1

Smart Garbage Management System

A Report Submitted
in Partial Fulfillment of the Requirements

for the Degree of

Bachelor of Technology
in
Computer Science & Engineering

by

Sujeet Kushwaha
Yash Verma
Surbhi Mayank
Vishal Verma
Revadi Sai Eswar

to the

COMPUTER SCIENCE AND ENGINEERING DEPARTMENT

MOTILAL NEHRU NATIONAL INSTITUTE OF TECHNOLOGY

ALLAHABAD, PRAYAGRAJ
 2

UNDERTAKING

I declare that the work presented in this report titled “Smart

Garbage Management System”, submitted to the Computer Science and
Engineering Department, Motilal Nehru National Institute of Technology,
Allahabad, Prayagraj, Uttar Pradesh (India) for the award of the Bachelor of
Technology degree in Computer Science & Engineering, is my original work.
I have not plagiarized or submitted the same work for the award of any
other degree. In case this undertaking is found incorrect, I accept that my
degree may be unconditionally
withdrawn.

November 2021
Prayagraj ,U.P. (India)

(Sujeet Kushwaha, 20184098)

_______________________________
(Yash Verma(20184154) )
_______________________________
(Surbhi Mayank (20184061))
_______________________________
(Vishal Verma (20184040))
_______________________________
(Revadi Sai Eswar (20184148)
 3

CERTIFICATE

Certified that the work contained in the report titled “Smart Garbage
Management System", Sujeet Kushwaha (20184098), Yash Verma (20184154),
Surbhi Mayank(20184061), Vishal Verma (20184040), Revadi Sai
Eswar(20184148) has been carried out under my supervision and that work
has not been submitted elsewhere for a degree.

Dr. Ashish Kumar Maurya

Computer Science and Engineering Dept.
M.N.N.I.T, Allahabad
November 2021
Prayagraj, Uttar Pradesh (INDIA)
November,2021
 4

PREFACE

Things that are connected to the internet and those devices controlled from
the internet are called Internet Of Things. So, basically in this system, the
bin is connected with the internet to display the exact information about
the dustbin level, to which area it belongs, the shortest distance to reach
that particular area. As the population nowadays is growing rapidly, this
smart garbage monitoring system will help to decrease the waste pollution
and spread of the deadly diseases in the cities.
 5

ACKNOWLEDGEMENT

The completion of this project required a lot of effort, guidance, and support
from many people. We feel privileged and honored to have got this all along with
the development of the project. We would like to express our gratitude to our
mentor, Dr. ASHISH KUMAR MAURYA (Professor) for his perennial support,
guidance, and monitoring throughout the making of this project. We would also
like to acknowledge and thank our professors, colleagues, and seniors for
supporting us and enabling us to successfully complete our project.
 6

CONTENTS

Introduction 7
1.1 Motivation 7
1.1.1 Lack of the proper monitoring system 7
1.1.2 Lack of optimal routes for collection 8
1.2 Objective 8

Preliminaries and Related Work 9

Proposed Work 11
3.1 Smart Bin Implementation Details 12
3.2 Admin Dashboard 14
3.3 User Application 16
3.4 Waste Collection Route Generation 17
3.4.1 Problem Definition 17
3.4.2 Objective 17
3.4.3 Dataset 17
3.4.4 Approach 18

Experimental Setup and Results Analysis 22

4.1 Experimental Setup 22
4.2 Result 23

Conclusion and Future Work 29

5.1 Conclusion 29
5.2 Limitations 29
5.3 Future Work 30

References 31
 7

Chapter 1

Introduction
In today's fast-paced life individuals scarcely have time to stop and configure
things manually and hence the idea of automation is by and large broadly
embraced. Lack of infrastructure and unsustainable practices have made waste
management worse, which leads to environmental contamination. The current
method of waste management requires workers to manually check the dustbins
and empty them. The garbage vehicle visits each and every dustbin to collect
waste on fixed schedules. Some of these dustbins get rarely filled and may not
require attention each and every time the garbage vehicle goes for collection.
The path followed by vehicles to visit these dustbins is random and inefficient.
This method of waste collection is cost-inefficient and time tedious. Authorities
should know the real-time status of the bins in order to efficiently schedule the
waste collection. The workers going for the collection should know the optimal
route to visit the bins in order to save fuel and time. The people living in the
locality had to manually locate an empty dustbin to dump the garbage. It would
be better if they know the nearest empty dustbin from their location so that they
could dump their garbage without overflowing any filled dustbin.

