Optimization of Waste Collection in Smart Cities with the use of Evolutionary Algorithms

Full text

978-1-7281-9352-6/20/$31.00 ©2020 IEEE
Optimization of Waste Collection in Smart Cities
with the use of Evolutionary Algorithms
Ilknur Aktemur
Istanbul Kultur University
Computer Engineering Department
Istanbul, Turkey
ilknuraktemurr@gmail.com
Kübra Erensoy
Istanbul Kultur University
Computer Engineering Department
Istanbul, Turkey
erensoykubra13@gmail.com
Emre Kocyigit
Yildiz Technical University
Department of Computer Engineering
Istanbul, Turkey
kocyigit.emre.30@gmail.com
Abstract—With the growth of population, there is an
inevitable increase in solid waste especially in urban areas. For
the municipalities, especially in smart cities, this becomes a major
problem in nature, and it leads to many socio-economic and
environmental problems. Thus, lowering our living standards. To
eliminate or minimize these problems, using the Internet of
Things (IOT) technology is the most advantageous solution for
collecting solid wastes within this scope. In this paper, we
proposed an optimal waste collection mechanism with the use of
some IoT devices in the garbage cans which show the level of
waste in them. For testing the proposal, we select a sample
environment as a specific region of Istanbul, which is named as
Bakirkoy. With the use of sensors, it is aimed to detect which
cans are needed to be visited. Then with the use of an
evolutionary algorithm, Genetic Algorithm, best path for visiting
these cans can be planned in a very short time. By using this
approach, it is aimed to effectively use the workforce/resources of
the smart cities and making less traffic jam on the roads.
Experimental results showed that the proposed system results
very good enhancement in the waste collection operations.
I. I
NTRODUCTION
Population growth in cities results consumption growth per
capita, industry growth in economy, and they cause serious
solid waste production which results a serious waste collection
problem especially in smart city environments. Solid wastes
produced by human activities cause environmental, communal,
and vital problems in central living areas and cities. In order to
minimize and solve this problem, many cities have turned to
information and communication technologies. Because
traditional city concept and infrastructure are not able to fulfill
the current needs and provide an efficient solution for this kind
of problems, Smart City concept is developed. A Smart City
benefits from different types of IOT sensors to collect data and
use it to manage city with productive, healthy and ecological
ways. Smart city can collect data from different types of
sources such as citizens, devices or any assets and process it to
manage transportation systems, waste collection, water supply
network, crime detection and variable community services as
shown in Fig. 1. Solid waste management has crucial role in
Smart City concept, and it is focused on a cleaner, tidy and
green environment.
In our project, based on the smart city theme, digital
transformation of Bakirkoy/Istanbul is planned. Solid waste
management has been created as an important part of this
digital transformation in the project location since the
containers used in the accumulation of wastes in cities cause
hygiene and bad odor problems.
Fig. 1. Smart City
IoT technologies play an important role in designing new
services in smart cities. At this project, this technology is
responsible and used for waste management, production,
resource allocation, storage, and waste collection and disposal
services. Waste management; creates policies of pollution,
traffic density, and recycling. There are several waste
collections points in our project location as shown in Fig. 2
Waste Collection Point and some of them are in dense housing
areas. Since size of Waste Collection Vehicles is quite large as
shown in Fig. 3 Waste Collection Vehicles, traveling through
narrow city streets can be extremely difficult. Moreover, some
of the containers which are in waste collection points are
sometimes not sufficiently full and since there is not a system
that check occupancy rates, Waste Collection Vehicles
periodically visit all points. This situation causes major traffic
problems during the collections and unnecessary or early visits
end up with high fuel consumption of this vehicles. In order to
prevent this kind of problems, waste collection management
provides meaningful solutions.
The purpose of our project in Bakirkoy/Istanbul is to
determine the location of containers to collect waste, especially
the route determination for garbage collector trucks,
considering environmental, economic, and social factors in
Authorized licensed use limited to: ULAKBIM UASL - YILDIZ TEKNIK UNIVERSITESI. Downloaded on July 11,2021 at 16:22:14 UTC from IEEE Xplore. Restrictions apply.

