists of an IR transmitter mounted on the base of the
streetlight, an IR sensor on the opposite side of the road which is connected
to an intermediate comparator circuit which is then further connected to
Arduino.

• During night, when there is no activity, the sensor receives the IR signal
continuously and hence, the comparator sends no signal to the Arduino
board. In this case, only 1 out of every 10 streetlights are switched on.

• Whenever there is an activity on the street, there is discontinuity in the

received IR signal and a signal is sent the Arduino board. In this case, 5
streetlights on either side of the sensor are switched on by the Arduino
board, and remain so for a fixed delay as to provide smooth transition for a
moving vehicle.
Hardware Requirement

• LED Array

• PIR motion sensor /IR/Ultrasonic

• LDR Sensor

• Arduino

• Solar panel and battery

Audio Signal Based Accident Detection

• Reduction of highway accident loss

• Immediate help at accident location on highway

• Reducing the number of highway accidents

• Leading to smart highway

Working Principle

• The input to the system is a 3-s segment of audio signal.

The system can be operated in two modes: the two-class
and multiclass modes. The output of the two-class mode is
a label of "crash" or "noncash". In the multiclass mode of
operation, the system identifies crashes as well as several
types of non-crash incidents, including normal traffic and
construction sounds. The system is composed of three main
signal processing stages: feature extraction, feature
reduction, and classification.

STEP 1: Network of audio sensors installed on the highway

collect the data which is first pre-processed, and their intensity is
calculated by local (nearer to sensor) Arduino.
STEP 2: If the intensity of captured data is higher than threshold
then Arduino sends this entire data frame to command centre
using wireless communication.

STEP 3: At command centre, from given data audio signal is

reconstructed. After reconstruction of the signal one can do the
resampling on desired rate (in this case it is 16 kHz).

STEP 4: After Pre-processing and resampling of the data,

Spectrogram analysis is done for accident and non-accident
audio signals.

STEP 5: Then MFCC of the data frame is calculated and KNN

classifier is used for classification. Training has been done with
numerous accidents and non-accident sound samples.

STEP 6: Classifier is providing very high efficiency classification.

Mel Frequency Cepstral Coefficient (MFCC)

Coefficients that collectively make up an MFC. They are derived

from a type of cepstral representation of the audio clip

The MFCC feature extraction technique basically includes

windowing the signal, applying the DFT, taking the log of the
magnitude, and then warping the frequencies on a Mel scale,
followed by applying the inverse DCT.

K-nearest neighbours (KNN)

KNN algorithm is a type of supervised ML algorithm which can

be used for both classification as well as regression predictive
problems. However, it is mainly used for classification predictive
problems in industry.

The following two properties would define KNN well −

Lazy learning algorithm − KNN is a lazy learning algorithm

because it does not have a specialized training phase and uses all
the data for training while classification.
Non-parametric learning algorithm − KNN is also a non-
parametric learning algorithm because it doesn’t assume
anything about the underlying data.

Figures below show spectrograms of accident and non-accident sound signals.

Hardware Requirement

• Microphone

• ESP32/Arduino

• Communication protocol

• Solar panel and batteries