1.1 Motivation
1.1.1 Lack of the proper monitoring system

The conventional method of manually monitoring dustbins does not take into
account the real-time status of the bin. The workers have no prior knowledge of
the waste level in the bins. They visit each and every dustbin to collect wastes
which leads to a lot of human effort, cost, and time. The people living in the
locality have to manually locate an empty dustbin to dump garbage which is
tedious and time taking.
 8

1.1.2 Lack of optimal routes for collection

The vehicles going for the waste collection visit bins in any order which is not
efficient both in terms of cost and time.

1.2 Objective
We aim to build a fully automated garbage monitoring system that uses IoT and
cloud computing to monitor the levels of waste in the bins and provide an
efficient and optimal route for waste collection. In this system, each bin is
installed with ultrasonic sensors, Arduino Uno, and ESP 8266 wifi module. They
capture the real-time level of the bin and transmit this data to the cloud. A cloud
platform is used here to store and visualize the data fetched from all the nodes.
The authorities are provided with an application through which they can manage
the installed bins within their locality more efficiently. They would be able to
generate optimal routes and allot these routes to vehicles for waste collection.
These vehicles are installed with GPS and their live location is shared with the
authorities. The people living in the locality are provided with an application
through which they can get the status of the bin located in their locality and can
fetch the location of the nearest empty dustbin from their location.
 9

Chapter 2

Preliminaries and Related Work

Currently, smart cities are highly populated due to rapid growth of population in
urban areas. This is causing an increase in waste disposal and causes harmful
effects to the human beings and environment. Due to lack of proper monitoring
of waste, the dustbins appear to be flooded and spill out onto the surrounding,
causing deadly diseases in the cities.

IoT
It describes the connection of hardware devices mounted/embedded with
sensors over internet. Devices connected to each other or any common server
with the help of internet.

Thingspeak
It is an IoT analytics platform service which allows us to store,visualize as well as
analyze the data in the cloud.

NodeMCU ESP8266
It is an open source both hardware and software development which is used for
the WiFi connection.

Ultrasonic Sensors
It consists of two ultrasonic transducers. The one acts as a transmitter that
converts electrical signals into 40KHz ultrasonic sound pulses. The receiver
listens for the transmitted pulses.
 10
Breadboard
Where every device connection exists. It is basically used for the connection
purpose.

Firebase
It is a google-backed application development software that enables developers
to develop android, IOS, web apps etc. It basically secures access to the database
directly from client-side mode.

The authors Samuel et al. [4] have proposed a sensor GSM integrated model for
smart garbage bin management implementation.
The authors Narayan et al. [5] presented both hardware and software approaches
to this smart waste management. The proposed system presents the use of IoT
technology and cloud computing that constantly monitors the level of waste in
the bin and sends it in real time. They have discussed how data analytics can be
applied on such systems.
The authors Tony et al. [6] have proposed a novel battery level aware clustering
family of schemes which is called BLAC and which also they have implemented
for a network of smart non-overlapping multi-hop clusters with energy concerns
that extends the lifetime. They have used ZigBee and for long range
communication with the base station over the GPRS.
The authors S. Vinoth et al [7] have presented an IOT based shrewd waste clean
administration framework which looks at the waste level over the dustbins by
utilizing sensor frameworks. When it was recognized promptly this framework
changed to concern approved through GSM/GPRS.
The authors Thangavel et al. [8] have proposed IOT-edged nodes that are
installed in the waste dustbins. Basically here they have proposed the system
where real-time data is sent to the cloud and from there it goes to various
applications. The cleaning workers reach to each waste bin to check whether the
level of garbage bin is filled or not which leads to excessive fuel consumption and
unnecessary man-power wastage. And because of this it may also happen that
the garbage bins are not cleared at all for a longer period of time.
 11