urban areas. IOT, which we use in our project, also avoids
waste of time and resources by calculating the passage of the
tippers of truck nets according to the occupancy rates, thereby
targeting waste efficiency. In addition, devices, and tools are
capable of sensing, calculating, and communicating about
waste levels in trash cans are used.
Fig. 2 Waste Collection Point
The IOT system in our project consists of three layers as
sensors, communication network, and control. The sensors
placed in the garbage cans in Bakırkoy/Istanbul measure the
waste volume here and are connected to the municipality server
via a wireless connection. Sensor communication layers
support the automatic decision-making process on the server
and trucks are guided. In addition, it ensures that truck routes
are organized according to changes such as road interruption
and regulation. In our study, an optimal road planning
algorithm has been developed in order to calculate the most
suitable waste collection routes in Bakırkoy/Istanbul. Our
algorithm produces solutions with linear programming fed by:
Fig. 3 Waste Collection Vehicles
- Using the data in the database of the geographic
information system, to determine the location of the containers
in the streets containing the information of the roads in the
cities.
- Current number and capacity of trucks
- Time and traffic factors
Our algorithm uses the above information to calculate the
path required to collect waste and minimize routes by sending
the appropriate number of trucks.
The rest of the paper is organized as follows. In the next
section, the related works about the waste collection and also
usage of genetic algorithms for solving these type problems are
listed. The proposed methodology is detailed with the
implemented algorithms in Section 3. In Section 4,
Experimental results are depicted and finally, conclusions and
future works are drawn.
II. R
ELATED
W
ORK
Waste management is a discipline that covers production,
consumption, waste generation, waste recycling and / or
disposal. Starting from the design phase of any product. Most
of the research done in the world on manual collection,
transportation and recycling of waste. It is difficult to see
successful examples in the field of waste management since the
past, especially in large and dense cities.
When the cities with
old past even from the industrial revolution were in the stage of
installation, their technology and population were quite
different from today. The systems which are tried to be
managed without the smart devices and advanced systems are
very insufficient for today’s conditions. However, as suggested
in this paper, studies using smart systems and advanced
technology have already begun. Recently, people have started
to use technology to manage waste in more effective ways. In
addition to this, smart city was started to be used in the world
in order to use resources efficiently. Therefore, smart city has
become a trend used by the whole world. In this context, it was
aimed to create cities with greener thanks to energy saving [1,
21] and higher human quality [2]. These utilize GIS
technology, which can be easily integrated with different
platforms. GIS provides the user with a deeper perspective by
making cross-modeling and building relationships to help users
make smarter decisions. As a result, 2 important keywords
emerged. They are digital data and ICT infrastructure. The data
received from the sensors can be stored, processed, and
traceable in the routing services thanks to IOT and its
applications for the waste management system [3].
When the use of sensors becomes widespread, Big Data is
used to process and store. Wide-term studies have been
conducted in the world to produce solutions on garbage
collection problems, which have a direct impact on city
management costs. Some of the algorithms which routing [4,
11] and clustering [5] aim to minimize the travelling cost. In
these studies, sensors that measure level of the waste bins are
proposed, in addition to the studies that make it possible to
detect the level of waste bins, there are also studies aimed at
saving money in waste collection without further analysis [6]
This can be even established by using an autonomous control
mechanism as mentioned in [7]. The purpose of this study is
not a new solution that will provide waste collection
optimization by route planning. The aim of this study is to
evaluate the positive and negative aspects of a system that can
decide which ones should be collected daily with the
intelligence it provides to the trash cans.
After the detection of the target waste bins the complexity
of the problem is solved with the use of evolutionary
Authorized licensed use limited to: ULAKBIM UASL - YILDIZ TEKNIK UNIVERSITESI. Downloaded on July 11,2021 at 16:22:14 UTC from IEEE Xplore. Restrictions apply.

algorithms. In the literature, there are many works about this
type solution. Mainly most of these problems are converted to
a standard traveling salesman problem and solved accordingly.
In [8], Turker et. al. solved the path planning problem of the
UAVs by using the simulated annealing algorithm. Usage of
UAVs made the solution easier, because there are very small
constraints in the solution. There is no physical obstacle in the
air (or a smaller number of obstacles). However, in the path
planning of cars, especially in a real-world, real-time
environment is a trivial issue to solve. And, finally use of this
type system generate better solution for small size problems.
Because Simulated annealing can be accepted as a local search.
Therefore, for big size problems, it can be hard to find an
acceptable solution.
Cekmez et. al. proposed a new model for solving similar
path planning problem with the use of multiple vehicles [9, 19].
In their system they also used the UAVs and they can decrease
the task completion time for their problem environment.
Although the use of genetic algorithms with a parallel
implementation platform results better solution this can be used
for a multi vehicle solution. Similarly, in [10], authors
proposed the parallel solution for the TSP like problems in
their domain. Also, they showed that the reached results
increased the performance of the whole control system. These
papers showed that the use of several vehicles is very critical
for finishing the task in a smaller time. Therefore, this project
can be enhanced to a huge smart city (as Istanbul) for
optimizing the waste collection mechanism in an efficient way.
Another approach about this topic is [20] the sensors inside
the trash detect the level of the waste bin and when it is full, it
transmits it to the trash truck driver with a message service. In
[12] RDIF and GSM technology is used during data transfer
between garbage truck and garbage bin. Garbage cases in the
waste bins are detected with the help of the sensor system and
transmitted to the authorized place via the GSM system. The
GUI is used to view the current status of the waste bins in
different regions, thus increasing the efficiency of the waste
collection mechanism. Smart waste bins were proposed to
perform these functions in [13] The data from the sensors are
received by the network to which the bins are connected and
visualized to show and analyze the waste situation in the city.
Furthermore, Artificial Intelligence has started to dominate
waste management field like other fields and [17] is an
example of waste management study which developed an
application of artificial intelligence. Machine learning approach
is also used for waste management related cases like in [18]. It
is quite possible that smart cities will be equipped with
artificial intelligence and solutions will be created with
approaches such as machine learning in all areas.
In [14] the authors stated that the deficient of these systems
is that the garbage truck does not aim for the best route to reach
these full garbage bins according to the data obtained from the
sensors. They argue that in order to really ensure effectiveness
in the garbage collection system, a system that will give the
best route to reach the garbage bins is, as well as showing the
real-time status of the garbage bins is required. But for this aim
firstly the best route algorithm should be developed.
In this project, there is a need for a system that will show
full waste bins in Bakirkoy/Istanbul and propose using an
optimized way to access full waste bins. With the study we
have proposed in order to reduce the problems related with
increasing population and waste amount in this district where
there is a dense settlement, a more economical waste
management system with advanced technologies and more
efficient optimization algorithms has been envisaged.
III. M
ETHODOLOGY
In this paper, it is aimed to optimize the waste collection
problem of smart cities by using some hardware components
and by using a genetic algorithm to minimize the path of waste
collection vehicles. It is evaluated that the efficiency in waste
management will be increased and the problems will be
reduced with our work equipped with devices suitable for the
smart city concept that is not included in the current waste
management and developed with optimization techniques.
A. Genetic Algorithm
The genetic algorithm is a heuristic search algorithm that
mimics the biological reproduction and natural selection
process belonging to the evolutionary algorithm class.
Evolutionary algorithms are used to find solutions to problems
where unknown information is missing. Inherently, genetic
algorithms have been used to find solutions to problems such
as traveling salesman problems (TSP). All these algorithms are
built on a random process, and instead of producing a single
solution to problems with limited information, they create a
solution set consisting of different solutions. This cluster is
named as the population in GA terminology. Populations
consist of sequences called chromosomes or individual, and
every element in the individual is called a gene. These
chromosomes form the first population of the problem and are
tested for quality. How well the possible solution for
individuals in the population is evaluated by the fitness
function and gets a fitness value. In the fitness function, a
selection is made among the high-value chromosomes. Then,
crossing and mutation operations are performed on these
selected chromosomes, and a new population is created by
choosing the best chromosomes from among them. This
process is repeated when the desired best result is achieved
during the operation of the algorithm, when the fitness value
remains constant or in some cases, to terminate the process as
shown in Fig. 4. Some parameters need to be defined before
the algorithm is executed; these are population size, probability
of crossing and mutation.