Chapter 3

Proposed Work
Our project tries to build a system that uses IoT and cloud computing to monitor
the levels of waste in the bins and provide an efficient and optimal route for
waste collection.Each bin is installed with an ultrasonic sensor and nodemcu
ESP8266 wifi module.The level of the garbage in the bin is then stored in the
cloud from where the data is shown to authorities and users on dashboard and
android app respectively.Figure 1 describes the architecture of the proposed
system. Following are the major steps involved:

● We collect the data from the dustbin whether it is full or not, its current
level of garbage.
● We feed the level of the dustbin into the cloud (ThingSpeak IoT cloud
platform) to store and visualize data obtained from the bins.
● We fetch the real-time data from the cloud into the admin dashboard which
is created to manage the installed bins. Through the dashboard, authorities
can generate optimal routes for the waste collections. They can also allot
these routes to the specific vehicles and track their real-time location while
vehicles go for the collection.
● We give the information of the nearest empty dustbin to the users where
they can dump the garbage.
 12

Figure 1:Architecture of the proposed system.

3.1 Smart Bin Implementation Details

The System uses a dustbin on which 4 Ultrasonic sensors are placed on the cover
side of the bin. Ultrasonic sensor provides a measuring range of 2 cm – 400cm. The
accuracy of sensors can reach up to 3mm. All 4 sensors are mounted on a disk
vertically to each other with equal distance from the center of the bin. Sensors are
connected to nodemcu microcontrollers. System is powered by a 5 V battery
connected to nodemcu.

Components Used
● Ultrasonic Sensors
● NodeMCU
● Connecting Wires
● Breadboard
 13

Working of Ultrasonic Sensor

We determine the level of filled garbage in the dustbin using ultrasonic sensors.If the
height of the bin is h and the height of garbage filled is g then height left is h-g.
● The ultrasonic sensor’s Trigger (Tx) sends pulses of ultrasonic sound towards
garbage every 10 microsecond.
● Receiver of senser (Rx) receives it. Time taken by pulse to go and coming back
is calculated.

Speed of sound in air = 0.034cm/ microsecond

Distance = speed * time taken by pulse coming back from garbage to sensor

D = 0.034 * sensor data/2

Now code written on Arduino ide is compiled and uploaded on the nodemcu system.
Once code is uploaded, sensors start sending data to nodemcu and nodemcu starts
uploading that distance on thingspeak using API.
 14
Figure 2:Smart Bin Circuit Diagram

3.2 Admin Dashboard

A dashboard for authorities to ease the management of bins. Through this they
can add the bins that they want to deploy and thus can monitor the filled levels of
bins in their locality.Here is the detailed features of the dashboard:
● Login/Signup for admin.
● Install, edit and uninstall a bin.
● Register a vehicle.
● Get the real-time status of bins and obtain their visualization on the google
map.
● Get the list of priority bins requiring collection and generate optimal routes
for their collection.
● Allot routes to the vehicles and show their real-time location on the admin
dashboard.

Figure 3: Database Schema

 15

Figure 4:Flow Diagram

Figure 5: Decision Logic For Monitoring Bins

 16

The activity diagram Figure 4 describes the logical decision involved in deciding
the priority bins whose collection needs to be scheduled.The bins whose last
collection has exceeded 2 days or have reached a threshold level are scheduled
for the collection.

3.3 User Application

Garbage management system proposed earlier does not give a move for users to
get all nearby bins status, so an application for users to get the location of
nearby dustbins . The user gets access to states of the status of bins. These bin
states are realtime status of bins, which has been sent to thingspeak by bins. The
data shown in application is updated by bins themselves after a regular interval of
time. It makes the process simple and easy and every user has an eye over the
nearby bins, and dumping garbage becomes reliable.Following are the features of
the app:
● Users can filter out bins on the basis of distance(radius) .
● Users can get the distance or directions to the nearest bin available.
● User can refresh the bins data regularly to get the real time status
 17

3.4 Waste Collection Route Generation

The vehicle routing problem with capacity constraints is used for dynamic route
generation.This problem is an extension of TSP(Traveling Salesman Problem) with
multiple vehicles taken into consideration instead of one,It also takes into
consideration the capacity constraints of the vehicle.
3.4.1 Problem Definition
● We have bins with capacity located at location , .
● We have a depot located at location , .
● We have vehicles each having capacity .
● Each bin is connected to another bin by an edge with a cost equal to the
distance between the two bins.
● Each vehicle starts and ends at the depot.
● Load of each vehicle should not exceed its capacity.
● Each bin is collected by exactly one vehicle.
● Load of a bin does not exceed the capacity of a vehicle.

3.4.2 Objective
Our objective is to reduce the total distance traveled by vehicles.

3.4.3 Dataset
We have used a benchmark dataset from VRP-REP (Vehicle Routing Problem
Repository). Each of the datasets includes the following attributes:
● Name of the dataset.
● Type of the dataset(CVRP).
● Dimension (n)
● Edge Weight Type(EUC_2D)
● Capacity of each vehicle
● Coordinates of n nodes
● A load of each node
● Depot
 18

For each dataset, the benchmarked solution is also available on VRP-REP which
can be used to verify the performance of our results. The benchmarked solution
contains the optimal distance and a minimum number of vehicles needed for
collection.Dataset consists of 100 instances with dimension varying from 10 to
1000.

3.4.4 Approach
We are using cluster first route second algorithm[17] to solve CVRP(capacitated
vehicle routing problem).It is a two phase algorithm:clustering followed by
routing.
● Clustering
○ Build the cluster using clustering algorithms.
○ Check the capacity constraint of the cluster.
● Construct a feasible route using TSP’s heuristics which can be further
improved using metaheuristics.

Since we are having multiple vehicles and each vehicle has a capacity constraint
on the load it can carry.We need to group all bins into clusters such that the total
load of each cluster does not exceed the capacity of the truck.We are using
sweep and nearest neighbor clustering methods as they are well suited for this
problem.Since the number of clusters to be formed is not known in advance we
needed to use algorithms which do not require the number of clusters as input.

In this analysis we are using christofides,nearest neighbor and greedy as tsp

heuristic to obtain the nearest optimal solution in a reasonable time.The obtained
solution is further optimized using metaheuristics(Guided Local Search,Tabu
Search,Simulated Annealing).
 19

Clustering Methods
We have used the following clustering methods to divide the nodes into clusters.
1. Nearest Neighbor Algorithm
● Randomly choose any unvisited node and assign it to a new cluster.
● From this node determine the nearest node and add it to the cluster if
the total load does not exceed the capacity of the cluster(equals to the
capacity of the vehicle).
● Repeat the above steps until all nodes are visited.
2. Sweep Algorithm
● Determine the polar angle of each node with respect to depot using
the formula 𝜃 .
● Sort the nodes in increasing order of their polar angle.
● Start visiting each node in increasing order of their polar angle and
keep adding them into the cluster until a load of all the nodes exceeds
the cluster capacity.
● Create a new cluster and repeat the above steps.

Figure 6:Clustering using nearest neighbor Figure 7:Clustering using sweep

 20

TSP Heuristics Algorithm

Following TSP Heuristic methods have been used

● Nearest Neighbor
○ Start from any random node.
○ Visit the nearest unvisited node.
○ Keep visiting the nearest node from the current node till all the nodes
are visited.
● Greedy
○ Sort all the edges based on the distance between them.
○ Select the shortest edge and add it if it does not create a cycle or
increases the degree of any node to more than 2.
○ Repeat step 2.
● Christofides
○ Build a minimum spanning tree from the set of all the nodes.
○ Select those nodes which are having odd degrees.
○ Create a minimum weight perfect matching.
○ Create a graph by adding edges of minimum weight perfect matching
with the minimum spanning tree to obtain a multigraph where each
node has an even degree.
○ From this graph create an eulerian circuit.
○ Create a hamiltonian circuit by skipping the repeated vertices.
 21