B. Objective
In this project, it aims to solve the routing problem taken
into consideration by minimizing the total distance traveled by
garbage trucks of different capacities. The municipality looks
at the occupancy rate in the garbage container on each street. It
then optimizes a specific route for the garbage truck based on
the data it receives. Here it is aimed to reduce the workload
between the tools. Traffic situation on the road is checked
according to the working time of the garbage collection trucks.
Based on this, we need street information to calculate road
length, street direction and connections.
Authorized licensed use limited to: ULAKBIM UASL - YILDIZ TEKNIK UNIVERSITESI. Downloaded on July 11,2021 at 16:22:14 UTC from IEEE Xplore. Restrictions apply.

C. Problem Constraints
Each vehicle takes a trip back to where it originally started.
Between trash cans, each truck goes to a different set. A truck
visits each collection point. The traffic network on the road is a
factor for us. Because there is no direct path between the
collection points, the roads have to cover a longer distance. The
garbage collection vehicle collects the garbage container it will
visit if the container is full, or if it is not sufficiently full, it will
stop by the next garbage container. The amount of waste
collected is discharged to garbage recycling points. Then,
garbage containers are updated. In this way, we observe the
occupancy rates in garbage containers. The vehicle cannot be
interrupted during waste collection on the street we use.
Because if the carrying capacity of the vehicle is not sufficient,
the vehicle must return to the garbage container and take the
garbage in the vehicle for recycling and must discharge the
waste.
Fig. 4 Genetic Algorithm Flow Chart
D. Chromosome
In the implementation of GA, the chromosome is called
sequences created by combining one or more gene structures.
In this project, the chromosome is defined to represent the
points that each truck must travel within a single chromosome
in the collection area.
E. Fitness function
The fitness function is one of the most important parts of
the GA approach, because it is the only method to determine
the chance of selection of the personal chromosome to the next
generation and to measure development over generations.
Using the fitness function, a number called fitness is assigned
to each chromosome in the population. The fitness of the
chromosome depends on how well the genetic algorithm must
solve the problem. The algorithm must find the minimum
distance for the total road propagation, so fitness is inversely
proportional to the distance traveled and therefore we take the
fitness function as 1 \ (the total distance of the route it will use).
Here, if the vehicle capacity is available, the total distance will
be between different waste bins. If there is not enough vehicle
capacity, the distance from a garbage bin to the nearest
recycling station is taken, and then the distance to the next
garbage bin. Then we add time, traffic and road factors to
address the actual distance, which shows the time spent on the
given distance.
F. Initialization
GA is based on the development of the population until a
certain condition is reached. The initial population should be
created with the onset of evolution renewal. There are some
different ways to create the first population; a random
chromosome sequence can be created in nature. For the most
part, there are two issues in genetic algorithms that are
considered for the start of the population: First, the population
initiation procedure and population size depend initially on a
parameter. This dimension is maintained throughout the
algorithm. Secondly, the first created population has two ways:
intuitive start and random start. Although the average fitness
value is high in the intuitive startup, it helps GAs find solutions
faster. However, this includes some of the solution area and
cannot find optimal solutions due to the lack of diversity in the
population. For this reason, random start is used in this article,
so the initial population is generated by the coding method
described earlier.
G. Selection
The selection step can be carried out using various
techniques. Using several methods, chromosomes with the best
fitness are selected to improve the population, giving a better
chance to the next generation. There are many selection
methods; the most used is roulette wheel and tournament
selection. Both methods provide good and different parents in
many situations. In this research, we used the roulette wheel
selection. This gives the possibility to select each chromosome
in the population.
This probability is proportional to fitness and roulette is
assigned to the slot of the wheel according to the fitness value
of each individual in the population. Therefore, good solutions
have a larger slot size than less suitable solutions. Roulette
wheel which is detailed in Fig. 5, repeats this process to select
the next parent.
Authorized licensed use limited to: ULAKBIM UASL - YILDIZ TEKNIK UNIVERSITESI. Downloaded on July 11,2021 at 16:22:14 UTC from IEEE Xplore. Restrictions apply.

Fig. 5 Fitness Values
H. Crossover
Crossover is used to change the programming of one or
more chromosomes from one generation to another. Crossover
is sexual reproduction. Random genes are taken from the
mating pool to obtain offspring with superior characteristics.