Meta Heuristics Algorithm

Following metaheuristics algorithms have been used:
● Guided Local Search
It is a penalty based method that works on the top of local search
algorithms.Penalties is used to help local search algorithms escape from
local minima.The objective functions are updated with the use of penalties
which allow the local search algorithm to escape out of the local optimum.
● Tabu Search
In this algorithm we escape local optimum by making some of the features
tabu i.e solutions with these features will not be produced for a certain
time.It consists of two phases:diversification to escape out of local minima
and explore other parts of search space and intensification to search in
detail a promising neighborhood.
● Simulated Annealing
In this algorithm we decide if a given solution in the neighborhood of
solutions should be accepted or not based on the probability which changes
with time.Initially the probability is such that at the beginning of the
algorithm a large part of the search space can be reached and converges to
0 in the end.
 22

Chapter 4

Experimental Setup and Results Analysis

4.1 Experimental Setup

● Arduino IDE
● ThingSpeak IoT Platform
● Visual Studio
● Django
● HTML,CSS,Javascript
● Mapbox API
● Firebase
● Matplotlib
● Numpy Library
● TSP solvers from or tools.
● Metaheuristics Library from or tools.
 23

4.2 Result
Figure 8 shows the average value of distance of ultrasonic sensors.The distance
thus measured is uploaded to the thingspeak cloud via its write API. Thingspeak
is connected to mobile, web applications, and bins.The distance measured is
clearly reflected in the thingspeak channel as shown in figure 9.In thingspeak the
field chart displays the bin levels varying with time.

Figure 8:Result on Arduino IDE

 24

Figure 9:Thingspeak Report

 25

Figure 10:Setup with bin

 26

Install bin All bins

Real Time Status Priority Bins

Registered Trucks Optimal Routes

Figure 11:Complete admin dashboard

 27

Setting range showing nearby available bins Directions to nearest bins

Figure 12:User Application

Table 1 : Computational Result of TSP heuristics with clustering algorithms.

TSP Nearest Neighbor Clustering Method Sweep Clustering Method
Heuristics
Average Cost Average Gap(%) Median Gap(%) Average Cost Average Gap(%) Median Gap(%)

Nearest 73001.65 19.455035230 18.464893352 81858.58 25.350601113 24.261018316

Neighbor

Greedy 72995.02 19.430423118 18.707867578 81855.29 25.333829703 24.251591839

Christofides 72981.36 19.378715521 18.468691164 81857.83 25.353327240 24.221823426

The results in terms of average cost, average gap, and median gap with respect to
the optimal value are given in table 1. Nearest Neighboring Clustering Method
with Christofides heuristic gives the best result.
 28

Table 2:Computational result of metaheuristics with nearest neighbor clustering

method and christofides .
Metaheuristics Nearest Neighbor Clustering Method With Christofides

Average Cost Average Gap(%) Median Gap(%)

Guided Local Search 72906.85 17.175539945 16.24552787

Tabu Search 72955.95 17.314481566 16.443221553

Simulated Annealing 72975.59 17.36261308 16.492127516

As we can see from table 2 the nearest neighboring clustering methods with
Christofides heuristic and guided local search metaheuristics gives the best
result in comparison with other algorithms.This combination of algorithms have
been used to generate optimal routes for garbage collection in our system.
 29

Chapter 5

Conclusion and Future Work

In this report we have proposed a system through which waste monitoring and
management can be efficiently achieved.

5.1 Conclusion
We have successfully implemented smart bins and are able to send their real time
level data to the cloud. We have also built a web app for admin and an app for
users . The applications are affordable and easy to use. The problem of excess
accumulation of garbage from bins across streets,houses can be minimized. The
user application can be easily installed by people and then they can dump their
garbage in an efficient and economical manner. We have also discussed the
various approaches to generate optimal routes for collection and achieved an
average gap of 17.17% with respect to the standard optimal solution.

5.2 Limitations
● The clustering methods discussed in generating optimal routes do not
always give the optimal number of clusters.
● The hardware components used have to be regularly monitored and
replaced when malfunctioning.
 30
5.3 Future Work
In Some places population density distribution is non-uniform which makes the
waste collection so complex and costly. With some improvement in the system,
we can allocate the bins on the basis of density distribution over an area , so that
areas with high population density will get a higher number of bins. This can be
done by analysis or can be done manually, it will lower the cost of collection as
well as human intervention. Advances in route planning and optimization in route
generation can make the process to be operated and real-time data while waste
collection. Improvements in bin design will help to utilize the full volume of bins.
One possible work can be done with the help of some advanced technologies like
AR can provide a 3d view inside the bins, which will be beneficial to both users
and municipality authorities. Some improvements in architecture to improve the
system scalability. To get over the security concerns. To get over some security
threats, proper authentication, authorization, some cryptographic methods, and
a user access control should be established to reduce attacks like MiTM, DDOS,
sniffing, etc. one future work possibly can be to use edge computing as much as
possible to make processing, network, and connectivity near the end-user.
 31

References
1. Cloud Computing Based Smart Garbage Monitoring System, 2016 3rd International
Conference on Electronic Design (ICED), August 11-12, 2016, Phuket, Thailand.
2. Internet of things for smart cities,” IEEE Internet Things J., vol. 1, no. 1, pp. 22–32, 2014.
3. Thingspeak Based Sensing and Monitoring System for IoT with Matlab Analysis”,
International Journal of New Technology
4. A Development Framework for Smart City Services, Lulea University of Technology.
5. Smart Bin Implementation for Smart Cities, International Journal of Scientific & Engineering
Research, Volume 6, Issue 9, September-2015, pp. 787-79.
6. Balancing Energy Consumption in Clustered Wireless Sensor Networks, ISRN Sensor
Networks, Volume 2013 (2013).
7. S. Vinoth Kumar ; T. Senthil Kumaran ; A. Krishna Kumar ; Mahantesh Mathapati: “Smart
garbage monitoring and clearance system using internet of things”2017 IEEE International
Conference on Smart Technologies and Management for Computing, Communication,
Controls, Energy and Materials (ICSTM)
8. Internet of things (IoT) based Smart Garbage monitoring system, Indonesian Journal of
Electrical Engineering and Computer Science Vol. 20, No.2, November2020, pp. 736~743
ISSN: 2502-4752, DOI: 10.11591/ijeecs. v20.i2. pp736-743
9. Smart Garbage Monitoring System using Internet of Things(IoT), 2021 5th International
Conference on Computing Methodologies and Communication (ICCMC) |
978-1-6654-0360-3/20/$31.00 ©2021 IEEE | DOI: 10.1109/ICCMC51019.2021.9418359.
10. S. E. Comert, H. R. Yazgan, S. Kır, and F. Yener, “A cluster first-route second approach for a
capacitated vehicle routing problem: a case study,” Int. J. Procure. Manag., vol. 11, no. 4, p.
399, 2018.
11. D.S. Johnson and L.A. McGeoch, “The Traveling Salesman Problem: A Case Study in Local
Optimization”, November 20, 1995
12. Christian Nilsson,” Heuristics for the Traveling Salesman Problem”.
13. Toth, P. and Vigo, D. (2002) The Vehicle Routing Problem, SIAM Monographs on Discrete
Mathematics and Applications, Philadelphia.
14. K. Braekers, K. Ramaekers, and I. Van Nieuwenhuyse, “The vehicle routing problem: State
of the art classification and review,” 2016.
15. M. Huang and X. Hu, “Large scale vehicle routing problem: an overview of algorithms and an
intelligent procedure,” 2012.
16. H. Gomaa, “Activity Diagrams,” Softw. Model. Des., pp. 89–92, 2011.
17. Serap Ercan Comert,Harun Resit Yazgan,”Effective cluster-first route-second approaches
using Metaheuristic algorithms for the capacitated vehicle routing problem”