As the expression of choice, there are multiple methods for
applying the leveling; the most common point is crossing and
two point leveling. In this project, these two projections are
used. The easiest crossover operator is one-point scale. A
random crossover point is selected in the main organism
sequence, and this point is called the cutoff point. Every point
that can be selected has an equal chance of being selected.
Error! Reference source not found. 6 shows two parents,
each represented by a series of characters with a total of 8
element sizes. To create the children, the algorithm copies the
first part of the parent and adds it to the puppy. He then takes
the second part of the other parent and adds them to the
offspring.
Fig. 6 One-Point Crossover
A two-point transition is the generalization of a single-point
transition. The difference is that the two-point transition
chooses two breakpoints, which divides the parents into 3
parts.
Fig. 7 Two-Point Crossover
In the two-point transition, the selection is made by mixing
each parent piece in the child. As a result of the elections, new
children are formed. In Fig. 7, the first part of the first parent is
copied and added to the offspring. The second part of the next
second parent is copied and added to the puppy. Finally, the
last part of the first parent is copied and added to the puppy.
Repeating this process by changing the order of the parents
produces the second offspring.
I. Mutation
The mutation is vital to improve research and is called
small random changes in the genetic material of a
chromosome; these changes preserve diversity in the
population. Mutations are not expected to bring the solution to
a better fitness, but it is necessary to improve them in the next
iteration. In Error! Reference source not found., swap
mutation is used. In this mutation technique, any two genes of
the chromosome are selected and exchanged between them,
these cases occur when a gene with TSP status is not allowed
to repeat.
Fig. 8 Mutation
J. Termination
Two conditions are used to terminate genetic algorithms:
1-The algorithm reaches the highest fitness solution
2-number of iterations reached
When the termination process is reached, the chromosome
with the best fit of the final population is chosen as the best
solution found by the genetic algorithm.
IV. E
XPERIMENTAL
R
ESULTS
In our experiment, Bakirkoy, Istanbul location, which is
shown in Fig. 9, was selected as project area.
Fig. 9 Bakirkoy, Istanbul Map
There are so many big size waste collection boxes and all
of them are potential points which are needed to visit by Waste
Collection Vehicles as displayed in Fig. 10. in this location.
Authorized licensed use limited to: ULAKBIM UASL - YILDIZ TEKNIK UNIVERSITESI. Downloaded on July 11,2021 at 16:22:14 UTC from IEEE Xplore. Restrictions apply.

Fig. 10 Waste Collection Boxes in Bakirkoy/Istanbul
Our proposal algorithm was executed on a PC, which is
configured as in Table 1. The computation time and speedup of
the applied algorithm are also tested on this machine. In the
hundredth generation, the route leading to the full waste bin
with the best chromosome is created, and this road is visualized
on the map. The route is represented in the map of
Bakırkoy/Istanbul, as can be seen in figure.
Table 1 Test Platform Properties
Computer Lenovo–HuronRiver
Platfor
m
Operating Syste
m
Microsoft Windows 10 Pro
CPU
Intel® Core™ i5-2410M
CPU @ 2.30GHz, 2301 Mhz,
2 Cores, 4 Logical
RAM 4,00 GB
IDE PyCharm 2020.1 EAP
Fig. 11 The route of Bakirkoy/Istanbul
In Genetic Algorithm, parameters are crucial for solution
success. Our proposed algorithm is performed with different
number of waste collection points according to Genetic
Algorithm parameters listed in Table 2.
Table 2 Genetic Algorithm Parameters
Paramete
r
Value
Parent Selectio
n
Roulette Wheel
Crossover Type Two Point Crossove
r
Population Size 100
Mutation Rate(%) 10
Elitism Rate(%) 2
Two-Opt Rate(%) 1, 5, 10
Fig. 12 Distance - Generations
Firstly, we calculated the path distances generation by
generation according to 52 waste collection points and obtained
feasible path distance results. Experimental results are detailed
as in Fig.12.
Genetic Algorithm is preferred owing to its feasible and
successful solutions. However, computation time is usually too
long. In order to decrease the computation time and create fast
solutions, we used Dynamic Programming principals. Our first
results are described in the first two rows of Table 3. In
addition to all, while avoiding too long computation time, we
benefit from 2-opt algorithm to obtain more successful
solutions in earlier iterations. Also, we added and changed 2-
opt rate to see its effect on computation time with same
iteration number. These results are described in the last three
rows of Table 3.
Table 3 Computation Time Results
No 2-opt 2-opt
Rate Iteration Computation
Time(sec)
1
N
o - 100 0,88470
2
N
o - 1000 8,93425
3 Yes 0.01 100 1,07324
4 Yes 0,05 100 1,72861
5 Yes 0,1 100 2,63584
After sensor system is adapted, there will be no need to
visit all 52 waste collection points that was displayed in Fig.
11. Therefore, we executed our algorithm with different
number of waste collection points. For example, if eight waste
collection points of were needed to visit according to their
occupancy rate, solution would be as displayed in Fig. 13.
Authorized licensed use limited to: ULAKBIM UASL - YILDIZ TEKNIK UNIVERSITESI. Downloaded on July 11,2021 at 16:22:14 UTC from IEEE Xplore. Restrictions apply.

Fig. 13 Route with 8 collection points
V. CONCLUSIONS
Waste management aims to minimize the impact on the
environment and economy in waste disposal in the whole cities
of the world. The most important percentage of the cost of the
solid waste management system is the collection of waste. The
success of an integrated, solid waste management system is
directly proportional to the success of the collection system.
So, in this proposed project, we aimed to find the best route to
reach filled waste bins according to the information of waste
level that we received from the sensors in the waste bins.
Thanks to the most effective route found cost and time loss are
prevented. It uses the genetic algorithm to find the best route
and the route found after 100 generations is accepted. The
program produces 100th generation in less than a second and
will give optimum possible solutions. The crossing rate and
mutation rate in the program randomly produced depending on
the suitability value. These rates are not defined as coefficients;
they are defined as variables. By that way, the random search
method, the structure of the genetic algorithm has been
preserved here as well. While trying to find the best route in
dense residential areas such as Bakırköy district, which is our
project area, the restrictions caused by the city infrastructure
established with old technology, narrow streets and the fact that
garbage collection vehicles are larger than the old narrow city
streets should be taken into consideration. In this study, a
vehicle moves around the project area. Working with multiple
waste collection vehicles can be expanded in much larger
areas.
The problem domain is the main set up in a relatively small
space. However, if the number of the waste box is increased,
this also makes the solution of the problem hard. Although the
use of a genetic algorithm finds an acceptable solution, due to
the increased complexity, there is a need to use some parallel
programming capabilities which can either be run on multi-
core structure as a new type CPU [15] or a parallel
programming environment like CUDA [16].
R
EFERENCES
[1] K. Muhammad, J. Lloret and S. W. Baik, "Intelligent and Energy-
Efficient Data Prioritization in Green Smart Cities: Current Challenges
and Future Directions," in IEEE Communications Magazine, vol. 57, no.
2, pp. 60-65, February 2019, doi: 10.1109/MCOM.2018.1800371.
[2] G. Piro, Cianci I., Grieco L., Boggia G., and Camarda P., “Information
centric services in smart cities”, Journal of Systems and Software, vol.
88, pp: 169–188.
[3] N. Mitton, Papavassiliou S., Puliafito A., Trivedi K. S., “Combining
Cloud and sensors in a smart city environment”, EURASIP Journal on
Wireless Communications and Networking vol. 2012, Article number:
247.
[4] T. Nuortio, Kytöjoki J, Niska H., Bräysy O., Improved route planning
and scheduling of waste collection and transport, Expert Systems with
[5] K. Byung-In, Seongbae K., Surya S., Waste collection vehicle routing
problem with time windows, Computers & Operations Research,
[6] Vicentini, Giusti A., Rovetta A., Fan X., He Q., Zhu M., Liu B.,
Sensorized waste collection container for content estimation and
collection optimization, Waste Management, Volume 29, Issue 5, May
2009, Pages 1467-1472, ISSN 0956-053X
[7] M. C. Olgun, Z. Baytar, K. M. Akpolat and O. Koray Sahingoz,
"Autonomous Vehicle Control for Lane and Vehicle Tracking by Using
Deep Learning via Vision," 2018 6th International Conference on
Control Engineering & Information Technology (CEIT), Istanbul,
Turkey, 2018, pp. 1-7, doi: 10.1109/CEIT.2018.8751764.
[8] T. Turker, O. K. Sahingoz and G. Yilmaz, "2D path planning for UAVs
in radar threatening environment using simulated annealing algorithm,"
2015 International Conference on Unmanned Aircraft Systems
(ICUAS), Denver, CO, 2015, pp. 56-61.
[9] U. Cekmez, M. Ozsiginan, and O. K. Sahingoz, "Multi-UAV Path
Planning with Parallel Genetic Algorithms on CUDA Architecture", In
Proceedings of the 2016 on Genetic and Evolutionary Computation
Conference Companion (GECCO ’16 Companion), Association for
Computing Machinery, New York, NY, USA, 2016, pp. 1079–1086.
[10] T. Turker, O. K. Sahingoz and G. Yilmaz, "GPU-Accelerated Flight
Route Planning for Multi-UAV Systems Using Simulated Annealing",
International Conference on Artificial Intelligence: Methodology,
Systems, and Applications AIMSA, 2016, pp. 279-288.
[11] S. Birtane and Ö. K. Şahingöz, "Data gathering from a large scale
Wireless Sensor network with means of Unmanned Aerial Vehicles,"
2016 24th Signal Processing and Communication Application
Conference (SIU), Zonguldak, 2016, pp. 2105-2108, doi:
10.1109/SIU.2016.7496187.
[12] S. Gupta and P. Kumar, “Real Time Solid Waste Monitoring and
Management System: A Case Study of Kanpur City,” International
Journal of Science, Environment and Technology, 4(2), April 2015.
[13] N. Sharma, N. Singha, and T. Dutta, “Smart Bin Implementation for
Smart Cities,” International Journal of Scientific & Engineering
Research, 6(9), Sept 2015.
[14] S.S. Chaudhari , V. Y. Bhole ,”Solid Waste Collection as a Service
using IoT- Solution for Smart Cities” Department of Information
Technology,Research,Jan2019.
[15] U. Cekmez, M. Ozsiginan, M. Aydin, O.K. Sahingoz, UAV Path
Planning with Parallel Genetic Algorithms on CUDA Architecture,
WCE, IAENG, Lecture Notes in Engineering and Computer Science, 1,
2014, pp. 551-557.
[16] E. Kugu and O. K. Sahingoz, "ACO algorithms with multi-core
implementation," 2013 7th International Conference on Application of
Information and Communication Technologies, Baku, 2013, pp. 1-5.
[17] T. H. Ali, A. H. Memon, H. U. Imad and S. H. Khahro, “Application of
Artifical Intelligence in Construction Waste Management: A conceptual
frame-work for effective waste management system”, 2019 8th
International Conference on Industrial Technology and Management,
2019, pp. 50-55.
[18] M. Graus, P. Niemietz, M.Rahman, M. Hiller and M. Pahlenkemper,
“Machine Learning Approach to Integrate WasteManagement
Companies in Micro Grids”, 2018 19
th
International Scientific
Conference on Electric Power Engineering(EPE), 2018.
[19] U. Cekmez, M. Ozsiginan and O.K. Sahingoz, "Multi-UAV Path
Planning with Multi Colony Ant Optimization", ISDA 2017:
International Conference on Intelligent Systems Design and
Applications, 2018, pp:407--417.
Authorized licensed use limited to: ULAKBIM UASL - YILDIZ TEKNIK UNIVERSITESI. Downloaded on July 11,2021 at 16:22:14 UTC from IEEE Xplore. Restrictions apply.

[20] K. Mahajan and J. Chitode, “Zig-Bee Based Waste Bin Monitoring
System,” International Journal of Engineering Sciences & Research
Technology, 3(2), Feb 2014.
[21] S. Ahmad, Imran, F. Jamil, N. Iqbal and D. Kim, "Optimal Route
Recommendation for Waste Carrier Vehicles for Efficient Waste
Collection: A Step Forward Towards Sustainable Cities," in IEEE
Access, vol. 8, pp. 77875-77887, 2020, doi:
10.1109/ACCESS.2020.2988173.
Authorized licensed use limited to: ULAKBIM UASL - YILDIZ TEKNIK UNIVERSITESI. Downloaded on July 11,2021 at 16:22:14 UTC from IEEE Xplore. Restrictions apply.